/ Table of Contents

The Intersection of AI and Simulation.......................................................................................................................................................................................... 02

Behind the Scenes:

Get to Know the Ansys SimAI Software-as-a-Service Platform..................................................................................................................................04

Simulate Securely with Ansys SimAI Technology................................................................................................................................................................ 07

Ansys SimAI Software in Action:

Ship Hull Design.........................................................................................................................................................................................................................................08

Vehicle Crash Predictions..................................................................................................................................................................................................................... 09

Design Iteration for Chips, Jets, and SUVs................................................................................................................................................................................. 10

Additional Use Cases..................................................................................................................................................................................................................................11

Get Started with Ansys AI Products................................................................................................................................................................................................ 16

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 01

/ The Intersection of AI and Simulation / KNOW THE TERMS:

Artificial Intelligence (AI): a computer-based

system or set of algorithms that can imitate
human behavior to process tasks. Despite its
name, AI is based on real — not artificial — data.

AI Training: the process of feeding large

amounts of data to an AI model to train it
to perform and complete tasks. Simulation
significantly improves AI training by providing
a comprehensive pool of data, speed, and
predictive accuracy.

Algorithms: procedures or instructions that

enable computers or machines to analyze data,
perform tasks, and make decisions.

Machine learning (ML): a subset of AI; a

computational method that uses data and
algorithms to train AI.
Across industries, product development teams are At the same time, while AI can enhance simulation,
constantly challenged to produce better designs at a simulation can also enhance AI with optimized training Deep Learning (DL): a form of ML in which a
faster pace to meet consumer demands. In the past, and predictive accuracy. The two technologies are computer or machine teaches itself without
design and development were limited by the accuracy interconnected, maintaining a symbiotic relationship in human intervention using neural networks.
and speed of individual engineers running simulations which each counterpart benefits the other.
by hand. Neural Networks (NN): deterministic
models that mimic and are inspired by the
Today, artificial intelligence (AI) technologies present a interconnected network of neurons in the
Enhance Simulation with the Power of AI
unique opportunity to accelerate production and enhance human brain.
simulation workflows. The potential expands even more • Faster: AI analyzes past simulations to quickly identify
when leveraging the nearly limitless compute power of complex patterns while incorporating new information
cloud-native AI technology. to examine correlations and relationships within data.

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 02

 Ansys AI: The Future of AI and Simulation / KNOW THE TERMS:
Has Arrived
At Ansys, we are revolutionizing engineering simulation Generative AI: AI that can generate new
with the power of AI. Expanded AI integration across content, including text, images, and other data,
our portfolio and customer community includes the by learning from previously generated data.
• Easier to Use: AI democratizes the use of simulation by
making it more accessible to nonexperts through user- Ansys SimAI™ platform, a cloud-native, physics-neutral
software-as-a-service (SaaS) application that combines Large Language Model (LLM): a specific
friendly web-based applications.
Ansys simulation with the speed of generative AI; Ansys application of generative AI that enables natural
• More Comprehensive: AI enables the integration of AI+™ product capabilities, which extend AI features language processing (NLP) and can decipher
multiple models to provide thorough representations of within Ansys products to enhance core functionalities; and understand nuances in language. This
complex systems. and the AnsysGPT™ multilingual and conversational extends to capabilities, including summarization
virtual assistant. and translation. One of the most popular
examples of an LLM is ChatGPT.
Enrich AI Training with the Predictive Accuracy
of Simulation AI-Powered/AI-Augmented/AI-Enhanced
Simulation: frequently used terms to describe
• Predictive Accuracy: Physics-based solving methods
simulation that has been optimized using AI.
and full-fidelity simulations provide high levels of
predictive accuracy to validate and optimize designs,
Cloud-Native AI: AI platforms that run in the
which increases the value and accuracy of AI
cloud, combining the power of AI with cloud
training data.
computing to scale, accelerate, and optimize AI
• Scope: Multiphysics simulation covers a wide range of training and performance.
engineering disciplines across all industries and enables
a near-limitless number of variables to test and validate, Software-as-a-Service (SaaS): a software
improving analysis for deeper design insight. This delivery model that enables users to access
enables a more diverse dataset for AI training. cloud-based applications via subscription.
• Speed: Simulation-based virtual testing is much faster
and more reliable than physical testing, saving
significant time during the data collection needed for
AI training.

