Agentic Workflow Overview

Understand the OODA loop: Observe → Orient → Decide → Act

The OODA Loop in Operations

Ona Terminal implements the OODA (Observe-Orient-Decide-Act) loop for autonomous operations management. This military-tested decision framework enables proactive, intelligent responses to operational challenges.

Why OODA for O&M?

Traditional O&M is reactive: equipment fails → scramble to fix → high costs.

OODA-based O&M is proactive: continuous observation → pattern recognition → optimal decisions → preventive action.

1. Observe: Data Ingestion & Normalization

Data Sources

🌡️ Weather Data

• Real-time weather station feeds
• Satellite irradiance data
• Weather forecast integration
• Historical weather normalization

⚡ Equipment Data

• SCADA system integration
• Inverter performance metrics
• String-level monitoring data
• Equipment error logs and alarms

📊 Operational Data

• Production vs. expected output
• Grid interconnection data
• Maintenance history and costs
• Warranty and compliance records

Data Processing

# Weather normalization example
🤖 | /weather-normalize --site SITE001 --period 2024-01-01:2024-01-31

Key Processing Steps:

1. Data Interpolation - Fill gaps in monitoring data
2. Weather Normalization - Adjust for irradiance and temperature
3. Quality Validation - Identify and flag anomalous readings
4. Historical Correlation - Compare against baseline performance

2. Orient: Fault Detection & Diagnostics

Pattern Recognition

🔍 Fault Detection Algorithms:

• String performance degradation patterns
• Inverter efficiency decline signatures
• DC combiner failure indicators
• Tracker alignment drift detection

🧠 AI-Powered Diagnostics:

• Equipment-specific failure modes
• Manufacturer error code interpretation
• Historical fault pattern matching
• Root cause analysis automation

Diagnostic Process

# Automated fault detection
🤖 | /fault-detection --equipment inverter --threshold 0.85 --site SITE001

# Specific equipment diagnosis  
🤖 | /diagnose-inverter --inverter-id SMA001 --symptoms "output 15% below expected"

Orient Phase Outputs:

1. Fault Classification - Type, severity, and probable cause
2. Equipment Impact - Affected capacity and performance loss
3. Failure Timeline - Predicted progression if unaddressed
4. Warranty Status - Coverage validation and claim procedures

3. Decide: Economic Analysis & Prioritization

Financial Optimization

💰 Energy-at-Risk (EAR) Calculation:

• Revenue impact of continued degradation
• Energy market price forecasting
• Weather-adjusted production loss
• Time-sensitive repair value

📊 Cost-Benefit Analysis:

• Repair costs vs. energy recovery value
• Preventive vs. reactive maintenance costs
• Warranty claim value optimization
• Resource allocation efficiency

Decision Matrix

# Economic dispatch optimization
🤖 | optimize maintenance timing considering weather forecast and energy prices

# Priority-based scheduling
🤖 | /schedule-maintenance --site SITE001 --optimize-for revenue --horizon 30days

Decision Criteria:

1. Financial Impact - Revenue at risk vs. repair costs
2. Urgency Level - Time sensitivity and degradation rate
3. Resource Availability - Crew schedules and parts inventory
4. Weather Windows - Optimal conditions for maintenance
5. Grid Constraints - System maintenance windows and curtailment

4. Act: Work Order Creation & Dispatch Tracking

Automated Work Order Generation

📝 Intelligent Work Orders:

• Equipment-specific procedures from manufacturer manuals
• Required tools, parts, and safety equipment lists
• Historical repair time estimates and cost projections
• Photo/video requirements for warranty documentation

🔧 Technical Instructions:

• Step-by-step diagnostic procedures
• Manufacturer-specific troubleshooting guides
• Safety protocols and compliance requirements
• Quality control checkpoints and testing procedures

CMMS Integration

# Create and dispatch work order
🤖 | /create-work-order --equipment INV001 --priority high --type "DC combiner replacement"

# Track dispatch progress
🤖 | /track-dispatch --work-order WO123 --technician-id TECH001

Integration Features:

1. Work Order Creation - Automated generation in existing CMMS
2. Dispatch Optimization - Route planning and resource allocation
3. Evidence Capture - Photo/video requirements for warranty claims
4. Completion Validation - Quality control and performance verification
5. Knowledge Capture - Lessons learned integration for model improvement

Workflow Visualization

graph TD
    A[Observe: Data Ingestion] --> B[Weather Normalization]
    A --> C[Equipment Monitoring]
    A --> D[Historical Analysis]
    
    B --> E[Orient: Fault Detection]
    C --> E
    D --> E
    
    E --> F[Pattern Recognition]
    F --> G[Diagnostic Analysis]
    G --> H[Warranty Validation]
    
    H --> I[Decide: Economic Analysis]
    I --> J[Cost-Benefit Calculation]
    J --> K[Priority Matrix]
    K --> L[Resource Optimization]
    
    L --> M[Act: Work Order Generation]
    M --> N[Dispatch Optimization]
    N --> O[Evidence Capture]
    O --> P[CMMS Integration]
    
    P --> Q[Performance Validation]
    Q --> A
    
    style A fill:#e1f5fe
    style E fill:#f3e5f5
    style I fill:#fff3e0
    style M fill:#e8f5e8

Real-World Example: String Performance Issue

Observe Phase

• Monitor detects String 12 underperforming by 18%
• Weather data shows clear skies (no irradiance issue)
• Historical data shows gradual decline over 3 weeks

Orient Phase

• AI diagnostics suggest DC combiner failure
• Pattern matches manufacturer TSB for this combiner model
• Warranty check confirms coverage expires in 45 days

Decide Phase

• EAR calculation: $2,400/month revenue loss if unrepaired
• Repair cost estimate: $1,200 parts + $800 labor
• Optimal timing: Schedule within 30 days to preserve warranty

Act Phase

• Generate work order with specific combiner part number
• Schedule technician with DC combiner replacement experience
• Include photo requirements for warranty claim documentation
• Validate repair with post-maintenance performance monitoring

Result: $28,800 annual energy recovery, warranty claim approved, 2-hour repair time vs. 6-hour reactive response.

Performance Metrics

Cycle Time Optimization

Total Cycle Time: 5-10 days → 2-4 hours (95% improvement)

What’s Next?

1. Configure Custom Models - Deploy your fine-tuned OODA models
2. Explore O&M Use Case - See complete business implementation
3. Master CLI Commands - Execute OODA workflows interactively

Configure Custom Models

Get Help & Stay Updated