Velocity correction for temperature

Sound velocity in metals falls as temperature rises: the material softens, the elastic moduli drop, and the wave slows. A thickness gauge calibrated cold but used hot measures the true time of flight and then multiplies it by the cold velocity, so it over-reads the wall. On hot piping and vessels this is the single largest source of error in corrosion monitoring, and it always errs on the unsafe side by making the wall look thicker than it is.

The correction is c(T) = c_ref · [1 − k·(T − T_ref)], where k is the fractional change in velocity per degree. For carbon and stainless steel the working rule is about 1 % per 55 °C, which is k ≈ 0.00018 per °C — roughly one metre per second per degree. Plastics change several times faster, which matters for the delay line and the wedge as much as for the part.

Two ways to deal with it: recalibrate on a reference block held at the working temperature, which is the preferred method and takes the couplant and delay-line behaviour into account as well; or measure cold, measure hot, and correct arithmetically. Either way the correction must be documented, because a repeat inspection taken at a different metal temperature will otherwise show a false change in wall thickness.

Above about 150 °C the probe itself becomes the limitation: standard probes debond, high-temperature couplant burns off, and contact must be brief. Use a probe rated for the temperature, keep the contact time short, allow the delay line to cool between readings, and treat the coefficients here as rules of thumb — verify them on a heated reference block whenever the reading is used for a fitness-for-service decision.

c(T) = c_ref · [ 1 − k · (T − T_ref) ]
Thickness over-read [%] = ( c_ref / c(T) − 1 ) × 100
True thickness = displayed thickness × c(T) / c_ref