Skip to content

Project Crucible

Jefferson edited this page May 15, 2026 · 2 revisions

Project Crucible

← Home


Overview

Project Crucible is a fire risk analysis pipeline that transforms NERIS (National Emergency Response Information System) incident data into actionable risk reports visualized in TAK (Team Awareness Kit). The project demonstrates how agent-based models, ISO 31010 risk methodology, and geospatial situational awareness converge to make previously uninsurable fire-risk properties insurable.

Core thesis: If the math doesn't fit the problem, fit the problem to the math.


The Pipeline

IoT / Fire Departments → NERIS API → ISO 31010 Analysis → TAK/iTAK → Decision Makers
Stage Implementation
Data Source NERIS API (https://api.neris.fsri.org/v1) — fire incidents with geolocation
Filter Arizona (state=AZ), fire incident types
Analysis Bow Tie, FMEA, Decision Tree, Markov classification
Output CoT Data Package (.zip) for iTAK on iPad
Visualization Fire markers with expandable risk reports in remarks

Agent-Based Model Context

Project Crucible is grounded in the principle that categorizing phenomena into modelable patterns enables inference about conflict resolution and resource allocation.

Sugarscape — Urban Distribution

Agents competing for scarce resources naturally distribute into concentric patterns mirroring real urban geography. Risk density follows the same pattern — properties in high-competition zones face different risk profiles than those at the periphery.

Recycling — Incentivizing Remediation

Models how incentive structures drive participation in environmental cleanup (riparian wash remediation). The parallel: if a captive insurance model shares underwriting profit with those who reduce fuel loads, prevention becomes self-sustaining.

Segregation — Coverage Withdrawal

Schelling's model shows how mild preferences produce dramatic boundary effects. Applied to fire insurance: carrier withdrawal from one zone cascades until entire regions become "uninsurable." The captive model breaks this cycle.


ISO 31010 Risk Analysis

Each fire incident receives four analyses:

Technique ISO Section Application
Bow Tie Analysis B.4.2 Preventative & reactive control effectiveness
FMEA B.4.3 Failure modes ranked by Risk Priority Number
Decision Tree B.9.3 Investment recommendation (simulated financials)
Markov Analysis B.5.9 LOW / MODERATE / HIGH risk state classification

Repository Structure

project-crucible/
├── src/
│   ├── config.py              # Configuration, env vars, simulated data fallback
│   ├── neris_client.py        # NERIS API client (OAuth2 + paginated fetch)
│   ├── simulated_data.py      # 40 realistic AZ fire incidents for demo mode
│   ├── risk_analysis.py       # ISO 31010 analyses + report generation
│   ├── tak_export.py          # CoT Data Package + KML export for iTAK
│   ├── main.py                # Orchestrator script
│   └── generate_pdf_report.py # PDF report with ABM context + TAK screenshots
├── context/
│   ├── 2026-03-18_AZCIA-Notes.md
│   ├── 2026-05-02_Crucible-Coalition_Forms_Email-Thread.pdf
│   ├── 2026-05-14_Project-Crucible-Design.md
│   ├── literature_map.md
│   ├── neris_onepager.html
│   ├── NERIS_Risk-Report-Guide.md
│   └── Project-Crucible_Coalition-Report.md
├── images/                    # TAK screenshots + ABM visualizations
├── output/
│   ├── az_fire_incidents.zip  # CoT Data Package for iTAK
│   ├── az_fire_incidents.kml  # KML fallback
│   └── Project-Crucible_TAK-Visualization-Report.pdf
├── requirements.txt
└── README.md

Quick Start

cd riskrunners/project-crucible

# Install dependencies
pip install -r requirements.txt

# Run pipeline (simulated data — no API credentials needed)
python3 src/main.py

# Generate PDF report
python3 src/generate_pdf_report.py

Using Real NERIS Data

export NERIS_CLIENT_ID="your_client_id"
export NERIS_CLIENT_SECRET="your_client_secret"
python3 src/main.py

Loading in iTAK (iPad)

  1. Transfer output/az_fire_incidents.zip to iPad (AirDrop)
  2. iTAK → ≡ menu → Data Packages → Import
  3. Select the .zip file — fire markers appear on map
  4. Tap a marker → tap callsign → scroll to Remarks for full risk report

The Apex Vision

The iTAK proof of concept establishes the data pipeline. The next evolution is Apex — the VisionPro platform for immersive 3D situational awareness:

  • Drones flown on regular schedules over fire-prone corridors
  • Video streams combined with NERIS history and IoT sensor data
  • Apex vantage point — "eyes in the sky" for fire marshals, underwriters, developers, and brush removal services
  • Collaborative remediation — civic services and insurance industry jointly reducing fire risk

The complete pipeline achieves what the standard market cannot: making previously uninsurable fire-risk properties insurable through continuous, evidence-based risk scoring that rewards prevention.


Connection to Captive Insurance

Project Crucible feeds directly into the Integral Mass Captive feasibility study. The same behavioral incentive from Medical Stop-Loss applies:

  • MSL: Employers invest in wellness → keep underwriting profit
  • Fire Risk: Communities invest in prevention → get lower premiums
  • GC Captive: Contractors optimize quality → keep the dividend

Stakeholders

  • Arizona Captive Insurance Association (AZCIA)
  • Crucible Coalition
  • Casualty Actuaries of the Desert States
  • Fire departments reporting through NERIS

Links


Project Crucible — Insuring the uninsurable through data-driven risk intelligence.

Clone this wiki locally