Today
Defects found. Causes investigated later.
- Defect data is used mainly for pass/fail and quality disposition.
- Root-cause work starts after a trend is already visible.
- Inspection and furnace or tin-bath data sit in separate systems.
Inspection finds the defect.Process history can explain why.
Modern float lines already generate continuous defect data from online inspection. INSUS connects that output with furnace, tin bath and line conditions so quality teams can review likely root causes and early drift signals. The inspection system still detects. People still decide what to investigate.
01
Defect patterns linked to process context.
02
Likely causes ranked for review.
03
No replacement of the inspection system.
The problem
Detection and diagnosis are different jobs.Most plants have solved the first.
Online inspection can detect and classify bubbles, seeds, distortion and tin-side defects as glass moves. The harder question is what upstream condition changed when the defect rate moved. Quality teams often cross-reference inspection results with furnace, tin bath and line data only after a trend is already visible.
01Defect detected
02Process data separated
03Root cause investigated manually
- 01Defect detected
- 02Process data separated
- 03Root cause investigated manually
The shift
From defect detectionto defect explanation.
Quality and process teams stay responsible for investigation and action. What changes is whether defect-rate changes arrive with ranked process context from this line's own history.
With INSUS
Defect changes come with process context
- Defect-rate changes are correlated with upstream conditions.
- Likely cause candidates are ranked for quality-team review.
- Historical patterns can support earlier drift warnings where the data allows.
How it works
From defect and process historyto a likely cause.
This fits OptiStudio as operational analytics and decision support above existing inspection and process systems. Day one is advisory: no automatic furnace or line setpoint changes.
Connect
Bring quality and process data together
Read-only links to online defect type, location, severity and timing with furnace DCS or historian tags, tin bath conditions, line speed and other available process records.
Understand
Learn relationships on this line's history
Analyze how defect-rate changes have related to upstream parameters on this plant, not from a generic defect-to-cause template.
Predict
Rank the most likely process causes
When a defect pattern changes, rank likely process causes by historical correlation strength for that production line.
Advise
Surface causes and early drift for review
Present ranked causes and, where history supports it, flag process conditions moving toward patterns previously linked to rising defects. Teams investigate and decide.
Verify
Improve confidence with team feedback
Compare flagged causes with actual investigation findings. Confirmed and disproved cases feed the plant-specific model.
Getting started
Start with one float lineand the data it already collects.
For float lines that already hold inspection logs and time-aligned furnace or process history. Delivered through OptiStudio as advisory quality analytics. Expert Services can help with discovery and validation against known incidents. No new cameras or sensors are assumed.
- What we need. Defect-inspection logs and time-aligned historian data: defect type, location, severity, furnace temperatures, tin bath conditions, line speed and other available process tags.
- What you see first. A correlation view that ranks likely process causes for defect-rate changes and flags early drift patterns supported by this line's history.
- How it grows. After validating flagged causes against known incidents on one line, extend across additional lines as confirmed and disproved cases accumulate.
Related for glass
Other glassstarting points
Defect root-cause analytics sits beside furnace energy and cutting yield on the same float line.
- 01Furnace Energy & Melting OptimizationRecommend furnace combustion and temperature setpoints from DCS and historian data inside quality constraints.
- 02Cutting & Yield OptimizationRecommend cutting patterns around defects and orders to reduce avoidable cullet from each ribbon section.
- 03Glass industryThe wider picture for float melting energy, cutting yield and defect analytics.
Questions glass quality teams ask us
Inspection finds defects; diagnosis often waits until trends are already visible. This use case correlates defect and process history so quality teams can review likely causes sooner.
Existing defect-inspection logs and time-aligned furnace and process historian data, including defect type, location, severity, furnace temperatures, tin bath conditions and line speed where available.
No. Inspection still detects and classifies defects. Quality and process teams investigate recommendations and decide what action to take.
No. The pilot analyzes data from existing inspection and process systems rather than adding new inspection hardware.
No. It is advisory decision support with read-only integration. Furnace optimization is a separate use case.
Flagged causes are compared with quality-team investigation findings. Confirmed and disproved cases improve plant-specific confidence over time.
Through OptiStudio as operational quality analytics. Expert Services can support discovery, data alignment and validation against known historical incidents on one line.
Glass · Quality & process analytics
Inspection knows what happened. Process history can show why.
Connect defect-inspection data with furnace and process history so rising defect rates become explainable signals your quality and process teams can investigate.