Hydro processing reactor integrity: The basics (November 2004)

Brian Cane ¹ and Julian Speck ²

¹ Brian Cane ( brian.cane@twi.co.uk) is responsible for TWI's services to process and power industry members.

² Julian Speck ( julian.speck@twi.co.uk) is TWI's structural integrity technology group manager, responsible for FFS and RBI activities.

Paper published in Inspectioneering Journal, November 2004.

Demanding cleaner fuels

Environmental regulators drive refiners to introduce bottom-of-the-barrel conversion hydroprocessing units, to produce cleaner fuels. These include hydrotreaters and hydrocrackers with reactors that operate at high temperature, andpressure and in the presence of hydrogen. (Catalytic reforming units are in the same category, with respect to the challenge of ensuring the integrity of reactor vessels).

The planning of reactor inspections is complicated because it requires consideration of the resistance to pressurised hydrogen, alloy stability at operating temperature, ambient temperature properties (for heavy-walled vessels),resistance to high temperature H ₂ /H ₂ S corrosion, and the resistance of austenitic alloys to stress corrosion cracking (SCC).

Vintage reactor designs

Generations of hydroprocessing technology have lead to advances in reactor design and fabrication:

1960s - Migration from cold-wall to hot-wall Cr-Mo alloys
1970s - Toughness requirements introduced for temper embrittlement
1980s - Steel composition control reduces temper embrittlement susceptibility
1990s - Acute composition controls try to eliminate temper embrittlement

2¼Cr1Mo has been the alloy of choice for reactors, for 40 years, and new grades of vanadium modified 2¼Cr and 3Cr alloys have now been introduced with high strength levels. This gradual evolution follows from a greater understanding of reactor in-service damage mechanisms, which has also led to significant advances in in-service inspection technology.

Critical pressure equipment

Reactor operating conditions now range from 400°C (~750°F) and 10MPa (~1.5ksi) hydrogen partial pressure, to 482°C (~900°F) and 35MPa (~5ksi) partial pressure. The nature of the inspections and assessments of this critical equipment and their timing in relation to design life, is influenced by three main factors: the nature of the plant and its operation, the economic consequences of failure or unplanned shutdown, and the safety hazards involved. (An early assessment would be necessary if a reactor has been subjected to abnormal temperatures above those specified in design).

Within a RBI framework, it's worth remarking that hydrogen has some unusual properties. It will ignite or explode in air mixtures of 4-75% (a flammability range ten times wider than gasoline), and it has a minimum ignition energy about 10% that of natural gas (a much weaker spark will ignite it). In addition, hydroprocessing units are often at the economic heart of a refinery.

Reactor damage mechanisms (DMs)

No attempt has been made to cover all reactor DMs in this article. API RP571 provides a useful listing of potential in-service DMs, if operators have not yet identified these during their in-house RBI assessments. These include: creep; sulfidation; hot hydrogen attack; high temperature H ₂ /H ₂ S corrosion; temper embrittlement; hydrogen embrittlement; sigma/chi phase embrittlement; brittle fracture; SCC of austenitic materials (due to chlorides and polythionic acids). Any attack varies in severity andlocation, depending primarily upon metal temperature. Some of the most peculiar reactor DMs are:

• Hot hydrogen attack results in internal surface decarburisation (on unclad reactors) and the formation of internal base metal methane micro-voids at grain boundaries, that may accumulate to produce macro-cracks.
• Another hydrogen effect is hydrogen-assisted crack growth (due to hydrogen embrittlement) that can occur when the vessel is cooled too quickly to ambient temperature, after the steel has absorbed hydrogen in service.
• CrMo alloys suffer temper embrittlement due the presence of specific 'tramp' elements in the steel and after years of exposure in the range of 370-570°C (~700-1050°F), resulting in an increased ductile-to-brittle transition temperature (a concern if shutdown and start-up thermal stresses are severe).
• High temperatures and unforeseen loads cause stress redistribution leading to accelerated creep cracking in the lower ductility regions of welds (which are micro-structurally complex, with varying mechanical and creep properties), Fig.1.

Cladding integrity

In clad vessels, the potential causes of cladding degradation include: microstructural changes (sigma/chi phase embrittlement, and sensitisation leading to loss of corrosion resistance to H ₂ /H ₂ S attack), low-cycle fatigue cracking (due to thermal gradients across thick sections), and loss of adherence (debonding).

Embrittlement and sensitisation of the 18Cr8Ni cladding can occur either during original PWHT or in operation. It has been known to lead to cracking in the cladding, and the subsequent penetration of these cracks into reactor walls. Some of the highly stressed areas where clad cracking occurs are in the ring grooves in RTJ flanges (due to the high stress concentration in the root of the groove), internal corners of large nozzles, and the corners of internal bed supports.

Thermal stresses and hydrogen drive the debonding mechanism (cracking along the weld overlay and the Cr-Mo base metal interface), which tends to occur several hours after a shutdown. The overlay is also susceptible to polythionicacid or chloride SCC, if it sensitises before or during service. Some operators wash all reactor circuit clad surfaces with an alkaline solution to neutralise any such SCC hazards during shutdowns.

