Simulation, FEA & Design-for-Manufacture

Welcome to the workshop. I'm Cad, your tutor, and for the next six modules we're going to turn your beautiful-looking parts into parts that actually survive and parts a factory will actually make without phoning you in a panic.

This course is for founders — you're building real hardware inside your Kaern startups, and you can't afford a bracket that snaps on a customer's desk or a housing that no injection moulder will quote. By the end you will be able to:

• Run a basic Finite Element Analysis (FEA) in cycleCAD and trust — or distrust — the colours it shows you.
• Set up sensible loads, constraints, and meshes instead of guessing.
• Read stress, displacement, and safety factors like a structural engineer.
• Redesign a part so it's cheaper, faster, and more reliable to manufacture.
• Pick materials and tolerances that don't blow your bill of materials.

We work hands-on. Every module has a worked example you'll follow along in cycleCAD, an exercise you do yourself, the mistakes I see founders make every single cohort, and a quick check before we move on. Bring a part from your own startup — we'll be analysing it by Module 6.

Let's get our hands dirty.

Module 1 — Why Simulate (and When Not To)

Learning objectives

• Explain what simulation buys you versus building-and-breaking physical prototypes.
• Identify which design questions FEA can and cannot answer.
• Estimate the cost of a late-stage failure versus a 20-minute simulation.
• Set up a simulation-ready project file in cycleCAD.

Lesson 1.1 — Simulation is a cheap argument with reality

Teaching script (Cad's voice):

Imagine you're packing for a flight and you sit on the suitcase to test the zip. That's a physical prototype — honest, but slow and you only get one suitcase. Simulation is like picturing the zip straining before you've even packed: cheap, repeatable, and you can try ten zips before lunch. Why does this matter to you as founders? Because your runway is measured in weeks, and every physical prototype is money plus shipping plus three days waiting for the courier. A simulation in cycleCAD costs you twenty minutes and electricity. Now — simulation is an argument with reality, not reality itself. It's only as good as the loads, materials, and constraints you feed it. A confident-looking colourful plot built on a wrong assumption is more dangerous than no plot at all, because it gives you false courage. So we treat every result as a hypothesis to be sanity-checked, never as gospel. The founders who win aren't the ones who simulate the most — they're the ones who know which questions are worth the twenty minutes. So let me ask you: of the parts in your current product, which single failure would embarrass you most in front of a customer?

Lesson 1.2 — What FEA can and can't tell you

Teaching script (Cad's voice):

Here's an analogy I love. FEA is like a weather forecast. It's brilliant at "will it rain roughly this afternoon" and useless at "will this exact raindrop hit my nose." FEA tells you regions of high stress, rough deflections, where things will likely fail first. It does not hand you the precise newton at which your part cracks on a Tuesday — real materials have flaws, weld lines, scratches, and bad days. Why does this matter? Because founders over-trust the third decimal place. They'll tell an investor "our bracket holds 412.7 N" when the honest claim is "around 400 N, comfortably above our 150 N requirement." The forecast framing keeps you humble and keeps your spec claims defensible. FEA is also blind to things you didn't model — if you forgot the vibration from shipping, the simulation can't warn you about a fatigue crack it never knew existed. So our discipline is: list every load before you open the solver, and label which ones you're choosing to ignore and why. What loads does your part actually see — not in the lab, but in the messy hands of a real user?

Worked example in cycleCAD

We open cycleCAD, create a new Simulation Study on a sample L-bracket. We don't solve anything yet. Instead we use the Study Setup panel to document: material (mild steel), the load case ("150 N downward at the tip, from a hanging tool"), and the fixed face ("bolted to wall"). I show how cycleCAD's Assumptions Log lets you write "ignoring shipping vibration — to be revisited in Module 6." We save this as bracket_v1.ccad. The point: a study starts as written assumptions, not colours.

Hands-on exercise

Open cycleCAD, import any part of yours (or the provided demo_bracket). Create a Simulation Study and fill in the Assumptions Log with: one material, one real-world load with its source, one fixed feature, and two loads you are deliberately ignoring. Do not solve. Save as yourname_module1.ccad.

Common mistakes

1. Treating a colourful plot as a guarantee instead of a forecast.
2. Quoting simulation numbers to false precision (412.7 N) in pitches.
3. Forgetting to list ignored loads, so they silently disappear forever.

Check for understanding

1. Give one design question FEA answers well and one it answers badly.
2. Why do we write the Assumptions Log before solving?
3. Your part survives in simulation but cracks in the field — name two reasons FEA might have missed it.