19
- Copyright - Risk-Informed Assessment of PWSCC Issue in CANDU Feeder Piping Xinjian Duan, Min Wang, Candu Energy Inc. Ming Li, Ontario Power Generation 17 th International Conference on Environmental Degradation of Materials in Nuclear Power Systems-Water Reactor, August 9-13, 2015, Ottawa, Ontario, Canada.

Risk-Informed Assessment of PWSCC Issue in CANDU Feeder Pipingenvdeg2015.org/final-proceedings/ENVDEG/presentations/... · 2015. 10. 19. · PWSCC observed in PWR/BWR primary heat

  • Upload
    others

  • View
    4

  • Download
    0

Embed Size (px)

Citation preview

  • ›- Copyright -

    ›Risk-Informed Assessment of PWSCC Issue in CANDU Feeder Piping

    ›Xinjian Duan, Min Wang, Candu Energy Inc. ›Ming Li, Ontario Power Generation

    17th International Conference on Environmental Degradation of Materials in Nuclear Power Systems-Water

    Reactor, August 9-13, 2015, Ottawa, Ontario, Canada.

  • Background

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 2

    PWSCC observed in PWR/BWR primary heat transport piping

    Dissimilar Metal Weld (DMW)

    CANDU feeder DMW: SA-106 Grade B to Alloy 600 with Alloy

    82/182 filler material

    Based on the EDY method, DMWs in some CANDU outlet

    feeders are approaching high risk category to be susceptible to

    PWSCC

    Cracking inspection on DMWs is challenging due to high dose

    and access limitations. As an alternative, demonstration of LBB

    mode of failure is required.

    oFor susceptible PWSCC, inspection is not required if LBB is demonstrated

    oFor active PWSCC, inspection (scope and frequency) is required

  • Objective

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 3

    Demonstrate LBB through a risk-informed methodology

    (composite of the three methods)

    oDeterministic LBB

    oAFEA/XFEM flaw evaluation

    oProbabilistic assessment

  • Methodology

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 4

    ›Deterministic LBB FFSG

    SRP 3.6.3

    ›FLAW Evaluation AFEA/XFEM

    ›Probabilistic Assessment PRAISE-CANDU 1.0

    Probabilistic

    Assessment AFEA/XFEM

    Flaw Evaluation

    Deterministic

    LBB

    • Factor on Crack Length • Factor on Load • Factor on Leak Detection • Factor on time from leakage to rupture • Consequential leakage

    • Realistic margin

    on time from

    leakage to

    rupture

    • Likelihood of

    failure

    • Effect of inspection

  • Deterministic LBB

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 5

    ›Key Inputs Geometry and material properties

    Loads

    oLoad Case 1 operational leakage calculation

    oLoad Cases 2/3 for the stability evaluation

    Leak detection limit: 25 kg/h based on operating procedure and OPEX

    Leak rate model (SQUIRT), leak rate factor of 5 as per the FFSG requirement to take into account the uncertainties in leak rate calculation

    Crack growth model (PWSCC and fatigue)

    ›Results LBB margins generally decreases with decreasing piping size

    40/44 feeders satisfy the LBB requirements

    4 feeder (two 2.5" + two 3") with reduced margins (3.7 – 4.8) on leak rate

  • Advanced FEA (AFEA) or Extended FEM (XFEM)

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 6

    ›Purpose Deterministic calculation

    To grow the crack of a postulated inner surface crack

    oConsidering the effect of Weld Residual Stresses (WRS)

    oCalculating the natural or non-constrained crack shape

    ›Methodology AFEA: same as MRP-216

    XFEM: new development

    ›Preliminary AFEA Results Sufficient time to credit operator action to the leakage

    oTime from incipient leakage to orderly shutdown leakage (5×25 kg/h): 160 days

    oTime from orderly shutdown leakage to immediate shutdown leakage (1800 kg/h): 58 days

  • Weld Residual Stresses-Overview

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 7

    ›Most uncertainty mechanical parameter

    ›Large experimental program to quantify this parameter

    Sample Fabrication

    oLiburdi Automation

    Measurements

    oNeutron Diffraction: Canadian Neutron Beam Centre and Rutherford

    Appleton Lab

    oX-ray Diffraction: Proto Manufacturing and Open University

    oContour Method: Open University

    FE Modeling

    oCandu Energy Inc (Abaqus) and Carleton University (VrWeld)

  • Probabilistic Assessment

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 8

    ›Purpose

    Provide additional technical basis for exempting inspection

    Provide supplemental information for deterministic calculation (deterministic

    LBB+AFEA/XFEM)

    oQuantify low probability of rupture

    ›PFM Code – PRAISE-CANDU 1.0

    Full compliance with CSA N286.7

    Joint effort between Structural Integrity Associates and Candu Energy Inc.

    State-of-the-art deterministic models and uncertainty treatment

    ›Advantages

    Uncertainties: aleatory, epistemic, and combined aleatory and epistemic

    Quantify the effect of crack initiation, inspection, and leak detection

  • PRAISE-CANDU 1.0

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 9

    ›Two-loop Architecture

    Same as xLPR 1.0/2.0

    Separation of uncertainties

    o Aleatory uncertainty

    o Epistemic uncertainty

    o Aleatory + Epistemic

    Important parameter/uncertainty study

    ›PC based, Monte Carlo only

    o Computationally efficient

    o Use of large number of CPUs

    ›Three Levels of Validations

    o Benchmarking with other generally accepted and documented PFM codes

    o Comparing PFM calculations with the rupture events of non-nuclear piping

    o Phenomenon based validation

    START

    END

    Epistemic

    Loop Done?