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 03

/ Get to Know the Ansys SimAI Software-as-a-Service Platform

By combining the power of artificial intelligence (AI) and multiphysics simulation, the or other sources and assess the performance of a new design within minutes. The
Ansys SimAI platform enables organizations to reach even greater levels of innovation software-as-a-service (SaaS) application combines the predictive accuracy of Ansys
at a rapid pace. The SimAI solution is a physics-agnostic and cloud-native platform in simulation with the speed of generative AI via the cloud — a combination that boosts
which you can train an AI model using previously generated data from Ansys products model performance by 10-100X across all design phases for computation-heavy projects.

To assess automotive aerodynamics, engineers often incorporate computational fluid

dynamics (CFD) simulation to analyze the flow field around a car.

Inform Machine Learning with Numerical Simulation

To appreciate the technology behind the SimAI solution, you must first understand Governing equations play a crucial role in the analysis of physical systems because they
the nature of numerical simulations and their complexities and purpose. Numerical enable models that can be used to analyze and predict behaviors that are otherwise
simulations model the behavior of physical systems in various engineering domains unobservable. By employing mathematical approximations, numerical simulations
without needing to conduct lab experiments. This enables engineers to calculate and solvers apply these governing equations through computer algorithms.
iterate system-specific quantities more easily and efficiently.

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 0 4

 New Design Performance Prediction

UPLOAD TRAIN PREDICT

Take a Physics-Agnostic Approach with the Ansys SimAI Solution 2. Physics-agnostic: The other approach is for the ML algorithm to learn a latent
representation of the physics directly from the solutions of numerical solvers.
When employing ML techniques, the goal is to enable an algorithm to learn a task or
These methods are agnostic to the underlying equations and solvers. For instance,
general rule, given a set of examples. There are several ways to do this when integrating
by using examples from previous computations, ML algorithms can learn a data-
ML with numerical simulation.
driven representation of a governing equation and use a new geometry and
Here are two categories of such methods: freestream condition as inputs — like the speed of a car — to deliver its flow field
as the output. In this way, physics-agnostic methods offer both speed and predictive
1. Physics-aware: The most straightforward approach is to replace the bottleneck in
accuracy.
standard simulation solvers with ML techniques that can output predictions faster,
such as accelerating the inversion of a large linear system of equations. However, The SimAI application takes a physics-agnostic approach, accelerating your simulation
most of these methods remove the solver involved in the computation while keeping workflow without compromising its predictive accuracy. Instead of relying on geometric
information about governing equations, e.g., by injecting penalization terms implied parameters to define a design, the platform uses the shape of a design itself as the
by the equations into the loss function (i.e., the amount of error in your training data). input. This facilitates broader design exploration even if the shape’s structure is
In other words, physics-aware methods focus on a particular physical task and get the inconsistent across the training data.
job done, but they might sacrifice some accuracy.
Essentially, the SimAI workflow is engaged in three simple steps: upload data, train AI
model, and predict. As mentioned, customers can train the AI model using previously
generated data.

