Finding your steel

The catalog is 2,060 grades#

The built-in catalog holds 2,060 grades. 2,059 are enabled; exactly one, 20mnb4, is currently disabled by a platform administrator. They are sorted into 43 families and 68 alloying-system groups, and by alloy type they break down as 1,907 steels, 106 stainless steels and 47 cast irons.

That is worth stating plainly because the twenty-grade table on the material catalog invites the opposite conclusion. Those twenty are a documented reference set: the grades Phases falls back to when the live catalog is unreachable, and — separately and correctly — the twenty grades resistance spot welding supports. They are not the catalog. Tool steels alone — general, hot-work, cold-work, plastic-mould and high-speed — come to 470 grades; structural and general engineering steels come to 1,091 between them.

So the useful question is rarely "is my steel supported". It is "which of 2,060 rows is my steel", and that is what this page answers.

Where the catalog lives#

Material Base in the left sidebar opens the workbench at /ws/<workspace>/materials. Its header reads Material base over the live count — something like 2,059 catalog grades · 3 in this workspace — so the number on screen is always the real one, not a number from a document. Three scopes sit across the top:

• Material — browse the catalog and inspect one grade
• Compare — put several materials side by side
• Family data — aggregate a whole family

Material Base needs a workspace, so guest users do not have it. See guests and limits.

Search first, facets second#

The Material scope leads with a search box, and two facets under it. With two thousand rows in the list, search is the way in and the facets are for narrowing afterwards.

Section headings in the list change meaning with the filter: while the Family facet is on All materials they are family names, and once you pick a single family they subdivide it by alloying group.

Two things that will otherwise look like missing data#

The Alloy type facet is not a filter you opted into. It defaults to the alloy type of whatever material is currently selected, and a family only exists inside one type. Land on the page with a steel selected and the 106 stainless grades and 47 cast irons are simply not in the list — not filtered out with a message, just absent. If 316L or a cast iron appears to be missing, change that dropdown before concluding anything.

Your workspace materials sit under their own alloy type. They are grouped as Workspace materials, which means they are equally invisible while the facet says Steel. Same fix.

A third, smaller one: family labels are imperfect and the product knows it — a grade can sit under an unhelpful label such as Other, where nobody would think to look for it. The Family facet therefore carries an All materials option that lists every grade of the chosen alloy type, so a badly filed grade stays reachable without knowing its name.

Naming a grade: what actually matches#

Two different lookups are in play, and they do not match on the same things.

Almost all of that work is done by the label, not by aliases. 2,059 of the 2,060 grades carry aliases, but for 2,034 of them the aliases are internal source keys — S355N carries 4884 and 4884_s355n — which nobody would type off a drawing. Only 26 grades carry anything else, and most of those are card filenames such as mat_DC04. Genuine cross-naming-system aliasing is the exception rather than the rule: 316L is the real example, answering to 316L, 1.4404, X2CrNiMo17-12-2 and S31603.

What actually makes a Werkstoffnummer or an AISI number work is that the catalog frequently files the steel under that number as its label — 398 rows are labelled as a plain Werkstoffnummer, 22 as a bare AISI/SAE number. So type the designation you have, but expect it to find the row that was named that way, not a cross-reference from some other naming system.

Do you have 1.2344?

Load 42CrMo4 and show me its composition.

Five limits on that, none of which the product hides:

• Matching is exact after case folding, not fuzzy. 42crmo4 resolves; 42 CrMo 4 does not reach the catalog lookup, because the catalog step only lowercases — it does not strip spaces or punctuation. Nor does it normalise a standard's own punctuation: X40CrMoV51 is a row, X40CrMoV5-1 is not. Type the designation the way the catalog prints it.
• Supplier names are stored with the umlaut dropped, not transliterated. The catalog holds 35 BOHLER … rows and 17 Dorrenberg … rows, and no row anywhere — label or alias — uses the oe spelling. So BOEHLER K110 finds nothing and BÖHLER K110 finds nothing; BOHLER K110 is the row. Some trade names are also filed bare, without the maker: the Böhler M200 mould steel is simply M200.
• An alias will usually not rescue a missed spelling. Because only a couple of dozen grades carry a real cross-system alias, trying "the other naming system" mostly fails even when the steel is present under a different row. Search the browser instead of guessing again.
• Aliases are not searchable in the UI. The catalog payload the browser and the configuration-card dropdown read carries the id, label, alloy type, family and group — not the alias list. So the Material Base search box will not find a grade by its Werkstoffnummer unless that number happens to be the label — which, for 398 rows, it is. Search the browser by label; the handful of real aliases only help you in chat.
• A name that matches nothing is refused, not guessed. You get back No material was identified. Call list_materials() and load the chosen material with material_id=<builtin grade id> or tenant_material_id=<workspace material id>. 'name' on its own is only the display label for the session entry and selects nothing. The agent is instructed never to invent a grade to cover a miss, which is why a near-miss produces a question rather than a silently wrong steel.

