"How We Calculate Risk" - our Default Risk Recipe

Last updated: April 9, 2026

SewerAI's risk assessment combines Likelihood of Failure (LoF) and Consequence of Failure (CoF) as equal components (averaged together) to generate comprehensive risk scores, which we call Total Risk. This methodology was developed through collaboration with civil engineers, utility customers, industry experts, and review of published research, including the 2020 ASCE meta review paper by Mohammadi et al.

The LoF describes the probability that a pipe will fail, and is driven by but not synonymous with the NASSCO inspection LoF. The CoF describes how bad it would be should the pipe fail.

Likelihood of Failure (70% of total risk)

The LoF assessment prioritizes recent inspection data while incorporating asset characteristics and SewerAI's unique benchmarking:

Primary Factor:

  • NASSCO LoF (70%) - Derived from the most recent pipe inspection observations, this forms the foundation of failure likelihood assessment

Supporting Factors:

  • PIONEER Structural Benchmark (10%) - SewerAI's proprietary ranking that compares each pipe against similar pipes (same material and diameter) across the millions of assets stored in PIONEER

  • Age of Asset (10%) - Traditional degradation indicator from GIS data

  • Diameter (4%) - Pipe size as a structural factor

  • Depth (2%) - Installation depth considerations

  • Length (2%) - Pipe segment length

  • Material (2%) - Pipe material type from asset records

Adaptive Weighting: When recent inspection data (NASSCO LoF) is unavailable, the system automatically rebalances the remaining factors to still provide meaningful LoF scores. An AI model is in development to more intelligently infer LoF for uninspected pipes.

Consequence of Failure (30% of total risk)

The CoF assessment focuses on impact severity using infrastructure importance and proximity analysis. Our default formula takes the highest score from any single factor as the consequence,

  • Diameter - Used as a proxy for network importance and number of upstream customers served

  • Hospital Proximity - Distance to critical healthcare infrastructure

  • School Proximity - Distance to educational facilities

  • Waterway Proximity - Distance to sensitive environmental water bodies

  • Road / Rails Overlap - What type of road surface, path or railway the asset sits under,

  • Land Use Overlap - Whether the asset intersects sensitive spaces like parks, airports or government buildings

All proximity data is calculated using national mapping datasets (USGS / Census TIGER files). We have plans to incorporate more sophisticated network analysis for better infrastructure importance assessment.

This balanced approach ensures risk scores reflect both the technical likelihood of pipe failure and the real-world consequences of service disruption.