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 0 5

Discover Ansys SimAI Architecture Similarly, the application uses an adequate representation / ANS YS AI+ TECHNOLOGIES
of 3D shapes to describe arbitrary or irregular meshes
SimAI architecture is based on a fusion of different
with complex geometrical variations that do not follow a Ansys AI+ technologies enable customers to
techniques that combine multiple deep learning neural
specific distribution, e.g., unparameterized geometries. augment simulation with add-ons, offering
networks. This type of architecture makes it possible
to capture all the important scales of physics. SimAI’s more options and variety in accessing artificial
To help quantify the uncertainty in the prediction, the intelligence (AI) capabilities across Ansys
architecture is composed of a significant number of
SimAI platform uses a unique confidence score to desktop products. Ansys AI+ technologies
nonlinear layers, including multiple parameters and
calculate the distance from the nearest known geometry include:
complex interactions between variables.
in a very high-dimensional space. • Ansys Granta MI AI+ module for greater
Instead of explicitly storing data points like pixel values insights into material-process-property
Explore AI at Ansys relationships
in an image, the architecture uses an implicit neural
representation to learn a continuous function that can While it’s good to know the basics of SimAI technology, • Ansys optiSLang AI+ module for sensitivity
generate these data points. This means that the SimAI the best part about the platform is that you don’t need to analysis, optimization, and robust design
solution can take a set of points obtained from previous know any of this to use it. The SimAI solution is intuitive • Ansys CFD AI+ module for turbulence
computations and generalize well to new geometries and and designed for users without coding experience, AI/ML modeling
freestream conditions. familiarity, or deep learning expertise. • Ansys SynMatrix AI+ module to accelerate RF
filter optimization
Model with Quick and Easy Geometry To learn more, register for the webinar “Introducing Ansys • Ansys Structures AI+ for faster thermal
SimAI: Cloud-native Generative AI for Simulation,” and strain calculations and prediction of compute
In fact, one of the SimAI platform’s strongest assets is its resources
discover other Ansys AI innovations, including Ansys AI+
use of regularization techniques to prevent overfitting and • Ansys Missions AI+ to enhance Digital Mission
product technologies and the AnsysGPT virtual assistant
improve generalization for new geometries. What does Engineering products
at Ansys AI — AI-augmented Simulation Technology.
this mean? Basically, overfitting is when an AI model’s • Ansys Electronics AI+ to forecast simulation
predictions are too narrowly focused on its training needs
data and included geometry, lacking any generalization
or scope to learn or generalize new geometries.
Regularization techniques are designed to reduce
overfitting.

The SimAI solution uses regularization techniques,

including local methods that are embedded directly in
the structure of the model and lead to a more stable and
expressive model. It’s the reason the application can work
with new geometries so quickly and easily.

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 0 6

/ Simulate Securely with Ansys SimAI Technology
/ ANS YSGPT

Ansys customers can trust that their Know Your Infrastructure AnsysGPT is a virtual assistant that garners
SimAI data is secure from initial upload The SimAI platform is multitenant with logical separation engineering expertise across physics domains,
throughout model building and prediction through tenant-based and user-based ownership. provides 24/7 comprehensive technical
runs. The SimAI platform is architected on Likewise, simulations are run on tenant-exclusive, support, and reduces response times.
AWS by keeping in account the security ephemeral-allocated instances and all read objects are
and architecture best practices from the copied, used, then destroyed. Unlike general AI virtual assistants that
AWS Well Architected (WA) Framework. The AWS WA use unsupported information, AnsysGPT is
Framework describes key concepts, design principles, As another layer of security, SimAI infrastructure is trained using Ansys data to generate tailored,
and architectural best practices for designing and managed using an infrastructure as code (IAC) model applicable responses drawn from reliable
running workloads in the cloud. It is based on six pillars: on Kubernetes (k8s) clusters using Amazon Elastic Ansys resources including Ansys Innovation
operational excellence, security, reliability, performance Kubernetes Service (EKS). What does this mean? Courses, technical documentation, blog
efficiency, cost optimization, and sustainability. Put simply, k8s clusters are groups of nodes that run articles, and other vetted resources.
containerized applications. This enables multiple
Train AI Securely applications to run securely in a more efficient and
To get started, simply select the files you wish to use scalable way. As a result, the benefits of the IAC model
and either upload them to the SimAI platform using are maximized.
a web application directly through your browser or by
using a Python software development kit (PySimAI In addition, SimAI user accounts are managed through an
SDK) compatible with Python version 3.9 or higher. The industry-grade authentication and authorization service,
PySimAI documentation is publicly available. Your data is which verifies access to the platform using a customer
securely stored in an encrypted data bucket in Amazon identity provider (IdP). Data transfer is only possible after
Simple Storage Service (S3) and transferred through an authentication and authorization. Further, your storage,
encryption tunnel protected by transport layer security i.e., your Amazon S3 data bucket, is protected with AWS
(TLS) 1.3 — the most recent and strongest version of the Key Management Service (KMS) encryption owned only by
TLS security protocol. ephemeral compute instances, which use the advanced
encryption standard (AES) 256. Access is also filtered and
Once your data is uploaded, you can configure a model restricted by AWS Identity and Access Management (IAM)
and initiate training with confidence that your data permissions. For more information, read our recent blog
remains secure throughout all stages of the SimAI “Maximize Simulations with Secure, Cloud-native AI.”
workflow. The SimAI architecture performing training and
inference is hosted in the AWS Europe region (eu-west-1,
Ireland), leveraging best security practices.

