Fred Starr recollects – hydrogen attack in steels
I am pretty sure that I am one of the few people who has worked on a corrosion problem that almost got them killed. It was not through a lab accident, but in my previous job on a steam reforming plant, where the frightening chaos stemmed from the perforation of a tube in a reformed gas boiler. This was, fortunately, nothing like the disaster at the Tesoro Anacortes Refinery in 2010, where fatalities did occur.
Unlike the reforming plant incident, where we had to risk our necks, there was an official report from an arm of the US Government. Its metallurgical findings about hydrogen attack should be known to everyone in the petrochemical field. Especially cost-driven plant designers.
I will be coming to Tesoro Anacortes later, but the common thread throughout this piece is the speed at which hydrogen moves through steel, at quite moderate temperatures. I learnt this soon after taking over the Met Lab at London Research Station, where my predecessor, Dr Peter Neufeld, had designed a really elegant experiment to assess corrosion in the aforesaid reformed gas boilers.
The corrosion was caused by deposits of wet potassium carbonate, at about 400°C, attacking the steel. The hydrogen liberated, diffused clean through the steel and out the other side of the test piece. So much so, its pressure could be measured.
Moving on, one of my jobs as a failure investigator was to give advice on hydrogen attack on the tubes of a shift convertor boiler. This particular type of boiler – one of the shell and tube type, a bit like the boiler in a steam locomotive – ‘saw’ a process gas containing hydrogen at a very high partial pressure.
Hydrogen attack is not the same as hydrogen embrittlement, whereby atomic hydrogen cracks high strength steels in a very brittle manner. Hydrogen attack is a high temperature form of corrosion, in which the hydrogen, that gets into the steel, reacts with iron and other carbides to form methane.
The methane can’t escape, so the steel ends up full of trapped pockets of the gas, greatly weakened, and forming blisters on the surface. To our surprise, long after a metallographic sample had been taken from a tube, after covering the surface of the specimen with oil immersion fluid, we could see bubbles of methane still emerging.
The lab work was purely confirmatory. The people running the plant and myself had agreed that plain mild steel was not adequate for the job. Simple iron carbides, present in the pearlite in carbon steel, are easily attacked if the temperature and hydrogen pressure are even moderately high. The next step was to look at the Nelson Curves to see what they advised about which tube alloy to substitute. We went for a 1Cr-0.5Mo steel – chromium and molybdenum carbides being more resistant.
The Nelson touch
George Nelson was an American, now deceased, who, during the 1950s, brought together information about how various steels responded to hydrogen attack. From his contacts in the refineries, he was able to delineate safe operating regions for plain carbon and low-alloy steels. For example, the curve for carbon steel shows that below 100psi there are no problems, but at pressures above 500psi (34bar), the temperature must be kept below 500°F (260°C). Even lower temperatures are needed at higher pressures. The resulting Nelson Curve is L shaped.
The curves are a great way to show what is happening, but they are quite empirical. Over the years there have been several attempts to derive a theoretical underpinning, without much success. It is an almost intractable problem, involving the stability of iron and other carbides, the diffusion rate of hydrogen, time of exposure, creep strength of steel at temperature, and the reaction kinetics of carbides with hydrogen.
I heard about the Tesoro Anacortes Refinery disaster through a recent zoom conference organised by European Technology Development. I followed it up via the American Chemical and Safety Board’s investigation report.
The report showed that, as a result of long-term hydrogen attack, part of the actual shell of a long thin pressure vessel, containing a heat exchanger, simply peeled open, releasing a mixture of hydrogen and naphtha, which immediately caught fire. Seven of the operating staff either died there and then, or succumbed to burns over the next few days. Pictures taken, after the incident, show that the hydrogen and naphtha had been expelled with such force that the bundle of heat exchanger tubes had been forced out of the opening.
The catastrophic peeling back of the pressure vessel was caused by the growth of metre-long cracks running longitudinally, and in line with a girth weld. The cracks were subsurface and seem not to have been picked up in previous plant inspections. But the main point is that, according to the Nelson Curves, they should not have been there. Consequently, no one looked too hard. The Safety Board report highlights eight other cases where the Nelson Curves have not worked out as they should.
I now recollect another failure on the steam reforming plant, where I worked, in which a big pipe, downstream of the shift converter, failed quite inexplicably. Could that have been hydrogen attack, I wonder? The thought also occurs that there is an even more well-known Nelson, Horatio the Admiral, who turned a blind eye in one sea battle and got away with it. But you would indeed be trusting to luck if you stick to obsolete and dubious hydrogen attack data. In a nutshell, don’t blindly follow the curves.