How the Analysis Works

Recipe Parsing and Material Resolution

Materials are matched to the internal database by name, alias, and supplier normalization. Each matched material carries oxide analysis, LOI correction, and source metadata. Unresolvable materials produce explicit missing-data warnings rather than silent fallback.

Oxide Calculation and LOI Correction

Batch percentages are converted to fired-weight oxide contributions after loss-on-ignition correction. Oxides are summed across all materials and reported both as mass percent and as molar equivalents. Calculation precision is limited by the accuracy of the source material analysis.

Unity Molecular Formula (UMF)

The molar oxide composition is normalized so RO (fluxes) sum to 1.00. This puts all recipes on a comparable scale regardless of batch size. RO2 (network formers: SiO₂, TiO₂) and R₂O₃ (intermediates: Al₂O₃) are expressed as ratios relative to unity flux.

• Si:Al ratio signals durability and surface character.
• Alkali to alkaline-earth flux balance affects thermal expansion.
• High-MgO vs high-CaO fluxes behave differently at temperature.

Limit Formula Comparison

UMF values are compared against published limit ranges for the stated cone and atmosphere. Out-of-range values trigger flags. The limit approach is a useful heuristic—it is not a physical model. It does not account for interaction effects, firing schedule, clay body, or application thickness.

• Silica below typical range: durability concern.
• Alumina below range: melt stability concern.
• High alkali flux: elevated thermal expansion proxy.

Thermal Expansion Proxy

Ceramic Stability computes a relative expansion index from oxide contributions using published coefficients. This is a proxy, not a measured COE. Coefficients vary by source and are temperature-dependent. The value is used comparatively—between recipe variants or against known body expansion ranges—not as an absolute prediction.

• Higher alkali (K₂O, Na₂O, Li₂O) generally increases expansion.
• Higher silica and alumina generally moderate expansion.
• Calcia and magnesia effects are temperature-range dependent.

Risk Rule Engine

Rules fire against the computed oxide profile, UMF, limit comparison, and available metadata. Each rule emits a structured risk object with category, severity, confidence, evidence list, and recommended actions. Rules are versioned; each report records the rule-set version used.

Risk categories

• Thermal fit: crazing, shivering, dunting
• Melt process: running, crawling, pinholing, blistering, over/under-melt
• Surface durability: leaching concern, cutlery marking, acid attack
• Color: burnout, reduction shift, devitrification
• Application: settling, hard-panning, flaking

Confidence Scoring

Confidence reflects how much relevant information was provided. It is not a quality score for the recipe—it is a score for the analysis.

• Very low: recipe chemistry only.
• Low: recipe plus firing cone.
• Medium: recipe plus body plus firing schedule plus observations.
• High: measured test evidence (leach, expansion) included.
• Very high: repeated batch data plus lab and long-term results.

Revision Candidates and Test Plans

For each elevated or high risk, the engine generates at least one revision candidate: a directional material change with expected oxide effect, likely surface tradeoff, and test priority. The test plan output lists specific physical tests matched to flagged risks, ordered by priority.