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 07

/ Ansys SimAI Software in Action

Regardless of your occupation or industry, there’s a modern-day unifying truth: artificial Main benefits of the SimAI solution in ship hull design include:
intelligence (AI) has the potential to transform your workflow, if it hasn’t already. Let’s • Accelerated performance assessment of new hull designs in a few minutes
explore how the Ansys SimAI application optimizes design workflows from chips to instead of days.
ships, vehicles, and beyond. • Informed decision-making in the early stages of design.
• Saved time for value-added tasks, including design exploration, which creates more
Inform and Accelerate Ship Hull Design time to innovate.
As environmental concerns grow around the world, more manufacturers are being
required to reduce carbon dioxide (CO2) emissions. Similar to the automotive and
aerospace industries, the maritime industry is no exception. An example is the
push from the International Maritime Organization (IMO) to reduce emissions from
international shipping.

To achieve this, one approach is to improve ship resistance by optimizing hull form
X-Wall Shear Stress
during design. This directly translates into energy efficiency and fuel savings. However, Isostatic Pressure
fast and efficient simulations are needed to evaluate several hull designs at once with so
many different variables.

Apply the Ansys SimAI Solution to Hull Design Wave Patterns (Elevation)
Employing the SimAI application, engineers created an AI model using 288
computational fluid dynamics (CFD) simulation results that included hull shape
variations and operational conditions such as draft and boat speed.

The simulations were split into three sets:

Ansys SimAI Prediciton: ~50 seconds CFD Calculation: 4 hours on 32 cores
1. Training set: contained 244 simulations used to train the model.
By combining traditional simulation methods like computational fluid dynamics (CFD) with the SimAI
2. Validation set: contained 20 simulations used to tune the model.
application, you can predict new geometry for ship hull design in less than a minute.
3. Testing set: contained 20 simulations used after the training phase to give an
unbiased estimation of the model performances on unseen data.

In less than one minute, the application predicted new hull geometry. In addition, the
SimAI resistance error was less than 5% compared to the CFD simulation results and the
wave pattern prediction was fully accurate.

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 0 8

Benefit Beyond the Hull the full 3D response of a new design in minutes. Additionally, as a physics-agnostic
Another advantage of the application is the reduction in preprocessing time needed to application, it can be trained to work with fluids, structures, electromagnetics, and other
make a prediction because a mesh or solver setup is not required. This saves time — from physics domains.
minutes up to hours — depending on the process automation and design complexity.
Apply the Ansys SimAI Solution to Crash Tests
Each SimAI prediction takes around 8 seconds, which unlocks new possibilities for faster SimAI software has proven effective in predicting how fluids move, especially in constant
and more efficient design iterations. For example, by incrementally shifting the longitudinal motion simulated by CFD solvers. To show how the software can handle other physics,
center of buoyancy (LCB) of the hull forward or backward, you can reduce wave amplitudes including structural applications, Ansys engineers used it to predict how a car bumper
and consequently optimize ship resistance. And, circling back to sustainability initiatives, would behave in a high-speed head-on crash. A simplified vehicle model, or sled, was
improving ship resistance through hull form optimization offers a large potential for fuel used to simulate the bumper impacting a rigid wall at 15.6 meters per second. The load
savings. case aimed to predict the deformed shape and rigid wall force for a given geometry.
To accurately predict the transient response, Ansys engineers introduced design
In this study, ship resistance was calculated for different positions of the LCB and minimum variations. Changing the thickness of seven parts within minimum and maximum
resistance was achieved in 10 minutes — a drastic reduction from the typical computation boundaries generated 98 different designs with 21 states each — for example, a crushed
time and effort required to optimize hull form. Further, these results can be obtained by one state — for a total of 2,058 training data points.
engineer within one working day.
Ansys engineers then uploaded the data to the SimAI platform,
To learn more, check out the white paper “How to Improve Hull Designs with Ansys SimAI.” where it was automatically divided into three subsets:
1. A training set to train the model
Improve Bumper Impact Performance with Optimized Crash Test
2. A validation set to adjust or
To meet shrinking automotive development timelines and ever-increasing safety
“tune” the model’s underlying
regulations, computer-aided engineering (CAE) has become an integral part of the structure to improve its
vehicle design process. Virtual analysis enables designers to reliably assess a vehicle’s performance
crash and safety performance while reducing costly physical testing.
3. A test set to determine how
well the model could handle
Ansys LS-DYNA® nonlinear dynamics structural simulation software is one tool that has situations that it hadn’t
become an industry standard for building high-fidelity crash models. However, the seen before
complexity of vehicle models has grown exponentially with details. Although this has
improved confidence in CAE predictions and reduced the number of physical tests,
it has made the models much larger and increased the need for greater compute
resources. A comparison between Ansys SimAI software and solver
predictions of a bumper’s deformed shape at a given time show
SimAI software bridges the gap between theoretical models and real-world the SimAI model predictions to be reasonably accurate.
complexities. It learns from existing simulations and uses that knowledge to predict

