July 9, 1990
Gerrit M. Buchheim Exxon Research & Engineering Co. Florham Park, N.J. Large-scale inspections during the past 5 years of pressure vessels operating in wet H2S service and an American Petroleum Institute (API) survey have revealed that, industry wide, about 25% of all vessels inspected contained cracks. Individual companies and locations have experienced cracking in as many as 45% or more of the vessels inspected.
Gerrit M. Buchheim
Exxon Research & Engineering Co.
Florham Park, N.J.

Large-scale inspections during the past 5 years of pressure vessels operating in wet H2S service and an American Petroleum Institute (API) survey have revealed that, industry wide, about 25% of all vessels inspected contained cracks.

Individual companies and locations have experienced cracking in as many as 45% or more of the vessels inspected.

Such vessels are found in fractionators and overhead systems for fluid catalytic cracking, hydroprocessing, gas processing, sour gas and water treatment, and others where high-sulfur feedstocks are upgraded. Many of the cracks were so severe that a significant number of vessels have been replaced.

The inspections were prompted by the catastrophic failure of an amine absorber vessel in Unocal Corp.'s Lemont, Ill., refinery on July 23, 1984.1 The cause of the failure was wet H2S cracking.

Data collected by the API survey, and data from inspections of vessels at Exxon Co. U.S.A. refineries, revealed the types of cracking encountered.

With those data as a foundation, the mechanisms that cause cracking2 3 are presented here, along with repair techniques and information that will help prevent recracking of repaired areas and cracking of new vessels.


Although recognized in the processing industry for many years, the extent of the problem of wet H2S cracking wasn't considered a major problem until the failure of the Unocal amine absorber.

It was considered more of a problem in oil and gas production.

Cracking induced by hydrogen in the presence of H2S has been researched extensively, and practical guidelines have been developed for steels used in production and pipelining. Those segments of the oil and gas industry have found that there are at least two types of cracking mechanisms: sulfide stress cracking (SSC) and stepwise cracking (SWC). In order to minimize the risk of SSC, the National Association of Corrosion Engineers (NACE) Standard MR-01-75 is used in the production segment to define maximum hardness for steels in wet H2S service.4

Hardness is a contributing factor to SSC susceptibility. To minimize SWC, improvements to the steel-making process were instituted because SWC is linked to steel cleanliness.

In refining, NACE Standard RP-04-72 was used to define acceptable hardness limits for carbon-steel weldments to avoid SSC.5 Hydrogen blistering, another type of hydrogen damage observed in refineries, had been observed for many years.

However, blistering was not considered catastrophic and, therefore, very little research was conducted to improve the resistance of steels to blistering. Furthermore, vessels operating in wet H2S service were not normally inspected for cracking.

As a result of the absorber failure and research conducted since the failure, the refining industry now generally accepts that there are actually three different types of cracking resulting from wet H2S environments.

They are: SSC (sometimes referred to as hard-weld cracking), hydrogen-induced cracking (HIC, also referred to as blistering or hydrogen damage), and stress-oriented, hydrogen-induced cracking (SOHIC).

All three types of cracking require the production of hydrogen atoms at the steel surface by a corrosion reaction in an H2S-containing, aqueous solution. Although there is no established lower limit of H2S concentration that causes cracking, the industry practice has been to assume a lower threshold of 50 ppm H2S- Concentrations below 50 ppm do not cause cracking.

The hydrogen atoms produced at the steel surface have a large affinity to combine together to form innocuous hydrogen gas. However, in the presence of FeS and/or cyanides, the hydrogen gas combination reaction is poisoned, allowing some of the hydrogen atoms to diffuse into the steel.

Only atomic hydrogen can diffuse into the steel because the molecular hydrogen molecule is too large. The way the hydrogen affects the steel differentiates the three types of cracking.


In SSC, the hydrogen atoms affect the flow and fracture behavior of the steel and result in what is generally termed as hydrogen embrittlement.

The atoms remain dissolved in the matrix and are highly mobile.

SSC is very dependent on the composition, microstructure, strength, and the applied and residual stress levels of the steel.

With the types of steel used in the refining industry, cracking is observed in the welds or heat-affected zones in the base material adjacent to welds (Fig. 1). This is where high-strength/low-ductility microstructures may be present and can be identified by high hardness.

Limiting hardness, therefore, is one practical method of reducing the susceptibility of a particular steel to SSC. Nonetheless, field experience and metallographic studies have shown that a macro-hardness limit, such as is called for in NACE RP-04-72, is insufficient to prevent SSC.

Small localized hard zones can be present in the heat-affected zone and can initiate SSC, even though the macro-hardness is low.

As a result, micro-hardness limits have greater technical basis, but measuring micro-hardness is not practical in production-weld testing. Proper welding electrode selection and welding procedures will minimize the hardness of welds and heat-affected zones.