    Yes

    No

    Sample Epistemic

    Random Variables

    Aleatory Loop

    Done?

    Yes

    No

    Sample Aleatory

    Random Variables

    Deterministic

    Models

    Time Loop

  • PRAISE-CANDU Benchmarking on PWSCC

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 10

    1E-08

    1E-07

    1E-06

    1E-05

    1E-04

    1E-03

    1E-02

    1E-01

    1E+00

    0 5 10 15 20 25

    P(L

    OC

    A)

    Time (years)

    PRAISE-CANDU

    WinPRAISE

    0

    0.2

    0.4

    0.6

    0.8

    1

    0 10 20 30 40 50 60P

    rob

    ab

    ilit

    y o

    f R

    up

    ture

    Time (years)

    95 percentile: PRAISE-CANDU 1.0

    95 percentile: xLPR 1.0

    Mean: PRAISE-CANDU 1.0

    Mean: xLPR 1.0

    ›Excellent Agreement with

    WinPRAISE

    xLPR 1.0

    PRAISE-CANDU vs WinPRAISE PRAISE-CANDU vs xLPR 1.0

  • PWSCC Initiation Model 1

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 11

    ›Amzallag et al (1999)

    i: temperature index, exp(-Qi/RT)

    i: stress index, pn

    im: material index

    A: model coefficient, including heat-to-heat and within-heat uncertainties

    QI: activation energy, calibrated to be 193 kJ/mol

    pn: applied stress

    𝑡𝐼1

    𝑖𝑖𝑖𝑚

    𝑡𝐼 =1

    𝐴𝑖𝑚𝑝𝑛e𝑄𝐼 R𝑇 for 𝑝 > 𝑡ℎ

  • Results

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 12

    The rupture probability is low, < 1.410-8

    HAZ (Alloy 600) has higher rupture probability than weld center

    (Alloy 82)

    1E-09

    1E-08

    1E-07

    1E-06

    1E-05

    1E-04

    1E-03

    1E-02

    1E-01

    1E+00

    0 5 10 15 20 25

    Pro

    ba

    bil

    ity

    Time (a)

    P(Init) P(TW) P(LOCA)

    1E-09

    1E-08

    1E-07

    1E-06

    1E-05

    1E-04

    1E-03

    1E-02

    1E-01

    1E+00

    0 5 10 15 20 25

    Pro

    ba

    bil

    ity

    Time (a)

    P(Init) P(TW) P(LOCA)

    Weld Center (Alloy 82) HAZ (Alloy 600)

  • Effect of Activation Energy

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 13

    Activation Energy, QI = 0, i.e., multipel pre-existing initial cracks

    The rupture probability is low, < 1.010-4

    1E-09

    1E-08

    1E-07

    1E-06

    1E-05

    1E-04

    1E-03

    1E-02

    1E-01

    1E+00

    0 5 10 15 20 25

    P(L

    OC

    A)

    Time (a)

    QI = 193 kJ/mol QI = 0

  • Effect of Leak Detection Limit

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 14

    25 kg/h vs. 50 kg/h, 125 kg/h, no leak detection

    Rupture probability increases

    1E-09

    1E-08

    1E-07

    1E-06

    1E-05

    1E-04

    1E-03

    1E-02

    1E-01

    1E+00

    0 5 10 15 20 25

    P(L

    OC

    A)

    Time (years)

    No leak detection

    OLRL = 125 kg/h

    OLRL = 50 kg/h

    OLRL = 25kg/h

    Power Law Model

  • Effect of In-Service Inspection

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 15

    No inspection vs. inspection every 3 years with good POD

    Rupture probability slightly decreases

    1E-09

    1E-08

    1E-07

    1E-06

    1E-05

    1E-04

    1E-03

    1E-02

    0 5 10 15 20 25

    P(L

    OC

    A)

    Time (years)

    No Inspection Good POD

  • Parameter Ranking

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 16

    Activation energy (activation energy 193 vs 0 KJ/mol ), leak

    detection capability (25 vs 125 kg/h) and welding residual stress

    are top three factors .

    0

    4

    8

    12

    16

    20Im

    po

    rta

    nce

    Factors

  • Uncertainty Separation

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 17

    Forty billion (40×109) Monte Carlo simulations

    Uncertainty of mean residual stress being epistemic uncertainty

    Rupture probability: 5.9510-8 at 95th percentile vs. 1.4010-8 at

    mean value

  • Conclusions

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 18

    A tiered composite RI-informed LBB assessment has been

    developed and applied to PWSCC issue on feeder DMW. The

    results have been conditional accepted by the regulator.

    Deterministic LBB assessment demonstrates 90% DMWs satisfy

    the LBB requirements.

    AFEA demonstrates sufficient margins on time from leak to

    rupture for the remaining 10% DMWs.

    Probabilistic assessment demonstrates the probability of rupture

    is low.

    The sensitivity study shows the PWSCC initiation model, leak

    detection capability, and WRS are important, while the in-service

    inspection has little impact on the rupture probability.

    The uncertainty can be separated through two-loop PFM code,

    PRAISE-CANDU 1.0 for prioritizing the resources.

  • Acknowledgement

    - Copyright - © [2015] SNC-Lavalin Inc. and its member companies. All rights reserved. Unauthorized use or reproduction is prohibited. 19

    CANDU Owners Group and Ontario Power Generation for the

    funding the work.