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 0 9

Reviewing the Results Shows Efficiency Gains within one minute, delivering a less than 0.5% temperature difference compared to
Most of the SimAI model predictions were within 10% of the crash solver’s predictions MAPDL while accurately pinpointing critical locations.
(Figure 1). The graph shows how the rigid wall force changes throughout the crash. The
solver’s predictions for a design at different points in time are on the X axis, while the An aerospace team experienced similar success producing jet engine bracket designs
SimAI model predictions for the same design and time stamps are on the Y axis. This more quickly to meet structure requirements. Using Ansys Discovery™ 3D product
suggests that SimAI software is accurate without overfitting too much. It also highlights simulation software and Ansys optiSLang® process integration and design optimization
areas where SimAI training might need improvement. Another comparison of the SimAI software, the team built an AI model using about 250 training samples, including
software predictions to the crash solver predictions showed similar accuracy (Figure 2). topologically diverse bracket designs. With insight from past simulations, the SimAI
application predicted a new shape and its physical behavior in less than a few seconds.

INPUT: Power Map OUTPUT: Temperature Distribution

Figure 1. A trend plot for Ansys SimAI software shows Figure 2. Comparison of rigid wall force
overall crash prediction accuracy as compared to the predictions between the crash solver (A) and
traditional crash solver. SimAI software (B).

For more information, review the white paper “Ansys SimAI Software Predicts Fully Ansys Mechanical: Ansys SimAI:
Transient Vehicle Crash Outcomes.” 30 minutes on 4 Cores Less than 1 minute

Accelerate Design Iteration for Chips, Jets, and SUVs SimAI software is impacting the automotive industry, too. One team integrated the
Faster design iteration is a goal shared by design and development teams in every software to assess the aerodynamic performance of a sports utility vehicle (SUV) and
industry from semiconductors and aerospace to automotive. reduce carbon emissions. Using approximately 50 CFD results from Ansys Fluent® fluid
simulation software, the group created an AI model that included car exterior shape
A recent semicondutor team was seeking a faster way to explore chip design. Their variations and topological changes, such as a rear mirror, ski rack, and spoiler. With the
existing approach slowed them down due to a dependence on domain discretization support of the SimAI platform, predictions for the new SUV geometry took less than one
and the number of elements required in chip modeling. It also offered limited scalability. minute. In addition, the application’s drag error compared to CFD was less than a 0.5%
Using Ansys Mechanical Parametric Design Language™ (MAPDL) and simulation results difference.
from Ansys Mechanical™ structural finite element analysis software, the team created
an AI model in the SimAI platform based on 2,250 diverse chip thermal simulation To learn more about the use cases in this section, read the blog: “From Chips to Ships:
datasets. The application predicted a temperature map for new input power maps Optimize Design with the Ansys SimAI Platform.”

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 1 0

/ Ansys SimAI Software in Action: Additional Use Cases

Structures
/ Wire Forming Process

CHALLENGE: Designing hairpin wires for electric motors requires a lengthy trial-and-error process to determine
optimal parameters.

SOLUTION: Used Ansys Workbench™ simulation integration platform, LS-DYNA software, and optiSLang software to
automate the simulation workflow, model the forming process, and create a database of simulation results to train
the AI model in the SimAI platform.

BENEFIT: The application predicted an as-manufactured shape of a wire centerline in less than one minute with a
shape difference of less than 5% compared to LS-DYNA software, reducing time and cost.

/ Crystal Plasticity Homogenization

CHALLENGE: Material properties for different microstructures are typically obtained from costly and time-consuming
experimental setups. Alternatively, a crystal plasticity simulation can be performed.

SOLUTION: Mechanical simulation results of 50 microstructures at 10 different strain levels were generated to train an AI
model using the SimAI application.

BENEFIT: EFIT The application predicted the mean stress-strain curve and the stress distribution in less than one
minute, delivering a less than 3% stress difference compared to the Mechanical simulation results, reducing time
and cost.

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 1 1

/ Ansys SimAI Software in Action: Additional Use Cases

/ Crane Hook Design

CHALLENGE: Crane hook design often involves multiple iterations to achieve desired performance and safety factors.

SOLUTION: Create an AI model in the SimAI application using 50 diverse hook design Mechanical simulations.

BENEFIT: The SimAI platform predicted stress and deformation results for new geometries within one minute with a
stress difference of less than 1% compared to the Mechanical simulation results while accurately pinpointing critical
locations.

/ Rubber Bush Force-stroke Curve Estimation

CHALLENGE: A rubber bush serves as a vibration isolator in automotive components like suspension systems and
engine and gearbox mountings. Simulation analyses of rubber bush properties are time consuming and difficult due to
hyperelastic material, large deformation, and nonlinear contact.

SOLUTION: Used about 10 evaluated designs from the Mechanical simulation results to generate a surrogate AI model
in the SimAI platform.

BENEFIT: The application can predict within five minutes compared to Mechanical software, which takes 1.3 hours.

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 1 2

/ Ansys SimAI Software in Action: Additional Use Cases

Fluids
/ Battery Thermal Management Conceptual Design

CHALLENGE: In the global race for electric vehicles, a fast turnaround time is necessary to stand out from the
competition and optimize cooling design for safety and performance.

SOLUTION: Approximately 10 CFD simulation results were used to create an AI model of a battery module in the SimAI
application.

BENEFIT: The SimAI platform predicted eight new cooling designs in less than one minute.

/ Power Inverter Cooling

CHALLENGE: With increasing power densities, classical thermal designs are not enough. New cooling systems,
including cold plates with nonconventional pin/fin designs, are needed to meet thermal requirements.

SOLUTION: About 30 Fluent CFD simulations were used to create an AI model of a forced convection cooling channel
installed on top of a hot power inverter, with large topology changes in the design of the pins.

BENEFIT: The SimAI application predicted a new cooling channel in 30 seconds.

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 1 3

/ Ansys SimAI Software in Action: Additional Use Cases

/ Centrifugal Pump Design Exploration

CHALLENGE: Centrifugal pump design exploration involves many parameters, including the number of blades and
leading and trailing edge angles at the hub and shroud.

SOLUTION: Approximately, 113 Fluent simulation results were used to create an AI model in the SimAI platform of the
centrifugal pump, with topology changes in terms of number of impeller blades and leading and trailing edge angles at
the hub and shroud.

BENEFIT: The SimAI application generated pressure and velocity field predictions in 30 seconds.

/ Solid Suspension in Stirred Tanks

CHALLENGE: During processing in stirred tanks, the distribution of the solid phase affects mixing scales, the availability
of solids to chemical reactions, and the overall performance of the tank.

SOLUTION: Approximately 28 liquid-solid multiphase Fluent simulations were used to create an AI model in the
platform of stirred tanks with topology changes, including the number of impellers, agitation rate, and the percentage
of solid loading variations.

BENEFIT: The SimAI application predicted the solid volume fraction field in 30 seconds.

/ Shell and Tube Heat Exchanger

CHALLENGE: Designing an efficient yet economical heat exchanger in a timely manner

SOLUTION: Using Discovery software and the SimAI application, an AI model of a shell and tube heat exchanger with
topology changes, including the number of tubes, baffles and their orientation, relative positioning, and inlet coolant
velocity, was created based on 250 conjugate heat transfer (CHT) simulation results.

