Brick Casting Process Parameter Recommendation

Problem

Brick casting quality depends on parameters such as temperature, duration, shift, and mold material. When values drift away from historically qualified ranges, defects such as waist inclusion, cap collapse, or abnormal weight can occur. This skill helps engineers turn experience-based questions like “what temperature and duration should this size, material, and location use?” into reviewable recommendations: similarity-based retrieval over qualified history, with core parameters, confidence, input warnings, and defect risk notes.

How It Works

The input usually requires six fields: material, location, spec model, net weight, casting amount, and sand mold material. Before fields are complete, the skill should ask for missing values rather than using defaults. The output is prioritized:
- Core recommendation: casting temperature, duration, furnace/box/shift information.
- Confidence level: high or medium-high can be adopted directly; low or very low requires manual review.
- Input warnings: inconsistent net weight/spec or casting amount below net weight must be flagged.
- Risk notes: common defects observed in similar historical samples.
- Matching layer: similarity threshold search and L1-L6 fallback logic.

Limits and Notes

It is suited for casting parameter pre-checks, defect troubleshooting, and shift handover support, but should not be the sole release decision. Low confidence, sparse history, or input warnings require human confirmation; do not invent values if the API fails. Small changes in spec, weight, or mold material can change similarity matching, so accurate entry matters.