Localized Loss Downstream of a Wash-Water Injection Point
A 6 in. Sch 40 A106-B sour water line (nominal 0.280 in., t_min 0.190 in.) runs from an overhead accumulator to a slop header at 150 psig and 130 F. Wash water is injected through a 1 in. quill at a tee, and the line then runs 30 ft horizontally to a 90 degree elbow. The wash-water rate is set by operations on an as-needed basis and has been cut twice in the past year to save utility cost. The CML at the 5 o'clock position, 1 ft downstream of the tee, read 0.260 in. three years ago; it now reads 0.185 in. The upstream CMLs and the 12 o'clock readings are essentially unchanged. The line is Class 1 and the stream carries H2S and ammonia.
1. State the most probable damage mechanism(s) and the process data you need before committing to a repair scope.
Read the worked answer
What the pattern tells you. Loss confined to the bottom of the line immediately downstream of the tee, with the upstream and top-of-line readings unchanged, is a flow-dependent, injection-related mechanism, not general sour-water corrosion.
- Ammonium bisulfide (NH4HS) corrosion in sour water, which is strongly velocity- and concentration-dependent.
- Erosion-corrosion / droplet impingement in the turbulent mixing zone. One foot downstream is in the worst of it.
- Poor mixing. If the quill does not disperse the water, the water can run along the bottom as a separate phase, so the metal sees an undiluted or locally concentrated stream while the top of the line sees almost nothing. That matches the 12 o'clock readings being flat.
Data to obtain before finalizing the repair:
- NH3 and H2S concentrations in the stream, and therefore the NH4HS concentration;
- wash-water rate history (the two reductions are the leading suspect, since less dilution means more concentrated NH4HS and possibly a different flow regime);
- wash-water quality, chloride and oxygen content;
- actual velocity and flow regime at operating and turndown conditions;
- pH and temperature;
- quill condition and orientation, and whether it can be pulled for examination.
The inspection finding alone does not tell you whether to change the metal or change the process. The process data does.
2. Can the line remain in service as found? Where do you place additional measurements, and what NDE approach do you use at the injection point?
Read the worked answer
No. 0.185 in. is below t_min of 0.190 in. The measured rate is (0.260 - 0.185) / 3 = 0.025 in./yr, so the location has been at or below t_min for some time already. This is a repair/replace or rerate decision now, not a monitoring decision.
Do not generalize from the one reading. The 0.185 in. value is a single point in the worst-known location; the rest of the line may be sound, which is exactly what makes a whole-line replacement argument and a 'the line is fine' argument both wrong.
Where to measure. Treat the injection tee, the 1 ft run and the elbow as one corrosion circuit:
- 360 degree readings (3, 6, 9 and 12 o'clock) at 0.5D, 1D, 2D, 5D and 10D downstream of the tee;
- the elbow extrados and intrados and the straight run just past it;
- upstream of the injection point, so you can separate injection-driven loss from general loss;
- the branch and the tee body itself, and the quill and nozzle if removable.
Method. Direct contact ultrasonic thickness with the insulation removed and the surface prepared, using the same instrument and technique as the historical readings so the comparison is valid. Radiography is a poor second choice through a tee and elbow; use it only for the branch weld. If the surface is pitted, use a grid rather than single points.
Immediate action. Replace or repair the mixing zone with corrected geometry (a properly sized static mixer or redesigned quill), then re-baseline with a new CML set.
3. Propose the durable fix. Justify each element by the mechanism you identified.
Read the worked answer
Restoring thickness without changing the mechanism simply restarts the clock.
- Process control. Restore and then control wash water: set a target NH4HS concentration and a minimum dilution flow, not an 'as needed' rate. An as-needed rate makes the corrosion rate a function of utility cost, which is the underlying cause here.
- Mixing. Confirm the quill disperses the water. Better a designed injection nozzle or static mixer than a straight quill discharging into a tee.
- Velocity. Check the mixing zone and elbow velocities against the range where carbon steel is normally acceptable in this service. If the velocity is high, increasing the spool diameter at the mixing zone is a legitimate corrosion control measure, not just a mechanical convenience.
- Material. An upgrade (e.g., Alloy 825 or a duplex stainless steel) can be appropriate, but only after a circuit-wide review: the elbow and downstream piping see the same fluid, so an upgrade confined to 3 ft of spool creates a dissimilar joint and moves the problem downstream. The upgrade also changes welding, NDE and inspection practice, so the WPS/PQRs and the inspection plan must change with it. It will not help at all if the real driver is a misdirected quill.
- Inspection. Create a permanent injection-point circuit with TMLs in the first 12 in. downstream of the injection point and at the elbow, monitored on a close interval until two consecutive surveys confirm the rate has fallen.