Nevertheless, it has been found that postweld heat treatment (PWHT) is necessary to minimize the susceptibility to SSC. PWHT serves two purposes: it tempers the microstructure and reduces the residual stresses.

SSC is essentially the same mechanism as what is known as delayed cracking in weldments. SSC and delayed cracking have the same metallographic appearance.

Therefore, for existing vessels, it is not possible to differentiate between a crack that formed during fabrication and one that formed in subsequent service.


In HIC, the hydrogen atom collects at imperfections or nonhomogeneities in the microstructure-elongated sulfide inclusion or oxides (Fig. 2). It occurs throughout the base metal of steel and typically does not occur in weld metal.

The interface between the inclusion and the steel matrix acts as a site for the hydrogen atoms to combine into hydrogen gas.

The hydrogen gas molecules are too large to diffuse back out through the steel and are trapped.

As more hydrogen gas collects at the interface, it builds up pressure and can cause yielding of the surrounding steel. That yielding creates lamination-type fissures.

These fissures are located parallel to the steel surface because the inclusions are elongated in the same direction during steel manufacturing.

If the fissures are located close to the inside surface of the vessel, the steel can bulge and result in classical hydrogen blistering.

Parallel fissures can also link up in the through-thickness direction without an apparent interaction with stress level. But this is very rare in the refining industry.

HIC is primarily a function of the steel cleanliness, which relates to the impurities in the steel and the method of manufacture. One way to avoid the situation is to use newly developed HIC-resistant steels, provided they are confirmed by qualification testing to be resistant to HIC.

These steels minimize sulfur and oxide levels and rely on special steelmaking techniques. Because HIC occurs in the base metal of the steel and is independent of stress level, PWHT will not improve resistance to HIC.


In SOHIC, the hydrogen atoms collect at imperfections in the microstructure, just as in HIC. But the fissures stack themselves in the through-thickness direction.

This occurs in the heat-affected zones or other areas of high stress concentration (Fig. 3). Gross blistering and severe HIC are not necessary for SOHIC.

Stress-oriented hydrogen-induced cracking (SOHIC) was first identified in the failed amine absorber. In that failure and several other instances, SOHIC was observed to be present in conjunction with sulfide stress cracking.

A crack might initiate by SSC, and as a result of the stress concentration, grow by SOHIC (Fig. 4). SSC is, however, not a precondition for SOHIC.

In general, cleaner steels that minimize HIC will also minimize SOHIC. But data suggest that these types of steels may be more susceptible to SSC.

Because SOHIC is stress related, post weld heat treatment (PWHT) will directionally improve resistance to SOHIC. But tests have shown that SOHIC can occur at nozzles, highly restrained welds, or other stress concentrations where PWHT has reduced residual stress levels.

Some of the latest generation of HIC-resistant steels appear to be truly resistant to SOHIC, particularly after PWHT. But qualification tests are required to ensure their resistance.


Since the amine absorber failure in 1984, thousands of vessels operating in wet H2S service have been inspected. Metallographic examinations of cracked vessels have confirmed many cases of SoHIC.

However, because SSC and delayed cracking have the same appearance, it is difficult to determine whether cracks occurred during fabrication or in service. Those vessels were not inspected after fabrication with wet fluorescent magnetic particle testing (WFMPT), a nondestructive test that has been proven to have a much greater sensitivity than radiography or magnetic particle examination for crack detection.

Therefore, it is possible that some cracks were present before the vessel entered service. As a reference point, new vessels that were inspected by WFMPT have not shown a high incidence of cracking.

Of course, fabricators probably increase their attention to welding quality control when they know vessels will be WFMPT examined.

A 1988 API survey provided some industry experience, and NACE Committee T-8-16a is also establishing an industry data base.6 The findings discussed here are based on the API survey, and on inspections performed at Exxon affiliated refineries worldwide.

The 1988 survey found, as noted, that about 25% of vessels inspected contained cracks, with 75% of these vessels having cracks deeper than the corrosion allowance for the vessel. About one third of the cracked vessels were replaced.

There appears to be large differences between the extent of cracking reported by different companies and individual refineries, ranging from 6 to 45% of vessels inspected (Table 1). These differences would relate to differences in the crude oils processed, the materials selection and fabrication procedures used, and possibly the extent of inspections performed.

The API survey reported that 25% of the vessels with cracks had undergone PWHT. However, it was not possible to extract cracking percentages for postweld heat-treated vessels.

General discussions within the refining industry indicate that postweld heat-treated vessels have only slightly lower cracking incidence and replacement rates than vessels not postweld heat treated. Future surveys will, it is hoped, be able to present a clearer picture of the role PWHT plays in the extent of cracking.

Research and conventional wisdom would predict a beneficial effect from PWHT in reducing wet H2S cracking. But a clear beneficial effect has not been observed in the field to date.

Explanations for this could be that: PWHT, in fact, does not significantly help; cracks were already present before PWHT; or that the PWHT temperatures were not sufficiently high to be effective.

The API survey included data from many different refinery units, such as cokers, FCCUs, pipestills, hydrorefining units, sour-water strippers, etc. But the data were not classified on a refining-unit basis.


Experience at Exxon refineries has been that most types of units processing wet streams containing greater than 50 ppm H2S have experienced cracking. Furthermore, within the accuracy of the data, the percentage of cracked vessels is quite similar among the various units. Most of the Exxon data were generated from inspections of FCCU overhead and light-ends systems. But the percentage of cracking found in the overhead systems of hydrotreaters is very similar.

One exception appears to be pipestill units. These have experienced only about half the cracking percentage of other units.

All types of conventional mild carbon steel, such as ASTM A 212B, A-285C, A-515 Grade 70, A-516 (all grades), and A-70, have experienced cracking (Table 2). Within the accuracy of Exxon data, other than for A-516 Grade 60 steel, all grades appear to have similar percentages of cracking incidence and severity.

A-516 Grade 60 steel is a low-strength steel and appears to have a lower incidence of cracking. Future survey results, it is hoped, will be able to clarify the extent that steel grade is a factor in cracking tendency and severity.

Because HIC-resistant steels, such as A-516 Grade 70, have only been in use for a few years, field inspection data are not yet available.

Vessels of various service lives have been found to contain cracks. Vessel ages range from less than 2 years to more than 40 years. The data provide no evidence that age itself is an important variable in wet H2S cracking.


Because of the large number of vessels operating in wet H2S service, and the inability to conduct inspections of these vessels quickly, each company and individual refinery has had to prioritize its inspections. Criteria differ somewhat for various locations.

In general, those vessels with a history of cracking or blistering are inspected first, followed by those with alterations that were not postweld heat treated. Finally, the rest of the vessels are inspected when they become available for inspection.

Some locations prioritize inspections based on the consequence of a failure. Other parameters, such as scheduled turnaround intervals, projected running plans, maintenance and inspection resources, etc., factor into the decision of if and when a particular vessel is inspected.

The majority of inspections are performed using wet fluorescent magnetic particle testing (WFMPT). That technique requires vessel entry.

Therefore, vessels inspected by this method are usually done at turnarounds. But due to long turnaround intervals and the number of vessels in need of inspection, many vessels have not yet been inspected for the first time.

In addition to the need for first-time inspection, reinspection may be required for some vessels that previously contained cracks or are subject to SOHIC. One future challenge is to define appropriate levels and intervals for reinspection.

Another factor to consider is the extent of the inspection of a particular vessel. Unless acoustic emission testing (AET) is performed (AET continuously monitors for cracks), only the welds and near-weld regions of a vessel are usually inspected.

Inspection coverage ranges from a few feet of weld to 1 00% of welds, depending on the refinery, vessel, and inspection technique used. The extent of inspection coverage becomes a very complex issue when large vessels, such as towers, are inspected because of the time involved to prepare the vessel for inspection, inspect it, and repair cracks.

Areas typically examined when less than 100% of the welds are inspected include the head-to-shell weld, a tray support weld, a nozzle weld, and a girth weld.


An appropriate inspection method must be selected that can effectively detect wet H2S cracking. It must also be one that keeps inspection time and cost under control.7 By far the most widely used method is WFMPT. It is the most sensitive method available, and it will detect even very small surface-breaking cracks.

One drawback is that it is so sensitive that interpreting indications is critical. The method usually requires that inspectors be qualified in the method.

Other drawbacks are that the method detects both active and inactive cracks, and vessel entry is required for its use.

In order to perform WFMPT inspection, the surface must be cleaned of all hydrocarbon residue and iron scale. Typically, the weld regions are cleaned by abrasive blasting. The major disadvantage of abrasive blasting, especially in trayed towers, is the time and cost associated with removing the spent abrasive.

Some experiments with waterblasting and hand-tool cleaning have shown that these cleaning methods may not reveal all of the cracks that abrasive blasting will reveal. Some locations use a flapper-wheel abrasive after blasting to increase the visibility of indications.

Other techniques that might offer promise are chemical cleaning and microabrasive waterblasting.

Once a vessel is cleaned, inspection should be performed within a few days to avoid rust forming on the cleaned surfaces.

Ultrasonic testing (UT), straight or angle beam, can be performed from the outside of the vessel while the vessel is in service. The method is not as sensitive as WFMPT and cannot detect relatively shallow cracks, and interpreting indications of HIC is very difficult. One advantage is that the method can detect subsurface cracking.

Some refineries use UT to inspect vessels prior to turnaround to try to identify vessels that might need replacement after WFMPT inspection is performed at turnaround.

Acoustic emission testing (AET) is gaining acceptance as a screening tool. It is, however, not usually used as a stand-alone method.

AET can be performed while the vessel is in service in some cases, but most inspections are done off-line. Vessel entry is not required unless there is a need for follow-up inspection.

AET can be used to inspect the entire vessel for cracking and can then pinpoint areas that need follow-up inspection by UT or WFMPT. AET will detect only active cracks and cannot determine their depth or length.

It will not identify inactive cracks that could later become active under certain operating conditions. The interpretation of AET indications is very complicated and there are only a few qualified vendors that offer the service.


Once a crack is found, the depth of the crack must be determined to decide how the vessel should be repaired. UT angle-beam testing can be done, or the crack may be simply ground out.

Many cracks are shallow and can be removed by light grinding. If the ground-out depth is within the vessel's corrosion allowance, no further action is typically taken. If, however, the ground-out depth exceeds the corrosion allowance, the area must be repaired by welding.

If welded repairs are required, it is prudent to conduct a weldability test and even a heat treatment test to determine whether the material can be repaired by welding and subsequently heat treated. There are many cases where a steel has repeatedly cracked after repairs were made because hydrogen was present in the steel.

If hydrogen is present in the atomic form, it can be baked out at about 450 F. But if it is molecular hydrogen, the steel will be difficult to repair without a high risk of recracking.

Use of weld procedures that reduce hard structures and use of PWHT are often required for repairing cracks due to wet H2S. It is also prudent to check welded areas and areas subjected to PWHT before returning the vessel to service.

The cracked area can also be strip lined with stainless steel, or it can be coated with refractory or metallic and nonmetallic coatings. There is a fair amount of experience with refractory coatings, but other coatings lack long-term field experience.


If a vessel has to be replaced, material selection of the new vessel becomes important. If it is replaced with standard steel material, with the advantages of low cost and quick delivery, future inspections will have to be conducted, and the potential for cracking is not reduced.

If the vessel is replaced with HIC-resistant grades, improved resistance to HIC and SOHIC will be provided. But not all HIC-resistant steels are, in fact, resistant, and qualification testing should be done.

Using resistant grades should reduce recurring cracking problems, but the materials cost about 30% more than standard steel. Delivery times are also increased, although because of higher demand that situation is beginning to improve.

Resistant steels are also more difficult to weld and may actually have increased susceptibility to SSC. PWHT is recommended for resistant materials.

Stainless steel or clad material eliminates future cracking tendencies, but material cost can be 300% higher than standard materials. Delivery times can also become long.


There are some important challenges facing the industry regarding wet H2S cracking. Inspection intervals need to be defined based on the likelihood that cracking will continue to occur in a particular vessel.

The effect that processing increasing amounts of sour crude will have on corrosivity and the potential for increased cracking tendency needs to be determined. There is also a need to determine the extent to which piping and liftings should be inspected for cracking.

And there needs to be an increase in supply of HIC-resistant steels that have been prequalified as to their cracking resistance.

Faster and cheaper surface preparation techniques should be developed for WFMPT. Further development of UT and AET inspections for cracking is also needed.

Some work should be done to explore the effectiveness of corrosion inhibitors to minimize hydrogen entry into the steel.

Finally, appropriate qualification tests for HIC-resistant steels should be developed, along with a determination of which HIC-resistant steels are truly resistant to SSC.


  1. McHenry, et al., NBSIR 86-3049, "Examination of a Pressure Vessel that ruptured at the Chicago Refinery of the Union Oil Co. on July 23, 1984," National Bureau of Standards, Washington, D.C., March 1986.

  2. Merrick, R.D., Materials Performance, January 1988, p. 30.

  3. Merrick, R.D., and Bullen, M.L., NACE Corrosion/89, paper No. 269, New Orleans, April 1989.

  4. NACE Standard MR-01-75, "Standard Materials Requirements-Sulfide Stress Cracking Resistant Metallic Materials for Oilfield Equipment," National Association of Corrosion Engineers, Houston.

  5. NACE Standard RP-04-72, "Standard Recommended Practice-Methods and Controls to Prevent In-Service Cracking of Carbon Steel Welds in P-1 Materials in Corrosive Petroleum Refining environments," National Association of Corrosion Engineers, Houston.

  6. API Survey, "Extent of Equipment Cracking in Wet Hydrogen Sulfide Environments," presented by T. McLaury at the American Petroleum Institute midyear refining meeting, Chicago, April 1989.

  7. Davies, R., et al., NACE Corrosion/90, paper No. 214, Las Vegas, April 1990.

Copyright 1990 Oil & Gas Journal. All Rights Reserved.