Efficiency Detection (PRRC)

The Performance Ratio Relation Correction (PRRC) is an intelligent feedback control mechanism that enables fair comparison of energy production components across your installation, even when they have different technical specifications, configurations, or operating conditions.

Purpose

PRRC solves a fundamental challenge in multi-component monitoring: How do you fairly compare components that are inherently different?

Without PRRC, it would be impossible to answer: "Is this component performing as well as it should relative to others in the system?"

How PRRC Works

Feedback Control Mechanism

PRRC is a continuously adapting correction factor that bridges the gap between theoretical simulation and real-world performance:

PRRC = Performance Correction Factor
Corrected Simulation = Raw Simulation × PRRC

The factor is kept within bounds that depend on the component type — the closer a component sits to the grid, the tighter its allowed range:

Key Properties:

• Starts at 1.0 (no correction needed)
• Adapts daily based on actual vs. expected performance
• Only updates during normal operation (prevents corruption during failures)
• Configuration-specific (resets when component specs change)

The PRRC Lifecycle

1. Initialization

When monitoring begins for a component:

• Query for stored PRRC history matching the exact configuration
• If recent values exist: continue from their rolling 7-day average
• If not found or configuration changed: start with the neutral PRRC = 1.0

2. Daily Adaptation

Every day a component produces normally, the system learns from it:

• Compares measured production against physics-based simulation
• Calculates the day's performance deviation
• Nudges PRRC toward closing that gap by a fixed, component-specific fraction of the deviation (weighted by data quality): 50% for strings, 40% for combiner boxes, 30% for inverters, 20% for feed-in meters
• Strings adapt fastest to capture each string's individual behavior; the higher aggregation levels move more slowly, keeping the plant model steady

3. Stability Through Averaging

The correction applied to simulations is not the latest daily value:

• A rolling 7-day average of the stored daily values smooths out single odd days
• The further the model is from reality, the larger each daily correction — so the first days after a fresh start sharpen the model quickly, with convergence emerging from the feedback loop itself
• Gradual changes (seasonal patterns, aging) are tracked without overreacting to noise

When PRRC Updates

PRRC only updates on days when the component was classified as producing normally — its production within the normal band for its component type — and the day was not flagged as anomalous (see below):

This prevents PRRC from "learning" incorrect patterns during failures or measurement errors.

Anomalous Days Are Skipped Entirely

Beyond the per-component-type normal band above, PRRC also refuses to learn on bad-weather or unreliable days:

• If more than 30% of a day's daylight is flagged as anomalous (snow, dew, fog, network outage, shutdown, or other excluded periods), PRRC does not adapt that day for any component in the plant.
• The correction is applied weather-conditionally — clear, hazy, and cloudy conditions are weighted differently so the model is not skewed by a single cloudy afternoon.

Why PRRC is Needed

The Challenge: Diverse Components

Energy installations contain components with inherent differences:

• Different specifications: Varying capacities, efficiencies, designs
• Different configurations: Various orientations, connections, topologies
• Different conditions: Environmental factors, maintenance levels, age

The Solution: Normalized Comparison

PRRC normalizes all components to a common reference scale:

• A high-capacity component vs. a low-capacity component can be compared
• Components with different orientations are evaluated fairly
• Efficiency can be assessed independent of design differences

Understanding PRRC Values

High Efficiency (PRRC ≥ 0.95)

The component performs close to or better than expected:

• PRRC = 1.0: Perfect match between simulation and reality
• PRRC > 1.0: Performing better than theoretical model predicts
• Indicates healthy operation and proper maintenance

Moderate Efficiency (0.80 ≤ PRRC < 0.95)

The component shows some performance degradation:

• May indicate gradual wear, minor issues, or suboptimal conditions
• Still within operational range
• Monitor for further decline

Low Efficiency (PRRC < 0.80)

Significant underperformance detected:

• Indicates serious degradation or persistent issues
• Requires investigation and possible maintenance
• Could indicate measurement or configuration problems

High PRRC (PRRC > 1.05)

Component exceeds expected performance:

• May indicate overly conservative simulation models
• Could reveal measurement calibration issues
• Review configuration accuracy

Configuration Dependency

PRRC values are configuration-specific and stored with detailed metadata:

What's Tracked:

• Component technical specifications
• Operational parameters
• Physical configuration details
• Connection topology

What Happens on Changes: When component configuration changes:

1. Previous PRRC values become invalid (don't match current config)
2. System starts fresh with PRRC = 1.0
3. The feedback loop re-converges over the following days
4. Prevents applying outdated corrections to changed components

PRRC vs. Component States

PRRC works alongside Component States to provide layered monitoring:

Example:

• Component A: PRRC = 0.85, Status = PRODUCING_NORMAL
• Component B: PRRC = 0.98, Status = PRODUCING_NORMAL
• Interpretation: Both are operating normally, but Component B is more efficient

Practical Applications

1. Early Degradation Detection

Track PRRC trends over time:

Component X:
Week 1: PRRC = 0.98
Week 4: PRRC = 0.92
Week 8: PRRC = 0.85

→ Gradual decline indicates developing issue before critical failure

2. Performance Benchmarking

Compare similar components:

Component Group A (same specs):
- Unit 1: PRRC = 0.97
- Unit 2: PRRC = 0.96
- Unit 3: PRRC = 0.82  ← Outlier requires investigation

→ Fair comparison identifies underperformer

3. Maintenance Effectiveness

Measure improvement after service:

Before maintenance: PRRC = 0.78
After maintenance: PRRC = 0.95

→ Quantifiable improvement validates maintenance impact

Best Practices

Initial Setup

1. Ensure accurate configuration: Verify all component specifications are correct
2. Allow training period: Wait at least 7 days before making performance judgments
3. Verify reference data: Ensure baseline measurements are reliable

Ongoing Monitoring

1. Monitor trends, not single values: Look for patterns over days/weeks
2. Compare within groups: Compare components with similar specifications
3. Investigate sustained low PRRC: Any component below 0.90 for multiple weeks
4. Use with other metrics: Combine PRRC analysis with component status and loss detection

Troubleshooting Low PRRC

When a component shows persistently low PRRC:

Step 1 - Check Configuration

• Verify component specifications match physical installation
• Confirm all parameters are accurate in the system
• Review recent configuration changes

Step 2 - Compare with Peers

• Do similar components show the same pattern?
• If yes → likely systemic issue (environmental, design)
• If no → likely component-specific problem

Step 3 - Review Component History

• Check maintenance records
• Look for recent events (weather, incidents)
• Examine physical condition

Step 4 - Validate Measurements

• Ensure sensors are functioning correctly
• Check for calibration drift
• Verify data quality

Relationship with Other Monitoring Features

PRRC integrates with the Digital Twin's monitoring system:

• Component States: PRRC provides the efficiency metric used in status determination
• Loss Detection: While loss detection calculates absolute energy losses, PRRC shows relative efficiency
• Digital Twin: PRRC is one of several signals the nightly watchdog uses for continuous monitoring

In essence, PRRC answers "Is this component doing its fair share?" by accounting for all legitimate reasons components should perform differently, revealing only actual performance issues that require attention.

PRRC vs. Edge Performance Ratio

Two distinct measures share similar names — keep them apart:

Related Features

• Digital Twin — the nightly watchdog and analysis engine PRRC feeds
• Component States — how components are classified as normal, degraded, or faulted
• Loss Detection — absolute energy-loss accounting with confidence buckets
• Digital Twin Architecture — the technical implementation behind these analytics