The Challenges of Using AI in Computational Engineering
with Jaideep Bangal of Altair
Artificial Intelligence (AI) is gradually permeating the engineering, additive manufacturing, and broader manufacturing industry, initially focusing on process simulation and optimization. However, its adoption in the design aspect of engineering has been somewhat slower due to a lack of commercially available software tools and usable training data.
Altair, a long-standing leader in simulation-driven design, is well-positioned to pioneer the AI-driven design space, thanks to their data-centric approach.
Jaideep Bangal, the Director of Simulation Driven Design and Manufacturing at Altair, will be showcasing Altair’s DesignAI at the upcoming CDFAM Computational Design (+DfAM) Symposium.
Following is an overview of his work at Altair, the challenges of using AI in computational engineering and what you can anticipate from their new AI engineering tools.
Tell us about your role at Altair, the products you have worked on in the past and are currently working on, along with an overview of your responsibilities and areas of focus within the company.
As the leader of a global team of experts at Altair, I focus on technical sales and strategy for simulation-driven design and manufacturing solutions.
My team helps engineers implement advanced design and manufacturing methods during product development to generate efficient designs while predicting and fixing manufacturing problems.
With my expertise in simulation, generative design, manufacturing, additive manufacturing, and computational fluid dynamics (CFD), I have worked on many products that have helped organizations improve their design and manufacturing processes.
These products include Altair® Inspire™, Altair SimSolid®, and, more recently, Altair® DesignAI™.
Inspire is a powerful yet easy-to-use generative design and optimization software that helps engineers quickly generate and explore structurally efficient concepts.
SimSolid is a structural analysis software that can analyze complex parts and assemblies quickly and accurately.
DesignAI is an artificial intelligence (AI)- and simulation-driven design tool. It combines physics-based, simulation-driven, and machine learning-based designs to create high-potential designs earlier in development cycles.
My skills and expertise have helped Altair improve its market position, develop innovative products, and build strong customer relationships. I am proud of the work my team, and I have done over the years to help over 1,500 companies take their first step into the simulation world, which allows them to create efficient designs, increase productivity, and realize higher profits.
Your presentation title at the upcoming CDFAM symposium is ‘The Challenges of Using AI in Computational Engineering.’ Please provide us with an outline of the topics you will address during your presentation.
As digital engineering advances, modeling, and simulation will converge with machine learning, AI, and high-performance computing (HPC) in solving the world’s most complex problems.
During my presentation at the upcoming CDFAM symposium, I will discuss how AI is used in computational engineering, including how it generates and optimizes designs, reduces simulation times, identifies optimal solutions, and predicts material behavior. AI-augmented computer-aided engineering (CAE) helps manufacturers discover machine learning-guided insights, explore new solutions to complex design problems through physics and AI-driven workflows, and achieve greater innovation through collaboration and convergence.
Additionally, I will discuss the challenges present in AI adoption, such as the difficulty of collecting meaningful data for training machine learning algorithms and capturing and understanding design intent.
I will also highlight specific AI tools and technologies that can be used in computational engineering, such as Altair® romAI™ for generating reduced-order models (ROMs), Altair® physicsAI™ for quick physics predictions, and DesignAI for combining physics-based, simulation-driven design with machine learning-based AI-driven design.
One of the major obstacles in adopting AI in engineering is the challenge of collecting meaningful data about design intent and resultant performance to train machine learning algorithms. What are your thoughts on how we can overcome this challenge?
Capturing and understanding design intent is crucial for developing machine learning models that can predict performance and optimize designs.
One of computational engineering’s challenges is dealing with large design spaces. AI can alleviate this burden by automating the design exploration process and identifying optimal solutions.
However, for AI to be effective, it needs access to large volumes of high-quality data. Additionally, the design space’s complexity can create computational challenges regarding the time and resources it takes to explore and evaluate all possible design options.
One potential solution to the challenge of collecting data for AI training is to leverage existing simulation tools and models. By using fast and accurate simulation software, users can generate large volumes of valuable data they can then use to train machine learning algorithms.
Can you provide an overview of which of Altair’s software applications integrate AI functionality first and what unique value they will give that cannot be achieved with existing methods?
physicsAI is a new tool designed to predict physics outcomes quickly. This technology leverages historical simulation data to deliver fully animated physics predictions in a fraction of the time it takes traditional solvers to do the same. Unlike previous machine learning technologies, physicsAI uses cutting-edge geometric deep learning to operate directly on meshes and CAD models, which generates even faster results.
DesignAI combines physics-based simulation-driven design and machine learning-based AI-driven design to create high-potential designs earlier in development cycles. With it, users can augment current product development practices and multiply the productivity of engineering teams with AI. DesignAI provides a cost-effective acceleration of engineering processes that integrate with existing engineering workflows and toolchains.
