Skip to main content

Phase 2: Observations, Alerts & Triage

Status: Future phase — planned to build on Phase 1

Overview​

Phase 2 adds intelligence to the asset condition data captured in Phase 1. When readings exceed defined thresholds or operators note unusual conditions, Phase 2 ensures those observations get visibility, structured triage, and follow-up — without requiring operators to change how they work or losing anything in the shift handover.

Core Capabilities​

Thresholds and Simple Rules: Define acceptable ranges for asset attributes (e.g., "discharge pressure should be 40–50 bar"). When readings fall outside these ranges during operator rounds, they are automatically flagged for attention.

Observations with Severity: Promote free-form operator notes and flagged readings into structured observations with severity levels (low, medium, high, critical). This ensures important issues don't get lost in unstructured notes.

Alert Views for Supervisors: Dashboards show all abnormal conditions across the plant, grouped by severity, asset, or location. Supervisors and engineers can quickly identify what needs attention.

Lightweight Triage: For each observation or alert, supervisors can acknowledge it, add comments, mark it for follow-up, or query AI agents (Phase 3) for intelligent recommendations. The goal is simple workflows that fit into existing shift handover and daily review processes.

Example: Digester Alert at Blueprint Paper Mill​

To illustrate how Phase 2 enhances Phase 1, consider this scenario at Blueprint Paper Mill:

Phase 1 (Current Capability)​

During the morning round, operator Sarah Johnson records:

  • Digester D-101 Pressure: 6.1 bar (normal range: 5.2–5.8 bar)

The app:

  • Highlights the reading in red 🔴
  • Prompts operator to add notes
  • Creates Observation #142
  • Supervisor sees observation in dashboard

Supervisor Maria Lopez manually reviews the observation, checks trends, and takes action.

Phase 2 (Future Enhancement)​

The same reading now triggers automated alert workflow:

1. Automatic Severity Classification

  • Reading is 16% above upper limit (6.1 vs 5.8 bar)
  • Rule engine classifies as High Severity (pressure excursions are safety-critical)
  • Alert automatically escalated to supervisor and shift manager

2. Threshold-Based Alerting

🚨 ALERT: Digester D-101 High Pressure
   Current: 6.1 bar | Threshold: 5.8 bar | Exceedance: +5.2%
   Severity: HIGH (Auto-classified)
   Time: 6:10 AM | Operator: Sarah Johnson

3. Contextual Alert Dashboard

Supervisor's dashboard shows:

⚠️ Active Alerts - Blueprint Paper Mill

🔴 HIGH PRIORITY (2 alerts)
┌────────────────────────────────────────┐
│ Digester D-101: High Pressure         │
│ 6.1 bar (exceeds 5.8 bar limit)       │
│ Unit A – Pulping | 6:10 AM            │
│ [Acknowledge] [Escalate] [View Trend] │
└────────────────────────────────────────┘

┌────────────────────────────────────────┐
│ Feed Pump P-404: High Vibration       │
│ 7.2 mm/s (exceeds 6.0 mm/s limit)     │
│ Unit A – Pulping | 6:15 AM            │
│ [Acknowledge] [Escalate] [View Trend] │
└────────────────────────────────────────┘

🟡 MEDIUM PRIORITY (5 alerts)
🟢 LOW PRIORITY (12 alerts)

4. Triage Workflow

Maria clicks "Acknowledge" on Digester D-101 alert:

  • Alert status changes to "Acknowledged by Maria Lopez at 6:12 AM"
  • She adds comment: "Pressure spike observed. Reducing white liquor flow by 10%. Monitoring every 30 min."
  • Alert remains open until resolved

5. Escalation Rules

If Maria hadn't acknowledged within 15 minutes, Phase 2 would automatically:

  • Escalate to shift manager
  • Send notification to reliability engineer
  • Mark alert as "Unacknowledged - Escalated"

6. Resolution Tracking

When pressure returns to normal:

  • Final reading: 5.6 bar ✅
  • Maria marks alert as "Resolved"
  • System records resolution time: 1 hour 30 minutes from initial detection
  • Metrics updated: "Digester D-101 - Average resolution time: 1.5 hours"

Key Phase 2 Enhancements​

Phase 1 (Manual)Phase 2 (Automated)
Operator notes abnormal reading+ Automatic severity classification
Supervisor reviews dashboard+ Real-time alert notifications
Supervisor manually triages+ Structured triage workflow (acknowledge, escalate, resolve)
No escalation mechanism+ Automatic escalation if not acknowledged
Manual trend analysis+ Alert dashboard with grouped priorities
No resolution tracking+ Metrics on response time and resolution

Phase 2 doesn't change how operators work—they still capture readings the same way. It adds intelligence for supervisors and engineers to ensure nothing falls through the cracks.

Why Phase 2 Builds on Phase 1​

Phase 1 establishes the data foundation: structured readings tied to asset attributes, with last known values and historical trends. Phase 2 layers intelligence on top of that same data without changing the underlying model. Operators continue to capture readings the same way; the system simply adds alerting and triage workflows for supervisors and engineers.

What Comes Next​

Phase 3 will add AI-powered insights by enabling natural language queries over observations, alerts, and asset data via MCP servers. See Phase 3 Features for more information.

See Also: