Biofouling Assessment Calibration (BAC)

GMI-MAR-027Maritime IndustriesIntegration-Required

Definition

Accuracy of CV biofouling assessment model across the fleet operating geography.

The Measurement Gap

What This Metric Measures

BAC measures the accuracy of the computer-vision-based biofouling assessment model across the geographic regions the fleet actually operates in. It compares model-predicted fouling scores against physical inspection results at hull cleaning events, segmented by the vessel's primary recent operating region: tropical, temperate, polar, or mixed.

The Decision It Supports

Is the fouling assessment model equally accurate across all operating regions, or does it systematically under-score or over-score fouling in certain geographies: leading to miscalibrated cleaning schedules? A model that overestimates fouling in temperate waters triggers unnecessary cleanings. A model that underestimates fouling in tropical waters allows hull degradation that increases fuel consumption and emissions beyond the point where cleaning would have been cost-effective.

Why Legacy Metrics Cannot See This

Legacy hull performance monitoring tracks speed-power curves and computes a fouling penalty as the gap between clean-hull and current performance. This tells you that the hull is fouled but not whether the CV model that predicts fouling severity is accurate. An organization might have a CV fouling model that scores 0.6 severity while actual fouling at cleaning is 0.9 severity. The speed-power curve eventually reveals the fouling, but by then the cleaning window has passed and the fuel penalty has accumulated. BAC catches the model error at the cleaning event (before the next fouling cycle begins) and segments it by geography so the organization knows exactly where the model is trustworthy and where it needs recalibration.

Formula

Primary Formula

BAC(r) = 1 − |Model_fouling_score(r) − Actual_fouling_severity(r)| / Actual_fouling_severity(r)

Variable Definitions

  • Model_fouling_score(r) = CV model's fouling severity score for vessels primarily operating in region r (0-1 scale)
  • Actual_fouling_severity(r) = actual fouling severity assessed at hull cleaning for vessels in region r (0-1 scale, standardized using a defined scale such as NSTM 009-32 or a 5-level severity scale)
  • r = geographic operating region (tropical, temperate, polar, mixed)

Units:

0 to 1 per region. A value of 1.0 indicates perfect calibration in that region. Compare BAC across regions to identify geographic bias in the model.

Edge Cases

Vessels operating across multiple regions should be classified by the region where the majority of operating time occurred since the last cleaning. Actual fouling severity assessment carries inherent subjectivity: standardize using a defined severity scale consistently across all cleaning events. Minimum sample of 5 hull cleaning events per geographic region per measurement period is required for meaningful comparison.

Proposed formula, pending validation pilots.

At a Glance

Unit
Score (0 to 1)
Trend Direction
Toward 1.0; Approaching 1.0 across regions
Reporting Frequency
Per hull cleaning event
Category
Digital Twin Fidelity & Model Validation
← Hull Performance Twin Validation Index (HTVI)Structural Twin Prediction Accuracy (STPA) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment