AI-Powered Sewer Inspection at Scale Under EPA Consent Decree

Overview

The City of Houston operates one of the largest wastewater collection systems in the United States by miles of pipe — a 6,000-mile gravity conveyance network serving 129,000 manholes. In April 2021, the City entered into a federal EPA Consent Decree (potentially the largest ever) following mid-2010s negotiations, driven by decades of infrastructure deterioration from massive mid-to-late 20th century growth, groundwater infiltration, rain-driven inflow, difficult-to-access backyard and backlot manholes and pipes, and the challenges of low-lying, flat topography. Now nearing the end of Consent Decree Year 4, the City is executing one of the most ambitious sewer assessment programs in the country — with SewerAI at the center of it.

The Challenge

Houston's sewer system faces a convergence of structural and operational pressures. Many assets installed during the city's rapid mid-to-late 20th century growth are now at or near the end of their rated useful life. The high groundwater table drives persistent infiltration, while rain events cause significant inflow. Approximately 60% of the city's 129,000 manholes are located in backlots — difficult to access, inspect, and maintain. Shallow pipe depths create frequent utility conflicts, and the relatively flat topography limits natural drainage. Against this backdrop, the EPA Consent Decree — which began in April 2021 — requires the City to systematically inspect and assess its entire collection system, with near-real-time reporting to federal authorities.

In 2020 and 2021, the City evaluated AI vendors from around the world. SewerAI was selected, and the engagement began in August 2021.

The SewerAI Engagement

The City of Houston's SewerAI program is one of the most complex and comprehensive AI-assisted sewer assessment deployments in the world. Ongoing condition assessment of CCTV videos is captured by approximately six CCTV and cleaning contractors, with three engineering firms involved in managing the AI-assisted NASSCO coding scope. The platform provides near-real-time project statuses for reporting to federal authorities and delivers critical alert notifications as defects are identified.

The scope has expanded significantly since the engagement began:

• 2021: SewerAI engagement begins; AutoCode AI-assisted NASSCO coding deployed across contractor network
• 2023: City revised contractor CCTV specifications to require use of PIONEER for AutoCode; Quality Assurance Inspection (QAI) added to SewerAI scope
• 2025: Laterals added to assessment scope
• 2026: Five-year engagement extension

Accuracy Study

In the first several months of the SewerAI-Houston engagement, an independent third-party audit was conducted across 1,500 surveys. The results validated SewerAI's performance at scale:

• 97% accuracy for Precision (True Positives)
• 8x greater accuracy for Recall (False Negatives) relative to the CCTV contractor dataset average (150 surveys)

The current engagement covers approximately 12 million linear feet of PACP assessments and ~12,000 MACP inspections annually.

Manhole Inspection at Scale

With 129,000 manholes in the sewer system — approximately 60% located in backlots — the City faces a unique inspection challenge. Data captured by several contractors is consolidated through the SewerAI platform, enabling the City to inspect approximately 10,000 manholes per year with dramatically improved economics:

• ~80% capex savings on inspection equipment
• ~40% savings per inspected manhole compared to legacy methods

Results

53% Increase in Inspected Miles Per Year

By integrating AI-assisted coding across its contractor network, the City of Houston has achieved a 53% increase in the miles of sewer inspected per year — a critical metric for meeting Consent Decree milestones and reporting obligations.

97% Accuracy — 8x Greater Than Contractor Results

SewerAI's AutoCode delivers 97% accuracy in defect coding — eight times greater than the average accuracy of contractor-submitted data. This level of precision is essential for a program where inspection results directly inform $6 billion in planned infrastructure investments over 15 years.

Financial Impact

The SewerAI engagement has delivered measurable financial returns across multiple dimensions:

• ~$1M per year in staff augmentation savings
• ~$100K in year-one QAI-based savings through accurate contractor pay-items
• 55.1% reduction in failure rate for contractor submittals

Supporting $6B in Infrastructure Investment

The data generated through the SewerAI platform directly supports the City's planned $6 billion in infrastructure investments over the next 15 years, providing the accurate, consistent condition data needed to prioritize repairs and demonstrate compliance with federal authorities.

In Their Own Words

"The data is readily available both in the app and via an API, making it easier for operations to make critical decisions required to maintain the collection system. SewerAI reduced the internal labor hours needed to code and review, and the data is more precise."

— Greg Eyerly, Senior Assistant Director, City of Houston