We have recently had a failure of the air coil installed in the primary reformer of our ammonia plant. The failure was preceded by tripping of the process air compressor and delay in restoring the air supply by 20 minutes. The coil failed at two points: at the inlet header and in the 3rd row from top. Material of construction of the top portion is P22. The header failed with a typical fishmouth failure which often results from overheating. The check valve in the air outlet was not sealing perfectly and hydrogen reverse flow is another possibility. Centrifugal process air compressor was loaded rather fast and traces of lube oil from seal leakage were found in the drains of the intercooler.
We suspect that either of three things could have happened: a) Overheating of the air coil due to inadequate supply of cooling steam through hand controlled valve; b) Failure of the check valve in the air outlet line leading to reverse H2 supply; c) Entrainment of oil in the air during start-up leading to auto-igniton and consequent explosion.
I would like to seek similar experience in other plants and ways in which air coil is protected during process air compressor trip. In particular, I would like to have your opinion on whether explosion in any pipe will be preceded by bulging of metal wall or would it be simple rupture with no effect on the immediate surrounding area of affected zone.
Regards and looking forward to receiving your comments,
Ashwani Kaul
I was present during a failure of the process air line to the secondary reformer on an ammonia plant a couple of years ago. In our case, like yours, the line ruptured during plant start-up on introduction of process air to the secondary reformer.
We did observe the fish-mouth type deformation that you describe at the point of separation. The force of the explosion was such that it caused a section of the 10" line to jump off the pipe supports and spiral around the CO2 absorber. We were fortunate that no one was injured.
The investigation revealed the following:
1. The check valve on the process air line failed allowing reformed gas to leak back into the air line forming an explosive mixture.
2. The protection steam supply was taken from a medium pressure boiler that recovered heat from the HTS effluent. Steam flow from this boiler was split between the primary reformer, LTS warm-up heater and the protection steam to air line.
At this point during plant start up, steam generation from the boiler would be low, therefore preferentially flow to the primary reformer, resulting in insufficient flow through the process air line. The protection steam flow was not metered.
3. Pressure taps for the process air differential pressure transmitter were not at the same elevation, giving an inaccurate indication of the differential across the process air control valve.
Can you describe the system you currently have installed for process air cut-in? Do you cut in process air using a small bypass around the main process air control valve?
Are you using a standard swing-type check valve or a piston assisted check valve?
I may be able to help you with this.
Kevin Nurse
-- 3:00 am process air compressor trips.
-- Primary reformer load
reduced.
-- Secondary reformer downstream process vent valve opened.
--
Steam opened to air coil manually thru CV from control Room.
-- 3:20 am
standby process air compressor taken on line.
-- 3:50 am loud explosion
heard. Air flow decreases to zero.
-- Primary reformer shutdown after 10
minutes.
Observations: 8" diameter hole in inlet header; fishmouth rupture in 3rd row from top (3" size); traces of oil in discharge piping; traces of hydrogen in header after opening the reformer furnace manholes.
The check valve is of horizontal swing type. We feel hydrogen explosion may not be the cause of rupture because hydrogen explosions are known to be more severe. Check valve on inspection was only minutely passing and no problems were experienced during earlier air compressor trips. Bulging observed on piping could be attributed to overheating only.
Carry-over of seal oil from overhead tank may be the reason as operation of CV was found faulty during inspection after rupture. Slug of oil could have reached inlet header and hit opposite end causing rupture. Or header could have ruptured when oil reached auto-ignition temperature at the inlet. We are trying to find ways of ascertaining this hypothesis.
Material of construction of coil is P22 (2.25Cr 1 Mo except for bottom 4 rows which is SS 321). What is the metallurguy in other plants? We had some interesting observations on choice of materials of construction.
Looking forward to comments/observations. Regards,
Ashwani
Ashwani:
It would be interesting to know what was taking place during the half hour interval between 3:20 am and 3:50 am.
Was air already inside the secondary reformer or were the operators building the differential in anticipation of cutting in air?
You mentioned seal oil carryover.
I have worked on seven ammonia plants: three Kellogg, three Braun and one Fluor. In these designs the process air compressor has a lube oil system but no seal oil system.
You also mentioned that the standby process air compressor was put online for the restart.
Is this standby process air compressor part of the original plant design or was it added to the system as a modification?
Does the main process air compressor have a seal oil system?
I will get back to you on the question on the metallurgy of the air preheat coil.
Kevin