You are on page 1of 2

Q&A

Advanced Creep Assessments with Case Studies

Nathaniel Sutton Seetha Kummari, Ph.D., P.E. Phillip Prueter, P.E.


Staff Engineer II Senior Engineer II Team Leader – Materials & Corrosion
Principal Engineer II
E NSutton@E2G.com E SKummari@E2G.com E PPrueter@E2G.com

Q1 Part 10 assessments are very sensitive to small changes and small variations in the creep ductility and creep strain rate in the Project Omega Model.
Do you have any guidance on how to vary these variables? Is it only linked to the amount of graphitization?

A1 In the absence of Omega/creep test data for the component, API 579 provides some guidance (e.g., a Level 2 assessment of a welded component
uses a ∆ΩSR of -0.5). In this assessment, we have tied the factors to the degree of graphitization as well as creep. With test data, ∆ΩSR and ∆ΩCD are
found from the difference between the predicted and measured values of Omega or strain rate at the test conditions.

Q2 Do you normally use the Omega method or LMP to evaluate creep damage? API 579 calls out both. Omega model is being used to simulte teritary
creep stage. Please clarify why Omega model for FFS is being used.

A2 Each model has advantages in certain cases. We frequently use the MPC Omega method as opposed to Larson-Miller, but we have used both in
many assessments. We generally consider the Omega multiaxial damage parameter, coupled with the effective stress expressions in the Omega
method, to be a more appropriate characterization for complex geometries with multiaxial stress states, compared to the Larson-Miller approach.

Q3 How is it determined to use Omega instead of LMP?

A3 Creep tests performed on ex-service material may or may not rupture, but almost always (if the test is properly executed) an Omega and Initial
Strain Rate parameter can be measured. So, available data from the test may encourage the use of the Omega method as opposed to LMP.
Additionally, 3-D FEA modelling often necessitates consideration of multiaxiality effects, which we typically address through the Omega method
multiaxial damage parameter.

Q4 How were the test plates provided to perform the Omega test? Where were they cut from?

A4 These plates were extracted from various locations throughout the vessel. One came from a tee-junction (intersection of a longitudinal and
circumferential seam); one came from a location where a stiffening ring had been attached, and the third came from a cooler, section in the vessel
where an internal reinforcing pad had been attached.

Q5 This damage mechanism is a function of stress and temperature. So component geometry and weld geometry play a factor in creep assesment
and damage potential?

A5 This is certainly correct. Complex component geometries can be accounted for through 3-D finite element simulations; if necessary, weld
geometries can be considered. In our assessment, weldment creep properties were utilized, as measured from the test samples, but the geometry
of the welds was not considered to play a role (welds observed were smooth, with no transitions, gentle contours, and no major concern for stress
concentration effects).

Q6 What is the remaining life of a vessel like this when creep is found? Would remaining life be component-based only or can the analysis assume the
estimated remaining life as a whole?

A6 In this assessment and in many similar assessments, we usually assess creep remaining life on a component-by component basis. For the case of 3D
FEA creep simulation, we can identify the life limiting location(s). If the owner-user inspects and repairs or replaces the most life limiting components,
the limiting remaining creep life of the vessel as a whole can be extended accordingly.

Q7 How can one account for different materials for shell and welds in a Level 3 FEA creep assessment, as in, for example, 0.5Cr-0.5Mo shell vs C-0.5Mo
welds?

A7 Although this was not performed for the case study presented, we have, in other projects, assigned unique material properties to different regions in
the FEA geometry. This is done, for example, to examine the effects of strain rate and creep strength mis-match between potentially dissimilar weld
and base materials. In advanced cases, we have performed material testing to determine HAZ-specific material properties.

Q8 Would you please provide more detail about the K-omega model creep subroutine?

A8 It is a custom Abaqus subroutine that we developed in-house to implement API 579 Part 10 Omega creep methodology for all points in FEA
accounting stress relaxation.

Q9 Have you consider thermal gradient through the vessel thickness?

A9 Yes. Thermal gradients through thickness were solved in the finite element heat transfer simulations.
Q&A

Advanced Creep Assessments with Case Studies

Q10 Where do you sample your stress at shell-to-nozzle junction when using shell elements?

A10 You can certainly sample the stress right at the junction. However, that would be conservative. Please note that we have not extracted stress
results in this analysis. Our final result was creep damage that considers stresses at the shell junctions as they are. However, with creep, these
high stresses relax quickly. We always prefer 3D solid models if we want to extract the stress results to compare with the Code allowable stresses.
Alternatively, you may consider moving away certain distance (we have used half of the shell thickness for some projects) from the junction to
remove conservatism with the shell FEA models. However, that distance needs to be established by comparing the shell and solid models.

Q11 How long is the solving time for a model like this?

A11 Simulation run time depends upon number of parameters such as mesh size, creep properties, and the creep damage expected at the end of the
time duration simulated. For the current project, the simulation times were anywhere from few hours to as much as one full day on our clusters with
20 CPUs.

Q12 How was steady state thermal used here? Was flow within vessel modelled as well? How was temperature applied here?

A12 We did not simulate fluid flow in this project. Rather, we estimated heat transfer coefficients using closed form hand calculations and based on the
flow information provided. The temperatures from the heat transfer simulations were mapped onto the creep simulations.

Q13 How did you apply heat source term for your heat transfer simulation?

A13 By estimating heat transfer coefficients using closed form hand calculations and based on the flow information provided.

Q14 In terms of location of highest creep damage, how useful is it to only perform simple static stress analysis and identify the location with highest stress
level instead of doing a detailed creep analysis?

A14 Please note that creep damage is not only function of stress but also a function of temperature. Additionally, stress relaxation plays a significant role
in how the creep damage accumulates over the time. In fact, the highest stress location in our current project did not come out to the most limiting
for creep damage based on the additional variables aforementioned. So, while the highest stress location gives an indication of possible limiting
location for creep damage, performing the detailed creep analysis such as performed herein is only the way to confirm this.

Q15 Should bending stress be considered during the calculation of the damage factor at an area of high stresses? Is there an assumption when bending
can be neglected?

A15 Considering bending stresses is important near structural discontinuities such as nozzle, stiffening ring, and skirt junctions. Bending stresses were
considered in the current project though finite element stress simulations. For the regions away from structural discontinuities, bending stresses
may be neglected.

Q16 What is the difference between creep and thermal damage?

A16 In general we consider thermal damage to be independent of stress. For example, in this assessment, graphitization progression is independent of
stress, while creep damage depends on stress.

Q17 Did you adjust the material properties, such as delta_sr and delta_cd, during the simulation to track graphitization? Did the simulation for the future
condition account for additional graphitization?

A17 Iteration was performed on the as-measured ∆ΩSR and ∆ΩCD properties to remove the effects of creep damage, but not graphitization. The
properties applied at the time of installation assume the same degree of graphitization as measured from the 2020 test specimens. From a historical
perspective, this is conservative. Since current and future operating temperatures are significantly lower than historical ones, it was thought that the
rate of future graphitization would be sufficiently slow and, therefore, justified to ignore. While we did not consider for the current job, it is certainly
possible to specify ∆ΩSR and ∆ΩCD as a function of time.

ANSWERS FOR TODAY. INSIGHTS FOR TOMORROW.


Sales@E2G.com // 216.283.9519
TM
© 2021 Copyright E²G | The Equity Engineering Group, Inc. | Trademark E²G | The Equity Engineering Group, Inc.

You might also like