CinderCore: refractory material properties

What a record contains

One row is one commercially described material grade - a brick, a castable, a board, a felt - as reported by a published source. The row does not describe a product you can buy; it describes a class of material with the numbers that were measured on it.

identity

Stable record id, grade name, family, and alias strings seen in other literature.

chemistry

Mass fractions for the oxides that decide the grade, plus free carbon where it applies.

thermal

Maximum service temperature with the atmosphere it was quoted for, and thermal conductivity at 1 000 °C with the method used to get it.

mechanical

Bulk density, open porosity, cold crushing strength, hot modulus of rupture, refractoriness under load.

durability

Thermal shock cycles survived under a stated protocol, and an ordinal slag resistance index defined by this project.

provenance

Source ids, the date the row was last touched, and a confidence flag of high, medium or low.

The schema page lists all 26 fields with units and types, and explains what an empty cell means - which is not the same as a zero.

The shape of the data

Service temperature is the field people reach for first, so here is the whole dataset reduced to it. Bars show the range of maximum service temperature across the records in each family; the tick is the median.

Where the numbers come from

Nothing here is measured by us. Every value is transcribed from a document that is either public or purchasable, and each document gets an id that the record points back to. The 38 sources break down as eleven textbooks and handbooks, nine standards and test-method documents, fourteen conference and journal papers, and four industry association data compilations.

Manufacturer brochures are deliberately excluded unless they state the test method and the test temperature. That rule removed roughly a third of what we first collected, and it is the single biggest reason the dataset is smaller than it could be.

How units were converted, how disagreements between two sources were resolved, and which parts of the data we do not trust are all written down on the methodology page, including a frank list of gaps.

What this is not

• It is not a design authority. Lining thickness, anchoring and expansion allowance need an engineer, not a table.
• It is not a substitute for a supplier datasheet on a specific delivered batch. Grades drift; two bricks sold under the same name can differ by a hundred degrees of practical limit.
• It is not complete. Entire properties that matter in service - abrasion, creep, alkali attack - are absent because the published data was too thin to normalise.
• It is not a benchmark. The slag index is an ordinal judgement defined by this project, and comparing it across chemistries is meaningless.

Confidence, in one picture

Of the 412 rows, 214 are flagged high, 137 medium and 61 low. The flag is a property of the record, not of the family: a shaky microporous panel sits in the same insulating bucket as a well-documented firebrick, so a filter on family alone mixes the two.