Grand Rapids, MI: Real-Time QA for Sewer Construction Using AI

As infrastructure programs scale, utilities are under increasing pressure to deliver high-quality assets while minimizing cost, risk, and disruption to the community. For growing cities like Grand Rapids, ensuring that newly installed sewer infrastructure meets expected standards before handover is critical.

By embedding Vapar AI into post-installation workflows, Grand Rapids is transforming how quickly and confidently it can verify construction quality, moving from delayed insights to near real-time decision support.

About the Client

Grand Rapids is the second largest city in Michigan with a population of 200,000 residents and serves a surrounding metropolitan area of over 1 million people. Its 1,100-mile sanitary sewer network transports and treats an average of 40 million gallons per day.

Through a successful 3-year sewer improvement project, the city eliminated all combined sewer overflows (CSOs), delivering improved environmental outcomes for the community.

The Environmental Services Department continues to rehabilitate end-of-life pipes and construct new sewers to support ongoing growth.

Project Snapshot

Grand Rapids conducts CCTV inspections after new pipe installations, but manual review processes delayed defect identification until days or weeks after installation and pipe has been backfilled.

This lag can create avoidable rework costs, contractor remobilization, and community disruption when installation issues are discovered too late. The utility needed a faster, more consistent way to review post-installation footage, so defects could be corrected while crews were still onsite before assets are signed off and handed on to operations.

The Challenge / Opportunity

Grand Rapids conducts CCTV inspections after new pipes are installed. However, traditional manual review processes created delays between capture and defect detection by city engineers.

This meant installation issues such as joint offsets, debris, grade inconsistencies, or construction damage were sometimes identified days or weeks after crews had demobilized from site.

When defects were discovered late, remediation required re-mobilizing contractors, reopening finished surfaces, and disrupting traffic or surrounding communities. The result was avoidable cost, reputational risk, and inefficiencies within the capital works program.

The utility needed a way to review installation footage rapidly and consistently so defects could be identified while construction crews were still onsite, enabling immediate correction before projects were signed off and handed over into operations.
‍

The Solution

Grand Rapids implemented VAPAR AI to automatically review post-installation CCTV footage shortly after inspection videos were captured. Using computer vision trained on millions of real pipe defect images, VAPAR AI rapidly identified construction-related issues that may have occurred during installation.

Instead of days or weeks between inspection and insight, the team now receives structured, inspection-ready outputs within hours. Defects are clearly flagged, time-stamped, and spatially referenced, allowing supervisors and project managers to quickly determine whether remediation is required.

This accelerated feedback loop enables installation issues to be identified while construction crews are still mobilized. As a result, defects can be corrected immediately before pavement is reinstated, projects are signed off, or assets are handed over to operations.

By embedding AI into the capital delivery workflow, Grand Rapids has shifted from delayed defect discovery to real-time quality assurance, reducing rework risk, lowering remediation costs, and improving confidence in newly installed infrastructure.

Results

• 200+ inspections processed efficiently
  Large volumes of post-installation footage were uploaded and analysed consistently, without adding review burden on engineering teams  

• 30,000 linear feet assessed with rapid turnaround
  Inspection data was converted into structured outputs within hours, significantly reducing time between capture and decision-making  

• 100 defects identified early in the construction phase
  Issues such as joint offsets, debris, and installation inconsistencies were flagged while crews were still onsite, avoiding delayed discovery  

• Reduced rework and contractor remobilization
  Early detection enabled immediate fixes, eliminating the need to reopen completed works and minimizing disruption to the community  

• Improved quality assurance across capital projects
  Standardized, AI-driven review ensured consistent defect identification and greater confidence in asset handover  

Conclusion

By integrating Vapar AI into its post-installation inspection workflow, Grand Rapids has fundamentally improved how it manages construction quality.

What was once a delayed, manual process is now a fast, consistent, and scalable system that enables real-time decision support. The city can identify and resolve issues earlier, reduce unnecessary costs, and deliver higher-quality infrastructure to its community.

This approach not only strengthens current capital delivery programs but also sets a new standard for how utilities can leverage AI to ensure long-term asset performance and reliability.

‍