Hacksaw, what was the failure mode? Were you able to inspect the thermowell? Was the failure at the tip or more towards the base of the well? What is the material of construction for the thermowell - metallic or ceramic?

If failure was towards the base, the failure could be caused by hitting the thermowell natural frequency for vibration. If my napkin calcs are right, you are looking at ~25 feet per second in the pipe, which is not insubstantial. Since you say the design conformed to B19.3, I will assume that the frequency/vibration angle was covered in the design. However, below is a paper from ASME I found that covers the issue, if it helps.

https://www.asme.org/wwwasmeorg/media/resourcefile...

Since this is a (steam?) condensate line, is there the possibility of two-phase flow during startup? I'm thinking there is also a risk of liquid condensate slugs causing impact failure on the thermowell. These slug impacts would cause huge stresses on the thermowell, and could also startup failure.

This was a documented failure at a Reactor Test Facility. The report was dated in 1962, so pre PTC19.3-1974, but designed using Murdock's design criteria that later formed the 19.3 requirements. The shorter drilled bar-stock design in the West-side Cooling Loop appeard safe. Interestingly, Brock on the 19.3 Committee at the time, made recommendations that might have addressed the failure risk (Available at ASME https://cstools.asme.org/csconnect/FileUpload.cfm?... ). The latter is also available at the Naval Post Grad School Library as a pdf.

https://www.osti.gov/biblio/4786636-spert-iii-ther...
https://www.osti.gov/biblio/4826907 SPERT III Facility

Apparently failed during warmup 2500psig/100DegF and took several days to reach operating conditins of 2500psig/650DegF. The fitting failed in less than an estimated 100 hours. The design appears to satify the current 19.3TW during warmup and normal operating conditions. NPS16 SCH 160 Main Run.

The photos show failure adjacent to a weld connection involving a pipe fitting.

What's puzzling, is that the design also passes the current 19.3TW criteria interms of avoiding resonance.

Agreed, field faricated designs, are inviting failure risk, but are commonly encountered in where "protection tube" designs are used.

Plenty of photos and information regarding the failure and the process design.

Here is the data I came up with

Straight U-Dim. Root Tip Bore End
9.000" x 0.8460" x 0.8460" x 0.5460" x 0.2500")
Root Fillet 0.000 in
Exposed Length 3.543 in
Material 304L S.S.
Modulus E 28100 kpsi
Metal Density 7.970 gm/cm3 at 70 Deg F

Main Run
Fluid Water
Pressure 2500.0 psig ( 2514.7 psia )
Temperature 100.0 Deg F
Spec. Grav @ Temperature 0.993
Viscosity 0.699 cP

Flow Rate 10000.0 gpm @ 24.9 ft/s
Pipe i.d./Wall Thickness 12.812 / 1.594 Inches
Pipe Rd / Tip Rd 3503802 / 231362

Resonance at service conditions: 255.7 Hz

Rosemount Twisted Square Thermowells eliminate over 90 per cent of dynamic stress, a leading cause of thermowell failures. https://www.process-worldwide.com/reducing-fatigue...

Have they published documented aerodynamic force coefficients such as steady state drag, osc. drag, osc. lift, as wells as the spectral content of these coefficients? Basically advanced flow test.