AI is reshaping engineering simulation

AI features are moving into mainstream simulation tools, Source: Gemini

Over the last couple of years, AI has moved from being an add-on idea to a built-in feature of mainstream simulation tools. 2024 in particular has been described as a turning point, with several major vendors rolling out AI-enabled capabilities directly inside their CAE platforms.

An Engineering.com round-up of “10 AI-powered simulation features to pay attention to in 2024” frames this shift as the start of a new era, where AI and simulation converge to help engineers explore designs faster, automate routine work and connect models to digital twins. McKinsey, in partnership with NAFEMS, has also reported that time to market has overtaken pure performance as the main value driver for simulation, and that there is strong interest in AI/ML-enhanced tools to support that push. What kind of AI is actually being added to simulation tools?

The recent wave of features is less about “AI designing everything for you” and more about supporting and accelerating what engineers already do. Examples include:

• Surrogate and reduced-order models: Training AI models on existing simulation data to approximate results much faster, enabling rapid design-space exploration and “what-if” studies.

• Automated workflows and pre-processing: AI to recognise recurring model patterns, suggest boundary conditions, or auto-configure meshing and solver settings for similar classes of problems.

• AI-assisted design exploration and generative design: Tools that propose geometry variants meeting certain performance targets, then use simulation (or surrogates) to rank and refine them.

• Simulation-aware digital twins: Embedding lightweight models into digital-twin frameworks so that real-time data from the field can be combined with physics-based predictions.

• AI assistants integrated into simulation platforms: For example, AnsysGPT, an AI-powered virtual assistant that helps users with support and documentation, and Ansys SimAI, which uses AI to accelerate simulation workflows. Altair’s HyperWorks 2024 release similarly emphasises AI-embedded design and simulation workflows, backed by HPC.

Alongside this, initiatives like Rev-Sim’s free “AI in simulation” learning programme show that the simulation community is actively trying to build shared best practice rather than experimenting in isolation.

How this changes the role of simulation

As AI features become standard, simulation is shifting in three noticeable ways:

• Faster loops, more scenarios: AI surrogates and automation mean engineers can evaluate many more design options in the same amount of time, which matches the growing emphasis on time-to-market and agile development.

• Wider access beyond specialist analysts: Simulation apps and AI-supported interfaces are making it easier for non-experts to run structured, pre-defined studies while still relying on validated underlying models.

• Closer connection to operations: When simplified or AI-accelerated models are embedded in digital twins, simulation stops being just a design-phase activity and starts to inform operational decision-making in real time.

None of this removes the need for high-fidelity, carefully validated models, if anything, it increases the importance of getting the physics right at the core.

Opportunities and caveats

There are clear benefits:

• quicker design exploration,

• reduced manual pre- and post-processing,

• better use of historical simulation data,

• and more consistent workflows.

But there are also questions that the industry is still working through:

• How do we validate AI-based surrogates and know where they are reliable?

• How do we avoid misusing AI tools outside the range of physics they were trained on?

• How do small and mid-sized engineering teams decide which AI features actually add value, versus what is marketing noise?

Most of the serious conversations now position AI as a complement to physics-based simulation, not a replacement.

Mansim’s perspective

Within this landscape, Mansim’s approach is to treat AI as an additional tool in the simulation toolbox, especially where it can enhance, not dilute, rigorous engineering judgement.

As Dr Sorosh Mirfasihi, Head of Consultancy at Mansim, puts it:

“AI won’t replace physics or engineering fundamentals – but it can absolutely change how quickly we get from a question to a reliable answer. At Mansim, we’re experimenting with AI-enabled workflows around things like design exploration and surrogate modelling, while keeping high-fidelity CFD and careful validation at the centre. The aim is not to automate the engineer out of the loop, but to give them better tools so they can focus on the hard decisions.”

For organisations navigating this transition, the key will be combining trusted physics models, good data practices and careful use of AI, so that simulation remains a decision-quality tool even as it becomes faster, more automated and more widely used.