Critical inspection locations

The TWI approach to reactor integrity management combines non-destructive examination (NDE), prior experience from previous reactor inspections, and fitness for service (FFS) assessment. The most critical areas requiring inspection relate to the head-to-shell junction, nozzle details, internal supports and the skirt-to-vessel joints. The maximum stress resulting from combined pressure, weight, wind, and thermal loads occurs at the skirt attachment, and so-called' vintage design details' are particularly susceptible to cracking, Fig.2.

The tasks involved in the process are: (1) assessment by calculation, based on historical operating data; (2) NDE using predictive in-situ metallographic replication to detect micro-cracks; and (3) NDE using conventional techniques(UT flaw detection, and internal and external surface inspection using MT or PT), to detect macro-cracks. The expected life of components is determined on the basis of published ASTM/ISO material property databases. (An elastic stress analysis is usually performed because time-dependent creep solutions require knowledge of the materials creep behaviour at critical locations). In the event conventional NDE detects macro-cracks, a disciplined UT campaign to size the cracks will be immediately implemented for FFS assessment purposes.

The complexities of NDE

The NDE requirements for reactors are not complicated; the conventional techniques described above are adequate for the majority of cases. The NDE issues are sometimes complicated by the advice of service providers, offering techniques like acoustic emission testing (AET), automated ultrasonic testing (AUT), etc.

Such advanced techniques have their place in reactor management. AET is a 'qualitative' technique that can detect small-scale damage during pressurisation and plant operation. It can be used to monitor all reactor circuit equipment simultaneously (reactors, pipework and exchangers). The benefit of using AET may be reduced downtime since subsequent conventional NDE is focused on locations of crack activity, identified by AE prior to the shutdown.

AUT techniques include external 'C-scan' mapping of the shell (to detect internal damage), and backscatter and velocity ratio measurements (to monitor the percentage of hot hydrogen damage). Both of these ideally require 'baseline'(reference) inspections; they become highly effective at future repeat inspections, to monitor any increase in damage. The time-of-flight diffraction (TOFD) technique is increasingly used to detect and measure macro-cracks, and asizing accuracy of ±1mm is claimed for site conditions. TOFD is the preferred UT technique if crack sizes are needed for FFS assessments or remaining life calculations.

When operators use advanced NDE techniques, they should be aware that original manufacturing flaws are the most common type of flaw found, since many vintage reactors were not subjected to the same level of pre-service inspection.

Crucial success factors

The design of all reactor NDE programs should account for the following realities:

• The prospect of accurately quantifying all stresses by calculation is remote, even if all external system stresses can be identified. In-service damage should therefore be directly assessed by physical examination.
• The FFS approach cannot completely deal with the complex nature of weldment properties, therefore NDE programs should always be implemented to compliment FFS assessments.
• Historical operating records sometimes insufficiently describe operating conditions for the purposes of life assessment, so direct physical examination is needed to provide confidence in future mechanical integrity.
• Only metallographic replication can detect the earliest stages of micro-damage (replication is in the order of 1,000 more sensitive than conventional NDE), to reliably establish future conventional NDE intervals.

Once micro-cracks have been detected, the subsequent period between crack initiation and final failure can be significant. Crack growth propagation rates may be determine by established FFS assessment procedures. One of the important ingredients in the assessment is relevant material creep crack growth data (in the 'aged condition', rather than of 'virgin material'). Reliable creep crack growth data on weldments can be quickly obtained from published sources. For example, Cambridge Scientific Abstracts' WELDASEARCH (www.csa.com), provides an extensive database of abstracts to articles, reports, standards, etc. from which the best data can be sourced.

Future inspections and assessments

The timing of subsequent actions will depend on what is discovered about the state of a reactor, taking into account the future operating regime. When there is little detectable deterioration, the next inspection plan is as before, Table 1. A thorough FFS assessment of remnant life requires an extension investigation. This can be very costly so this is normally only required when micro-cracking is detected. Regular physical monitoring is always the preferred safeguard against failure.

Table 1 Reactor inspection and assessment philosophy

Repair welding

When cracks are found in reactors, repair welding is usually carried out. The excavation and removal of all creep damage prior to welding is essential for a successful repair. Temper-bead welding procedures are frequently used as an alternative to PWHT. The design of the repair procedure must account for toughness requirements, residual stresses and hydrogen embrittlement susceptibility.

A case study

The inspection of a hydroprocessing reactor with Type III nozzles, Fig.3, was carried out after about ~100,000 hours of operation.

Extensive cracking was found in the HAZ on the nozzle-side of the top inlet nozzle weldment. Excavation of the area and subsequent replication revealed cracking to be greater than 5mm in depth. A FFS assessment indicated that crack propagation would be rapid leading to failure during the next unit run. Reverse FFS assessment of crack growth rates (from measured crack depth) to determine stresses (the crack driving force), indicated that excessive pipework stresses as well as the geometric stress concentration created by the nozzle neck transition, contributed to rapid accumulation of damage.

It was recommended that the HAZ should be deeply excavated and weld repaired, with additional weld 'reinforcement' added to reduce the stress concentrating effect of the transition. It was also suggested that the design and maintenance program of the pipe constant load supports should be reviewed, Fig.4.