F91, Grade 91, modified 9Cr-1Mo-V. It has the best creep strength in the chrome-moly family by a wide margin, which is why high efficiency power plants use it for main steam and hot reheat piping. It also has the least tolerance for heat treatment error of anything in the family, and the failures are expensive.
Where the strength comes from
F9 and F91 have nearly the same chromium and molybdenum, around nine percent and one percent. The difference is that F91 adds vanadium, niobium and controlled nitrogen.
Those elements form very fine, very stable carbonitride precipitates distributed through a tempered martensitic structure. The precipitates pin dislocations and resist the slow rearrangement that is creep. The result is a large improvement in creep rupture strength over plain 9Cr-1Mo.
The practical payoff is thinner walls for the same duty. Thinner walls mean lower thermal stress on startup and shutdown, which means a plant can cycle faster, which is exactly what modern grid operation demands.
The strength is in the microstructure, not the chemistry. The chemistry makes it possible. A specific normalize and temper makes it real. Damage the microstructure and correct chemistry will not save you.
The required condition
Grade 91 is supplied normalized and tempered. Austenitize high enough to dissolve the carbonitrides, cool to form martensite, then temper high enough to toughen it while precipitating the fine particles that do the work.
The tempering temperature has to be high, and that is the source of the problem, because it puts the temper close to the temperature at which the steel starts turning back into austenite.
The overheating trap
Exceed the lower critical temperature during any heat treatment, including post weld heat treatment, and part of the material re-austenitizes. On cooling, that region forms fresh untempered martensite: hard, brittle and with none of the engineered creep strength.
The critical point: another temper does not fix it. Tempering untempered martensite gives you tempered martensite, but not the specific precipitate structure that came from the original normalize. Recovery requires a full re-normalize and temper, which on an installed component in a power station is usually not practical.
So the failure mode is a component that looks fine, has the right chemistry on its MTR, and has lost a large fraction of the creep strength the design depends on. It will not fail at hydrotest. It will fail years into service, and the investigation will find a PWHT chart with an excursion on it.
Why nickel and manganese are restricted
Both lower the lower critical temperature. Grade 91 specifications therefore restrict the sum of nickel plus manganese, and it is a real restriction rather than a formality.
The logic is straightforward. PWHT already runs close to the critical temperature. A heat running high on nickel and manganese pulls that temperature down, narrowing the gap between an effective temper and accidental re-austenitization. Push far enough and the window closes entirely, meaning there is no temperature that both tempers properly and stays safe.
Restricting the sum keeps a workable window. It is one of the few cases where a chemistry limit exists purely to protect a downstream process.
Hardness tells you what happened
Hardness is the practical check on Grade 91 condition, and it is informative in both directions.
- Too hard suggests under-tempered material, or fresh martensite from an overheated PWHT. Brittle, and not what the design assumed.
- Too soft suggests over-tempering. The material is tough enough but the creep strength has been degraded, which is arguably worse because nothing about handling it feels wrong.
On a creep service grade, soft is not safe. Both limits matter, and both should be on the specification.
Type IV cracking
The long term failure mode specific to this family. Under sustained creep loading, cracking develops in the fine grained region of the heat affected zone beside a weld, where the thermal cycle left a microstructure weaker in creep than either the parent metal or the weld metal.
It appears after long service rather than during fabrication, and it is the reason Grade 91 systems get so much attention paid to weld placement, weld count and inspection planning. Fewer welds in high stress locations is a design strategy, not just good practice.
What this means for ordering flanges
- Specify F91 with the governing specification and revision, not just "Grade 91".
- State the nickel plus manganese limit if your specification imposes one, and expect it to be certified.
- State the hardness range, both limits.
- Ask for the heat treatment chart in the documentation package, with actual austenitize and temper temperatures and hold times.
- Confirm your fabricator has a procedure qualified on F91, at the actual thickness. A procedure qualified on F22 does not cover it, and the PWHT window is the reason.
- Plan the PWHT before the joint is welded, including calibrated control, thermocouple placement and recording. With a window this narrow, discovering the plan afterwards is not an option.
And a note on Grade 92
Grade 92, the 9Cr-2W variant, pushes creep strength higher again and turns up on newer high efficiency plant. Everything said above about processing sensitivity applies at least as strongly. Availability in flange sizes is narrower than F91 and lead time reflects it. Send the requirement and the specification and we will tell you what is realistic.
See preheat and post weld heat treatment for the fabrication detail, and chrome-moly grades for the family reference.