Repeat Repair Detection Suppression Index (RRDSI)

GMI-TF-014Fleet & TransportationIntegration-Required

Definition

Percentage of assets with multiple work orders on the same DOT-regulated system within a configurable window where the CMMS repeat repair feature did not trigger, because AI-driven routing used different component codes, shops, or mechanics across the work orders.

The Measurement Gap

No metric tells you whether AI-driven work order routing is suppressing the CMMS repeat repair detection on DOT-regulated components. Fleets increasingly rely on AI to auto-assign component codes, route work to available shops, and distribute repairs across mechanics, but these optimizations can inadvertently scatter what is functionally the same recurring failure across different data pathways, preventing the repeat repair algorithm from connecting the dots. The vehicle keeps coming back for the same brake system problem, but the CMMS sees three unrelated repairs.

Three legacy metrics are blind to this suppression:

  • Repeat repair rate is the metric most directly corrupted. If AI routing assigns different component codes to what is functionally the same system-level failure, the repeat repair algorithm never fires. The rate looks low, not because repairs are durable, but because the detection mechanism is broken. The metric is blind precisely because its own input data has been fragmented.

  • First-time fix rate appears healthy because each individual repair did resolve the immediate symptom. The brake caliper was replaced, the brake line was bled, the brake pad was swapped: each "fixed" the reported problem. But the underlying brake system failure keeps recurring, and first-time fix rate cannot see patterns across work orders with different component codes.

  • Work order completion rate shows 100% completion across all three work orders. Every repair was finished on time, parts were consumed, labor was recorded. The metric sees three successful maintenance events: it has no mechanism to ask whether those three events represent a single unresolved failure.

Decision Supported:

Is AI-driven work order routing fragmenting repair history in ways that prevent the CMMS from detecting recurring failures on DOT-regulated systems?

Formula

Plain Language

RRDSI counts how often assets have multiple repairs on the same DOT-regulated system within a defined time window where the repeat repair detection did not trigger because the work orders used different component codes. You divide the suppressed repeats by total system-level repeats to get the suppression percentage.

Notation

$$RRDSI = \frac{SuppressedRepeats}{TotalSystemLevelRepeats} \times 100$$

Variables

Variable Definition Unit
SuppressedRepeats Count of asset-system pairs with 2+ work orders within the repeat repair window where different component codes prevented repeat detection from triggering Integer
TotalSystemLevelRepeats Count of all asset-system pairs with 2+ work orders within the repeat repair window when evaluated at the system level (e.g., brake system, steering system) rather than component level Integer
Repeat Repair Window Configurable time period for evaluating whether work orders constitute a repeat, typically 30, 60, or 90 days depending on fleet policy Calendar Days
Component Code The specific part or assembly code assigned to the work order, which may differ across work orders even when the system-level failure is the same Code
System Level The higher-level DOT-regulated system classification (brakes, steering, lighting, coupling) used to aggregate component-level work orders Classification

Proposed formula, pending validation pilots.

At a Glance

Unit
Percentage
Trend Direction
Lower; fewer suppressed detections
Reporting Frequency
Quarterly
Category
Data Pathway Integrity
← Motor Pool Maintenance Funding Ratio (MPMFR)Consist Minimum Inspection Margin (CMIM) →

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