ROMs are a computationally efficient way to incorporate detailed 3D simulations into a 1D environment for system-level studies. Simulation tools like Altair® EDEM™ or Altair CFD™ offer detailed investigations of non-linear systems, but long simulation times typically limit the analysis to individual components or subsystems. romAI leverages 3D simulations as training data to generate dynamic ROMs, requiring fewer simulations than traditional methods. romAI works with any solver and provides accurate results within and outside the training space. The same machine-learning technique for generating ROMs can also be used for system identification from test data.
Altair RapidMiner – Altair’s data analytics and AI platform – reduces traditional data analytics and machine learning application barriers. Designed to be accessible for all user levels while remaining scalable, governed, and safe, Altair RapidMiner’s low-code platform enables users to develop data pipelines and machine learning models without needing in-house data science expertise.
In which applications or industries do you anticipate the first adoption of AI design tools and technologies? It would be great to get a sense of the potential impact of AI on these sectors, as well as any examples of AI adoption and their impact on processes, products, or services.
AI tools and technologies are already well used within various industries, including manufacturing, healthcare, finance, automotive, aerospace, and more.
AI’s potential impact on these sectors is immense, from improving quality control and predictive maintenance to reducing production times and costs. For instance, the aerospace and defense sectors use AI to streamline the design process, reduce manufacturing errors, and optimize supply chain operations. AI can also monitor aircraft components’ health in real-time, allowing organizations to perform predictive maintenance, reduce downtime, and improve safety.
The successful adoption of AI in manufacturing depends on several factors, including data quality, access to HPC resources, and the ability to integrate AI technologies with existing manufacturing processes. However, the potential benefits of AI for manufacturing are significant; we already have several customers in this sector that utilize our AI solutions and expect continued growth in this area in the coming years.
Altair was one of the pioneers in providing commercially available topology optimization software packages to the engineering market, which has helped accelerate the interest in adoption due to additive manufacturing. However, the most viable business applications are casting and forging rather than in AM. Do you expect to see a similar trend in the adoption of AI, where AM may create flashy images and headlines, but other processes will be used to manufacture the functional parts?
Remembering that AM is still a relatively new and emerging technology is essential. On the other hand, casting and forging are well-established manufacturing processes with a long history of successful applications. As a result, it is not surprising that we are seeing more viable business applications for AI in casting and forging, where the technology can be used to optimize existing processes, reduce waste, and improve product quality. Essentially, more things are made using traditional processes than additive manufacturing, and as such, the potential benefits of improvements to high-production-volume processes will likely be more significant.
However, as the additive manufacturing industry matures and evolves, we may see new AI applications tailored to this technology’s unique requirements and capabilities. Ultimately, the success of AI in additive manufacturing will depend on the industry’s ability to innovate and improve while also addressing the technical and regulatory challenges currently limiting its widespread adoption.
We have seen more advanced design tools, such as topology optimization and computational and generative design, moving from early adopters in AM to more established manufacturing methods because the manufacturing infrastructure, procurement, and validation processes are all mature and accepted, with fewer bureaucratic and political hurdles. What will it take to achieve the same level of acceptance for additively manufactured parts, or should we focus on existing processes?
The road to broader acceptance of additively manufactured parts is multifaceted and will require several actions from different players in the industry.
One critical aspect will be increasing the standardization and certification of additive manufacturing processes, materials, and components.
Education and awareness will also be vital in driving the acceptance of additively manufactured parts. Designers, engineers, and other stakeholders must be educated on the capabilities and limitations of additive manufacturing and how to design for the technology to maximize its potential.
While focusing solely on existing processes may be tempting, it is essential to recognize that additive manufacturing can offer significant advantages in specific applications and industries. Instead, efforts should be made to bridge the gap between the current state of additive manufacturing and the requirements for mainstream adoption.
At Altair, we have seen faster progress with AI in more established manufacturing processes. AI is being used more in established manufacturing processes because of the existing infrastructure and procedures. This makes integrating AI into existing workflows and data systems easier.
That said, additive manufacturing is still a relatively new technology, and as it continues to mature and gain acceptance, we can expect to see more widespread adoption of AI and other advanced digital technologies in the additive manufacturing space as well.
With your expertise in computational design, engineering, and advanced manufacturing, we are thrilled to have you speaking at CDFAM. What topics or areas of interest are you most excited to learn and engage with during the event?
As an individual with expertise in computational design, engineering, and advanced manufacturing, I am delighted to participate in CDFAM and engage with fellow professionals in the industry.
I particularly want to hear from Nathan Shirley about the untapped potential of design engines and additive manufacturing.
I’m also looking forward to Professor Neil Gershenfeld’s insights on the roadmap to replicators. I’m confident each topic from the event’s distinguished speakers will be interesting, and I am excited to gain the knowledge and insights every discussion will surely deliver.
Register to attend CDFAM to hear from Jaideep and other leading experts integrating AI in engineering and advanced manufacturing.
Supporting Sponsors
