From requirements to a shortlist you can defend
Material selection is where most projects quietly go wrong — a grade chosen because it was used last time, and discovered to be wrong at qualification. Start from the constraints instead.
Selection is a window, not a lookup
"Under 3 g/cm³ and above 300 MPa" is a box drawn on a property chart. What falls inside it is the shortlist; everything else is out, however familiar it is. The selector does this across far more properties than two — but this is the shape of it.
What you put in
Prose, not a form. "A seawater pump impeller running at 60 °C with cavitation, welded repairs in the field, and no more than €40/kg" is a complete brief — function, environment, constraints and the property targets that follow.
| Function | What the part does and how it is loaded — static, cyclic, impact, thermal cycling. |
|---|---|
| Environment | Medium, temperature range, pH, chlorides, UV, radiation, contact with other metals. |
| Property targets | The numbers that must be met: strength, stiffness, toughness, conductivity, service temperature. |
| Constraints | Manufacturing route, joinability, mass budget, cost ceiling, regulatory or standards requirements. |
What you get back
| Ranked candidates | Three to six materials, best fit first, with the reasoning for the ranking made explicit. |
|---|---|
| Property ranges | Realistic ranges in SI units with the condition attached — not a single number pretending to be exact. |
| Trade-offs | What each candidate costs you. The one that wins on strength usually loses somewhere else. |
| Processing and joining | How it is made, how it is welded or bonded, and the traps in each route. |
| Sources | Each database, paper or dataset consulted, plus an explicit entry for anything answered from general knowledge. |
Why the ranking is argued, not asserted
Each candidate comes with the property values that put it where it is and the weakness that keeps it from being obvious. Sources are listed per result: where a number came from OQMD it says OQMD, and where it came from published typical ranges it says so and tells you to verify.
- 01 Brief Function, environment, constraints
- 02 Screen Property windows across the field
- 03 Rank With trade-offs made explicit
- 04 Verify Against databases and your own tests
Every candidate can be looked up in the materials library, its class behaviour, and — through structure search — the computed phases behind it. The shortlist is the start of the argument, not the end of it.
"Bracket, aluminium or better, 200 °C continuous, salt spray exposure, bolted to CFRP, needs to be machinable in small batches, mass matters more than cost."
"What is the best metal?" — nothing to rank against, so nothing worth ranking.
- AI materials chat — for the why
- Material creator — when nothing fits
- Materials library — the reference data
Questions about the selector
Does it just pick from a fixed list?
No. The 37-entry library on this site is reference data you can browse, but the selector reasons over the whole materials space and queries live structure databases while it works. Candidates outside the library are normal.
Why a shortlist instead of one answer?
Because material selection is a trade-off, not a lookup. A single recommendation hides the alternative that would have been better once you weighed cost or joinability differently. The shortlist makes the trade visible so you can make the call.
Can I export the comparison?
Yes. On the Pro and Team plans results are saved to your history and export into a report you can take into a design review.
How accurate are the property ranges?
They are typical published ranges for orientation, and they behave like datasheet ranges: heat treatment, section size and product form move them. Use them to narrow a field, then confirm against the standard and the supplier's certified data.
Put an engineer's AI on the problem
Free plan, no card required. 20 AI requests a month, and the whole knowledge base stays open to everyone.