Ultrasonic Thickness, Velocity, and the Calibration Trap

Thickness is velocity times transit time over two, and the gauge has no way of knowing what material it is sitting on -- so a gauge calibrated on a steel block and used on aluminium reports a number that is precise, repeatable, documented, and wrong by about 7 percent. THE ERROR IS A PERCENTAGE, AND IT SCALES WITH THE WALL. That is worth stating plainly because it is easy to carry an absolute figure across from one thickness to another: 7 percent of a 5 inch section is nearly four hundred thousandths and 7 percent of a half-inch wall is thirty-four, and both are serious for different reasons. The figure that matters is the one computed at the wall actually being inspected. THE TWO DIRECTIONS ARE NOT EQUALLY DANGEROUS. A gauge reading LOW retires equipment early, which costs money. A gauge reading HIGH leaves a wall in service that is already below its retirement limit, and the report says it passed. Against a stated limit this reports the thinnest wall that still reads acceptable, because that gap is the exposure the calibration error creates and it is invisible in the reading itself. Coating is the other silent overread. Sound crosses paint at roughly a third of steel's velocity, so in single-echo mode a coated wall reads thick by considerably more than the coating's own thickness -- again in the unsafe direction. Echo-to-echo mode times between two back-wall echoes and ignores everything in front of them, which is why it is the mode for any coated corrosion survey, and why a survey done in the wrong mode on a painted tank is not conservative. The rule that follows from all of it: CALIBRATE ON A KNOWN THICKNESS OF THE ACTUAL MATERIAL, and re-calibrate when the material changes. A velocity table is a starting point, not a calibration, because velocity varies with alloy, temperature and grain structure. This is the velocity relation and its consequences. It does not select a transducer, frequency or couplant, address surface preparation, curvature, or high-temperature measurement (velocity falls with temperature, and a hot survey needs its own correction), interpret A-scans, distinguish general wall loss from pitting (a gauge finds the thickness under the probe and a pit between readings is missed entirely), evaluate laminations or inclusions, or determine a retirement limit. API 570, API 653 or the applicable inspection code, the equipment's fitness-for-service assessment, and the certified inspector govern.

thickness = velocity x round-trip transit time / 2. The velocity error is a PERCENTAGE, so its absolute effect scales with the wall being inspected; the thinnest wall that still reads at a retirement limit is limit x gauge velocity / actual velocity.

The ultrasonic pulse-echo thickness relation. Velocities are ENTERED (steel about 0.232 in/us, aluminium about 0.2490) because they vary with alloy, temperature and grain structure -- a published table is a starting point and not a calibration.

One relation and two entered velocities.

Estimate. AHJ and licensed professional govern.

Field names used by the API: transit_time_us, gauge_velocity_in_us, actual_velocity_in_us, nominal_wall_in, retirement_limit_in, coating_thickness_in, gauge_reading_in, true_thickness_in, error_pct, error_on_nominal_in, thinnest_passing_in, coating_overread_in

• Velocities are entered they vary with alloy, temperature and grain structurea calibration block of the actual material
• A gauge finds thickness under the probe a pit between readings is missed entirelythe inspection plan and grid
• Not a fitness-for-service assessment and it does not set a retirement limitAPI 570 / API 653