BENEFIT: The application predicted new heat exchanger geometry in less than one minute, reducing computational
time by 99%.

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 1 4

/ Ansys SimAI Software in Action: Additional Use Cases

Electromagnetics and Optics

/ Permanent Magnets in Consumer Electronics

CHALLENGE: Sizing and selecting materials for various electromagnetic (EM) functions while reducing cost.
SimAI Maxwell 3D Difference
SOLUTION: Up to 100 simulations from Ansys Maxwell® electromechanical device analysis software were used to create AI models Magnitute of magnetic flux density on the container
of different fidelities in the platform to predict magnetic field distribution and magnetic forces. wall due to a cylindrical magnet

BENEFIT: The application evaluated each new design for one magnet in 20 seconds versus five minutes with Maxwell simulation.

/ Predicting Traction Motor Performance for Electric Powertrains

CHALLENGE: Conducting rapid predictions of motor EM torque, efficiency, losses, and forces/stresses with design changes.

SOLUTION: Up to 100 Maxwell simulations were used to create AI models in the platform of different fidelities.

BENEFIT: Each new EM design was evaluated in minutes — 20X to 50X faster than conventional simulation techniques.

/ Optical Systems in Harsh Environments

CHALLENGE: Automotive and aerospace optical design need to consider diverse environmental conditions, including factors such
as fog and haze.

SOLUTION: Using Ansys Speos® optical illumination design and simulation software and the SimAI application, 10 existing optical
simulation datasets representing various environmental conditions with input variables such as fog density and haze levels were
used to create an AI model.

BENEFIT: The application enabled the team to explore more operational environments in less time with an accurate illuminance
field prediction and just a 2% error in the calculation of the max illuminance for an unseen new design.

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 1 5

/ Get Started with Ansys AI Products Today Powering Innovation That Drives
Human Advancement

Ansys is dedicated to advancing the customer experience, democratization of simulation, and next generation of When visionary companies need to know how
innovation with artificial intelligence (AI). their world-changing ideas will perform, they
close the gap between design and reality
From enhancing Ansys simulation with the power of AI to enriching AI with the predictive accuracy of Ansys simulation, with Ansys s imulation. For more than 50 years,
the mutually beneficial relationship between AI and simulation is unlocking new possibilities for engineers and Ansys software has enabled innovators
designers across industries and engineering disciplines. across industries to push boundaries by
using the predictive power of simulation.
/ Watch a Webinar From sustainable transportation to advanced
semiconductors, from satellite systems to life-
• Introducing Ansys AI: A New Chapter in Engineering Simulation
saving medical devices, the next great leaps in
• Introducing Ansys SimAI: Cloud-native Generative AI for Simulation human advancement will be powered by Ansys.
• Engineering in the AI Era
• Accelerate Your Simulations Webinar Series Ansys and any and all ANSYS, Inc. brand,
product, service and feature names, logos
/ Read an Article and slogans are registered trademarks or
• Explaining SimAI: How AI is Applied to Numerical Simulation In Practice trademarks of ANSYS, Inc. or its subsidiaries in
the United States or other countries. All other
• From Chips to Ships: Optimize Design With the Ansys SimAI Platform
brand, product, service and feature names or
• Maximize Simulations With Secure, Cloud-native AI trademarks are the property of their respective
• The Intersection of AI and Simulation Technology owners.

/ Stay Informed ANSYS, Inc.

• Visit and bookmark Ansys AI — AI-augmented Simulation Technology Southpointe
2600 Ansys Drive
• Visit and bookmark Ansys SimAI | Generative AI for Accelerated Simulation
Canonsburg, PA 15317
• Sign up to get Ansys AI updates emailed directly to you U.S.A.
• Check out the Ansys Advantage blog and never miss a new post with a subscription 724.746.3304
• Explore Ansys Advantage magazine, subscribe, and look out for our upcoming AI issue
Visit www.ansys.com for more information.

© 2024 ANSYS, Inc. All Rights Reserved.

/ COM BINE AI WITH SIMUL ATIO N FO R NEXT-LEVEL I N N OVAT I ON // 1 6