When more than one row answers to a name#

Names resolve in a fixed order: this session's materials first, then your workspace's materials, then the built-in catalog. A workspace material named 42CrMo4 therefore shadows the catalog grade for everyone in that workspace — a good argument for naming workspace materials after the coil or the supplier rather than after the standard.

Inside the catalog, ids, labels and aliases all fold into a single lookup keyed by the lowercased string, so a string registered against more than one grade can only resolve to one of them, and the name alone will not tell you which. This is not hypothetical: 311 lookup keys in the current catalog answer for more than one grade. 4140 is one of them — it is both its own row and an alias of 1.1209.

Two distinct causes sit behind that, and they call for different responses.

Near-identical designations really are separate steels. S355N and S355J2G3 are two materials with two cards, not one grade written two ways.

The same steel is often filed several times, once per data source. 42CrMo4 and 1.7225 are two rows carrying an identical analysis — 0.42 C, 0.75 Mn, 1.10 Cr, 0.22 Mo — but different references, Saarstahl 1980 against HTDG 2008; 1.2344 and X40CrMoV51 are likewise identical at 0.40 C, 5.15 Cr, 1.35 Mo, from IMS 2001 and Bofors 1975. 4140 is a third take on the 42CrMo4 family, at 0.39 C, 0.82 Mn, 1.00 Cr, from Climax Molybdenum 1977. They are not cross-linked, so the designation you type decides which source's data you get. Where that matters, compare the candidates before running, and read the reference line on the card.

When it matters, confirm rather than assume:

• Ask Phases to echo the composition back, and read the base grade id in its reply.
• Open the row in Material Base and read the alloy-type and family badges in the detail header.
• Put the candidates side by side, in the Compare scope or from chat: Compare 42CrMo4 with 34CrNiMo6.
• After a run, the exact card used is stored with the run as mat_Steel.k. Catalog cards carry $ Reference: and $ Confidence level: comment lines in their header naming the atlas the data came from, which is the one place the provenance of a specific result is visible. See sessions, runs and artifacts.

Do not ask for the whole catalog#

The instinct is to ask what steels do you support? and read the answer. With 2,060 grades that is the wrong move, for two independent reasons.

list_materials returns the entire enabled catalog in one call. There is no limit, no offset, no family filter and no search term — its only knobs are scope (all or workspace) and the two flags include_composition and include_phases, both of which default to false precisely because the payload is already around two thousand rows.

And the agent's own built-in grade list changes shape above forty enabled grades: instead of every name it carries a summary by alloy type and family, with counts and up to five example labels each. The agent is separately instructed that for a large catalog it should summarise with counts and a few examples rather than enumerate. So a broad question is answered with a summary by design. That is a sensible answer — it is just not an answer to a question about your steel.

Ask narrowly instead:

Do you have 16MnCr5?

Which bearing steels are in the catalog?

What Cr-Mo-V hot-work tool steels do you have?

Compare 42CrMo4 with 34CrNiMo6.

One thing to discount if you see it: list_materials also returns a fixed note saying all grades are available for HT Sim, CCT, RSW and V-Gleeble. That note is wrong for RSW and for the V-Gleeble deformation modes, both of which are limited to twenty grades. The capability matrix is the accurate account.

Picking a grade from a chat configuration card#

Every simulation configuration card in chat carries a searchable material picker over the whole catalog, which is usually the fastest route from a conversation to a specific grade. Its placeholder counts whatever is actually in the dropdown, your workspace materials included, so the number moves with the workspace — Search 2,061 materials… on a workspace holding two of its own. It shows at most 80 rows with the footer Showing 80 of N matches. Type to narrow the list., and it groups results as Active material, then Workspace Materials, then Alloy type · Family.

One trap is worth its own line: the RSW plate picker offers the whole catalog too. Nothing in that dropdown restricts you to the twenty grades spot welding supports, so you can pick 16MnCr5 as a plate material and only discover the problem at run time, when the run is refused with "…has no validated RSW bulk-property mapping" followed by the twenty that are supported. A dropdown being willing is not a capability check.

If your steel really is not there#

Two routes, in this order.

Look for the family rather than the name. A designation can be absent while a metallurgically equivalent grade is present under another naming system, and the catalog's families and groups are the map for that — see families and data sources.

Then supply the chemistry. Phases will run a composition from your mill certificate, mount it on a verified template and tell you how far the two sit apart. Read using your own chemistry before you do, particularly the part about the twenty element rows a card has: an element outside them is dropped without a warning, so a cobalt or a calcium addition silently becomes zero. A named catalog grade, where one exists, is always the better answer.

Finally, being in the catalog is not the same as being runnable. All 2,060 cards are verified against the same schema with no integrity errors, but transformation kinetics, RSW bulk properties and V-Gleeble bulk flow surfaces are resolved per run, and each simulator reports what it could and could not find. That is the capability matrix, and the boundaries of what the models claim at all are in known limits.