Plenaries

Failure Modes Come First: Simulation Driven Design Beyond Verification

Dr. Lukasz Skotny

CEO / Lead Designer / FEA Course Creator at Enterfea

Very often, FEA, and even basic engineering design is used way too late in the design process. This is easily apparent by the quality of models and their details we receive for analysis (and subsequent issues we discover that are hard to fix due to time constraints, and things already being done). But at the same time, we mostly use FEA to "verify" a given solution. This is of course a powerful option, but it assumes that we already know the solution, and we just want to confirm that it works.

It's way better to design for possible failure modes, starting from simply models toward more complex ones. But this requires experience, and understanding of the structure we're designing. In this keynote, I will showcase some typical approaches to FEA (and their risks), what needs to be understood to fix this, and how to more effectively gradually solve problems for faster and better solutions in the FEA itself.

This approach can really help to "build solutions using FEA", rather than just "check with FEA if a given solution works" which is critical in the simulation driven design.

LLMs, Agents and Engineering Simulation: What Actually Works Today?

Dr. Marcin Wierszycki

Managing & Technical Director at Technia Poland

Large Language Models (LLMs) are rapidly changing the way engineers work. However, the field is evolving so quickly that it is often difficult to separate real capabilities from marketing claims.

This presentation will provide a practical introduction to modern AI tools and agent-based workflows from an engineering perspective. We will explain what LLMs are, how different models and providers compare, and why concepts such as model size, quantization, context windows, and tool integration matter in practice.

The session will cover how AI systems interact with engineering tools, databases, documents, code, and simulation environments through technologies such as Model Context Protocol (MCP) and tool calling. Real-world examples will demonstrate where AI can already support engineering work, including simulation setup, automation, post-processing, reporting, and knowledge management.

The goal is to provide engineers with a realistic understanding of what can be done today. The presentation will discuss the current limitations of AI systems as well.

Conference home pageKeynotesSpecial sessionsVenueAbstract submissionGuide for attendeesRegistrationSponsorship opportunities

Media Sponsor

Conference cancellation policy

Presenters of the seminar can submit their manuscript to the NAFEMS Journal EMAS, and have the standard publication fee waived.