Kennel the Robot Dogs and Turn Off the AI Models: Old School Ways to Fix an Inspection Budget

Key takeaways

• Validate: stop inspecting based on bad data.
• Coordinate: bridge the gap between resource planning and NDE execution.
• Optimize: plan inspections around location and access, not just due date.

Where does an inspection budget actually go?

Industry attention is on robotics, drones, AI, and advanced RBI models. Those tools help, but they amplify whatever foundation they sit on. The two cost drivers that most often go unaddressed are:

• Data mismanagement, which produces unnecessary inspections and false risk profiles.
• Resource mismanagement, which makes the field work of collecting inspection data far more expensive than it needs to be.

How bad data drives unnecessary inspections

Under API 570 and API 510, the next inspection date for a corrosion monitoring location is commonly set at half the remaining life: (t_actual − t_required) / corrosion rate × ½. Each input to that calculation can be wrong, and an error in any one makes inspections too early or too late and inflates the facility's apparent risk.

Symptoms of resource mismanagement

Planners, craft support, and inspectors often work disconnected from one another, and it shows in the field:

• Scaffolding missing when needed, built long before NDE, or torn down before the job is done; rope access and drones not considered.
• Insulation not removed before inspection or not reinstalled after; UT ports missing.
• NDE crews taking only one or two readings per permit, working from paper forms with manual data entry, without knowledge of previous readings or access requirements.

How to fix it

Fix the data

Systematically find and correct bad data, end-of-life concerns, and undocumented replacements in the existing inspection data. Validate incoming data with data loggers, a QA/QC process, and machine learning where it helps.

Improve resource coordination

Optimize inspection plans by asking whether every CML adds value and whether valuable CMLs are being executed effectively. Group work by access and location rather than due date, and attack break-ins, permit time, and other efficiency losses. Treat inspection as a construction activity on a critical path, with just-in-time insulation removal and scaffolding.

Case study: one tower, five piping circuits

A distillation tower had CMLs coming due across five circuits on different schedules: two tower feed circuits driven by remaining life, a naphtha draw, an overhead line needing rope access, and an insulated reflux line with expensive scaffolding. Planned by due date, the work required seven permits, two separate scaffold builds, and one unnecessary inspection. Planned by location and access, pulling forward CMLs that would come due soon, the same coverage took three permits, one combined scaffold (or rope access), and no unnecessary inspections.