Professional Documents
Culture Documents
PVP2020-21106 INCEFA-PLUS Project
PVP2020-21106 INCEFA-PLUS Project
tte
Pressure Vessels & Piping Conference
nd
ee
PVP2020
July 20-24, 2020, Virtual, Online
R
ea
d-
O
nl
PVP2020-21106
y
C
op
y
INCEFA-PLUS PROJECT: LESSONS LEARNED FROM THE PROJECT DATA AND IMPACT
ON EXISTING FATIGUE ASSESSMENT PROCEDURES
Matthias Bruchhausen
European Commission, Joint Research Centre
Petten, The Netherlands
1
Contact author: sam.cuvilliez@edf.fr
where Xi denotes the ith RV associated with the ith factor amongst FIGURE 2: Illustration of the cumulative probability
b, c and d, i and i respectively denote the mean and the density functions associated with MVDS and A’ random
standard deviation of these normal distributions, and t95 ≈ 1.645 variable
denotes the 95th percentile of the standard normal distribution.
Thanks to the lognormal hypothesis, i and i can be determined The Fen-threshold is then defined as follows, so as to
in equation (3) using the range of values given in TABLE 1 for quantify the mismatch between experimental results obtained in
each parameter. The lower and upper bounds of the ith range of a PWR environment on a rough specimen and a life prediction
values are assumed to be respectively the 5th percentile (Xi,5) and that uses a Fen factor and the design air curve translation factors:
95th (Xi,95) percentile of the associated lognormal distribution.
𝐹𝑒𝑛−𝐶𝐶792 𝑁𝑡𝑒𝑠𝑡 1
The next step consists in the statistical combination itself, which 𝐹𝑒𝑛−𝑡ℎ𝑟𝑒𝑠ℎ𝑜𝑙𝑑 ≝ (7)
𝑇𝐶𝑙𝑖𝑓𝑒 𝑁𝑑𝑒𝑠𝑖𝑔𝑛
reflects the successive application of each reduction factor to the
life predicted with the mean air curve in the ln(N)-space:
where Fen-CC792 is the environmental correction factor used in
𝐴′ ≝ 𝐴 − ∑𝑖=1,𝑛 𝑙𝑛(𝑋𝑖 ) (4) ASME code-case N-792-1 and recalled in equations (8) to (11),
Ntest is the experimental life of the specimen (N25 for instance),
where the RV A’ characterizes the position of the new left-shifted Ndesign is the life obtained for a strain amplitude equal to the one
fatigue curve. The global translation factor on life TClife-global is applied during the test using the code design curve, and TClife is
then obtained for a given percentile of A’ (5% in practice). In the translation coefficient on life (statistical combination of a
NUREG/CR-6909, Monte-Carlo simulations were carried out by given set of sub-factors amongst a, b, and d, depending on the
randomly drawing values for A and Xi according to their assumed surface condition). In equation (7) TClife is calculated the same
distributions in order to determine TClife-global. However, since A’ way as TClife-global using equations (2) to (6), but the difference is
is defined as a linear combination of normal independent RVs, that the sub-factor pertaining to surface finish has been excluded
from the statistical combination in equation (4), since its effect
A’ is also normally distributed with a mean ’ and a standard
is already accounted for through Ntest (the test being conducted
deviation ’:
on a rough specimen). Consequently, the red part of equation (7)
is intended to make the water data comparable to the design
𝐴′~𝒩(𝜇′, 𝜎′2 ), with:
curve prediction in air (blue part).
𝜇′ = 𝐴𝑚 − ∑𝑖=1,𝑛 𝜇𝑖 (5)
𝜎′2 = 𝑠𝑑2 + ∑𝑖=1,𝑛 𝜎𝑖2 𝐹𝑒𝑛−𝐶𝐶792 = 𝑒𝑥𝑝(0.734 − 𝑇 ∗ 𝜀̇ ∗ 𝑂 ∗ ) (8)
TClife-global can then be calculated with the following closed-form 𝑇∗ = 0 (𝑇 < 150°C)
expression, also illustrated on FIGURE 2: 𝑇 ∗ = (𝑇 − 150)⁄175 (150 ≤ 𝑇 < 325°C) (9)
𝑇∗ = 1 (𝑇 ≥ 325°C)
𝑇𝐶𝑙𝑖𝑓𝑒−𝑔𝑙𝑜𝑏𝑎𝑙 = 𝑒𝑥𝑝(𝐴𝑚 − 𝐴′5 ), with:
(6) 𝜀̇ ∗ = 0 (𝜀̇ > 0.4%.s-1)
𝐴′5 = 𝜇′ − 𝑡95 𝜎′
𝜀̇ ∗ = 𝑙𝑛(𝜀̇⁄0.4) (0.0004%.s-1≤𝜀̇≤ 0.4%.s-1) (10)
𝜀̇ ∗ = 𝑙𝑛(0.0004⁄0.4) (𝜀̇ < 0.0004%.s-1)
where A’5 denotes the 5 percentile of the normal distribution
th
However, as the sub-factors on life were revised in Rev.1, and IP 0.3 300 0.01 5.07 3
since this is completely uncorrelated with the Fen expression IP 0.32 300 0.01 5.07 1
used, it could be useful to take advantage of these revised sub-
IP 0.6 300 0.01 5.07 4
factors since the corresponding TClife values given in TABLE 3
are lower than the ones reported in [14] (3.7 and 3.0 respectively
Surface Nb. of
a (%) Fen-CC792 Fen-threshold (1) Fen-threshold (2)
condition tests
2.91 or 3.35
Ground 15 0.3 2.50 2.91
or 5.07
Ground 1 0.313 5.07 2.87 3.35