View From The Penthouse

The DNFM Quarterly Newsletter

DAVID N. FRENCH METALLURGISTS
Ph 502.955.9847 Fax 502.957.5441 www.davidnfrench.com
Dr. David N. French Dr. David C. Crowe
Creep Fatigue
Creep Fatigue at a Lug Attachment adjacent carbon steel tubes because stainless steel has a higher coeffi- volume of oxide ‘jacks’ the attachment away from the surface. Oxide can
    cient of thermal expansion than carbon steel. Stresses between these be seen in the photo, between the tube and attachment, and this could
A creep failure was found at an attachment in a reheater with carbon and stainless tubes would be transferred by the ties and attach- have influenced the failure.
type 304 stainless steel tubes which had been cycled to an operating ments. In the example, the attachment was nickel-base Alloy 600, a
temperature of 1000 oF some 800 times over a period of 16 years. The good choice to resist high temperature oxidation. Its coefficient of ther-
mal expansion is closer to that of carbon steel than to stainless steel, so Creep and Remaining Life
picture below shows the cross-section at the weld to the Inconel 600
attachment. The weld composition matched as-deposited analyses for stresses will be induced at the weld when the attachment is heated. The The amount of creep fatigue damage is dependent on the
Inconel 182. There was cracking at the toe of the attachment weld, and Alloy 600 weld is stronger than the tube material, so strains at the junc- frequency of the stress and thermal cycling, magnitude of the transient
it had propagated through the tube causing the failure. The cracking had tion with the stainless steel are shared unequally, with the stainless steel loss and temperatures, and the properties of the material. In attempting
initiated at the end of the attachment where stresses would be highest. yielding before the Alloy 600. The cracking initiated at the end of the to evaluate the combined effects of creep and fatigue on component life,
There was also cracking of the weld outward from the unwelded area attachment where the weld was not carefully wrapped around the attach- researchers have developed rules for adding damages of these separate
beneath the attachment. ment, resulting in a notch at the toe of the weld, presenting a stress processes together, for use in design. At high cycle frequency, fatigue is
riser. most important, but at lower cycle frequency and withhold times when
Drivers stressed, creep becomes more important. The condition of the material
can also have an impact: initially some creep softens the material,
Failures of tubes at attachments can involve a number of thereby improving the fatigue life. But if creep has progressed to the
factors. Temperatures are high in service, especially at the attachments point where cavitation has occurred, then the voids in the material will
because, compared to the tube wall, the attachments are further from reduce the fatigue life.
the heat sink (the steam or water inside the tube). In the example, tem- The damage summation method is the simplest method for
peratures of the attachment may have been elevated because there was estimating creep fatigue life. In this method, Robinson creep damage is
a lack of penetration at the base of the attachment, so it would not be added to Miner fatigue damage values. In some cases, this approach is
cooled as well as it would have if it had been in full contact with the tube. too conservative, but in some cases it may overestimate the remaining
Additionally, as temperatures increase, the strength of the tube or at- life. Other more complex approaches provide increasingly accurate
tachment material decreases. evaluations. Practically speaking, any of these methods are difficult to
The cracking of the tube adjacent to the weld is illustrated in Stresses can be high depending on the restraint required, the
the view below. This intergranular cracking is characteristic of creep. apply after the fact, where the thermal and stress history of a component
design of the attachment and its condition. A worn or poorly made at- is based on corporate memory and someone’s best guess. Metal-
Separated grain boundaries can be seen in addition to the main cracks. tachment can have stress concentration points at the weld termination or
The cracking was not confined to the heat affected zone of the weld. lographic examination of representative components provides a very
at other imperfections. This was a factor in initiating cracking in the good way to check the condition of tubes to look for creep fatigue dam-
When heated, this stainless steel tube would have expanded more than example lug attachment. During start-up, stresses can be especially high age. Non-destructive testing may be used to detect cracking after initia-
as components heat up, not necessarily all at the same rate, producing a tion.
transient thermal gradient. And when a unit experiences many cycles in Ways to Minimize this Problem
its life, these stresses present a low cycle fatigue on the attachment and
Good attachment design is vital to minimize stresses. Slip
the tube.
type attachments may be used to accommodate differential thermal
Another effect involves the thermal gradient in a tube when
expansion. Terminations of attachments should taper to the surface to
heated. The surface will heat faster than the core of the component,
reduce the localized stress concentrations at the termination.
putting the surface into compression. When the core reaches operating
Lack of penetration in attachment welds can result in them
temperature, the surface will then be in tension, and this could add to
running hot, or increase stress concentrations. Good weld design and
other operating stresses to drive fatigue or creep.
adherence to welding procedures are essential.
Carbon steel and stainless steel have significantly different
Weld metal should be a match to the tube, or weaker so that
coefficients of thermal expansion, so stainless steel will expand (and
the failure occurs in the attachment rather than in the tube itself.
contract) at a greater rate for a given temperature change, producing a
Regular visual inspection followed up with spot checks using
tug-of-war with any carbon steel component to which it is tied.
magnetic particle or penetrant testing is recommended in order to find
Oxide formation may be a factor where welds are incomplete,
these problems.
allowing the environment to enter and form oxide underneath an attach- Transient stresses can be reduced by starting up more
ment. The volume of oxide is greater than the metal it replaces, and that slowly.

Topic Ideas ? Please email [email protected]