Take your engineering game to the next level! Join engineers worldwide in harnessing the power of engineering automation to boost productivity and achieve superior results. In less than 5 minutes, unlock new possibilities and elevate your productivity. These are where engineering tools are being used around the globe. Ready to join us and revolutionize your approach? Let's engineer success together.

The figure above displays the global distribution of Engineering Tools users. The size and color of the dots represent concentration.

by Ian Rivard, SeStructures LLC — May 2026

I get asked variations of the same question all the time. “What size model should my workstation handle?” “Are we behind the curve running half-million-node decks?”

I’ve been doing FEMAP scripting and structural FEA for over two decades now, working with aerospace primes, new space, and defense contractors. Engineering Tools has a lot of users across that spectrum, and over the years I’ve talked with a lot of them about how they work. None of these observations are scientific — but the patterns hold across companies and across tool types.

Here’s what I see in 2026, and how it’s shifted since I started.

Two industries, one toolkit

Before I dig into the observations: the most useful frame I’ve landed on is that the structural FEA community I see is really two distinct communities sharing a lot of the same software.

On one side: traditional aircraft and air-vehicle work — the primes, the legacy programs, the lifing-and-fatigue houses. Static stress, fatigue analysis, classical structural certification. Workflows that have been refined over forty years, locked in by DERs and validation requirements. Slow to change tools because the tools are part of the certification basis.

On the other side: new space — satellite primary structure, launch vehicles, smaller aerospace shops, the constellations. Heavily dynamics-driven (random vibration, modal survey, shock response). Younger workflows, less legacy weight, more willing to adopt new tools because they’re not validating against forty years of process.

These two communities answer almost every question about “the industry” differently. When I say something like “frequency response is the dominant workflow,” that’s true and not true depending on which side of that line you’re standing on. The new space side runs modes-and-random for a living. The aircraft side runs static stress and fatigue for a living. Both are right; they’re just doing different jobs.

The honest disclaimer about my own data: the user base that adopts free FEMAP scripting toolkits skews toward the people willing to try new things. That’s disproportionately new space. The aircraft side of the house tends to use more conservative, older workflows by design. So when I report “what I see,” I’m seeing more of what new space does, and somewhat less of what aircraft does. Adjust accordingly for your own context.

With that in mind, here’s what’s shifted.

Models have gotten bigger — but not in the way you'd expect

When I started, a “real” production model was 50,000–80,000 nodes. A half-million-node model was a flagship wing-box or a full satellite primary structure, and you ran it overnight.

Today, the boring-day-to-day model is more like 100,000–200,000 nodes. The flagship has crept up to a couple million. Roughly one in four sessions I see involves a model with more than 500K nodes. That used to be “the big one.” Now it’s just the Tuesday job.

This trend hits both communities, but for different reasons. New space gets bigger because the models are getting more comprehensive — fewer sub-models, more whole-vehicle FEMs. Aircraft gets bigger because the detail in critical zones (joints, splices, cutouts) keeps growing.

What isn’t changing is that elements are still mostly Quad4 plates. Shell-dominated structural FEA hasn’t been disrupted by all the compute we’ve thrown at it — engineers just put more shells in. The “let’s switch to all-solid second-order tets” revolution that everyone predicted ten years ago hasn’t really happened in production work. Tet10 is up, but Quad4 still outnumbers every other element type by an order of magnitude in the work I see.

If you’re sizing a workstation, the rule I tell people is: 32 GB of RAM is the floor for anything serious, 64 GB is the typical sweet spot, and 128 GB is normal for the heavy users I work with. I have people running 256 GB on truly large models. The crossover happened maybe four or five years ago.

Composites are everywhere now — and the tooling hasn't caught up

The biggest workflow shift I’ve watched, by a wide margin, is the spread of composites. A decade ago, composite layups were a specialty — you saw them at the big aerospace OEMs, a couple of motorsport teams, maybe a wind-turbine shop. Most engineers I worked with did metallics and never touched a layup property card.

Today, somewhere around 40% of the structural models I see have at least one composite layup defined. Aerospace structure is using it. Automotive structure is heading there. Even bracket and fitting work is increasingly hybrid (sandwich panels, bonded inserts, co-cured stiffeners). New space adopted composites aggressively because mass-fraction matters more than process-cost. Aircraft adopted them too, but more selectively and more recently — the certification basis for composite primary structure took longer to solidify.

The native tooling around composites has not kept up with this. Most of the composite-specific work in FEMAP still requires either hand-flying through dialogs or scripting. If you’re a metallics-trained analyst still avoiding the composite property cards, you’re already behind. That gap is going to widen.

Frequency response is the silent majority — for new space

This is the observation most likely to start a fight in a comments section, so let me caveat it carefully.

In the work I see, frequency response and modal analysis together are the dominant workflows — by a significant margin over linear static. Random vibration is the next-biggest. Linear buckling shows up regularly. Nonlinear-static is a real but minority workflow. Heat transfer? In four years of conversations with FEMAP users, I’ve talked to maybe three people who were doing serious thermal in FEMAP.

But that’s heavily new-space-flavored data. The traditional aircraft side of the industry runs a very different mix: linear static for the bulk of stress work, fatigue analysis (often outside FEMAP proper), occasional modal for flutter and ground-vibration, and the nonlinear cases where they matter. Static stress isn’t dying — it’s still the foundation of every certification report on every production aircraft. It’s just that the analysts doing it tend not to be the early adopters of free FEMAP scripting toolkits.

What that means practically: if you’re training the next generation of analysts, the answer depends on which industry you’re training them for. New space wants modal mass participation, mode tracking, frequency-response post-processing, and PSD analysis. Aircraft wants static stress, hand-checks, fatigue spectra, and a deep relationship with the certification basis. Both are legitimate career paths. Neither is going away. The mistake is assuming your industry’s defaults are universal.

FEMAP versions: a slow march

I see users on every FEMAP version from 11.0 up to the current release. That’s over eleven years of versions still in active use, in the same engineering community. The reasons are typical: corporate IT freezes, license constraints, validation requirements that lock a project to a version, sheer inertia.

The versions that dominate today’s work are the 2022.x, 2301, 2306, 2401/2406, and 2506 families. Older versions are slowly aging out, but slowly. Aircraft tends to lag a release or two behind new space — again, certification stickiness.

What I'd tell a new structural analyst

• Get good at composites early. Both sides are heading there; it’s not optional anymore.
• Don’t chase faster solvers; chase smaller models. Most of the speedup wins of the last decade have not come from buying more cores. A clean 200K-node model that converges to the right answer beats a sloppy 2M-node model every time. Solve your model in core memory. Use M.2 fast drives — this is one of the biggest speed boosts since we left spinning disks behind.
• Learn the workflow your industry doesn’t default to. If you’re in aircraft, take a modal class. If you’re in new space, run a classical fatigue spectrum just to know what it feels like. Cross-training makes you more valuable on the boundaries, where the interesting work tends to live.
• Get scripting fluent. Whether it’s the FEMAP API, a Python toolkit on top of bulk decks, or the Engineering Tools macros — hand-flying through dialogs is no longer competitive. The analysts who scale to large models scale because they automate.