LIVE · 4s
Root Cause Analysis

Root Cause Analysis Workspace

Investigate incidents with AI-assisted hypotheses, correlations, and structured RCA artefacts. All conclusions require human review.

Active incidents
Incident · rca-002
in progress· A. Kowalska (Process)
Seal defect spike on Filler F-101 (Berlin)
0.8 pp FPY drop · 1,420 units scrapped (last 24h)
RCA timeline
  1. 00:46 · 11-05quality
    FPY drop detected on Filling Line 01 (97.0 → 96.2)
  2. 01:16 · 11-05alarm
    Seal temperature instability alarm
  3. 01:46 · 11-05ai
    AI anomaly an-002 raised (conf 0.78)
  4. 02:46 · 11-05operator
    Operator note: 'cap fit feels tight, possibly batch-related'
  5. 03:46 · 11-05rca
    RCA opened by Process Engineering
Hypotheses (AI-assisted · human-verified)
Correlation analysis
Evidence panel
Humidity > 62%RH destabilizing seal temperature
Supporting evidence
  • · Pearson r=0.71 humidity↔defect (14d)
  • · 2 quality holds in 72h above 62%RH
Contradicting evidence
  • · No defect spike at last humidity peak Apr-26
Recommended next step
Re-run on shift A only; controlled humidity test
Fishbone (Ishikawa) — contributing factor categories
Machine
  • Seal head wear (184d in service)
  • Heater PID tuning
Method
  • HVAC setpoint not re-validated post-season change
  • No humidity compensation in recipe
Material
  • Cap supplier batch B-7714 dimensional variance
Measurement
  • Humidity sensor calibration (2026-04-10)
Environment
  • Hall humidity 64%RH (limit 60%RH)
  • Outdoor temperature swing
People
  • Shift handover note inconsistency
5 Whys assistant
AI-assisted 5 Whys · drafts require human confirmation
  1. 1
    Why did FPY drop? — Seal defects increased on Filling Line 01.
  2. 2
    Why did seals fail? — Seal temperature became unstable.
  3. 3
    Why did temperature destabilize? — Ambient humidity above the validated operating window.
  4. 4
    Why is humidity affecting it? — HVAC setpoint drifted; control loop slow to recover.
  5. 5
    Why did HVAC drift? — Recent setpoint change after spring transition not re-validated.
AI investigation brief
conf 74%

Top hypothesis: "Humidity > 62%RH destabilizing seal temperature". Evidence is suggestive but not conclusive — at least one verification step is recommended before sign-off.

  • Top correlation
    —
  • Evidence quality
    Multi-source (sensor + alarm + operator + AI model)
  • Governance
    RCA conclusions and corrective actions require human approval. No automated OT writeback.
Governance
AI proposes hypotheses, correlations, and report drafts. A qualified engineer must verify and sign off before any corrective or preventive action is taken.