ML18270A091

From kanterella
Jump to navigation Jump to search
09/24/2018 Overview of Draft NUREG-2228
ML18270A091
Person / Time
Issue date: 09/24/2018
From: Benson M L
Office of Nuclear Regulatory Research
To:
Michael Benson
Shared Package
ML18270A088 List:
References
NUREG-2228
Download: ML18270A091 (29)


Text

Overview of Draft NUREG

-2228Michael L. BensonOffice of Nuclear Regulatory ResearchU.S. Nuclear Regulatory CommissionPublic Meeting to Discuss Draft NUREG

-2228 on Weld Residual Stress Validation and NRC Research Program on Extended Finite Element AnalysisRockville, MDSeptember 24, 2018 Public Release of Draft NUREG

-2228*This meeting allows open discussion of the issues addressed in the NUREG, making the public comment period more productive

  • Introduces a validation method for finite element prediction of weld residual stress

)*Public comment period formally announced in the Federal Register on September 13, 2018

-

Reference:

83 FR 46524 (ML18239A171) day public comment period

-Submit comments by November 13, 2018

  • NRC posed four questions for the public

-Hardening law recommendation

-Choice of benchmark

-Choice of metrics

-Feasibility of implementationPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 2

Outline of Draft NUREG

-2228 1.Introduction

-Brief background on WRS research program

-Report scope 2.Phase 2b Round Robin Study

-Mockup fabrication

-Round robin modeling study

-Overview of measurement techniques

-Overview of measurement and modeling results 3.Uncertainty Quantification Methodology

-Overview of the Sandia method

-Outliers-Results: bootstrapped confidence bounds and tolerance boundsPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 3

Outline of NUREG

-2228 4.WRS Impact on Flaw Growth Calculations

-Superposition of operating stress

-Identification of important features of the stress profile (e.g., 0

-stress points)

-Relation of flaw growth behavior to stress profile features 5.Validation Procedure and Finite Element Guidelines

-Model-measurement comparisons: difference in means and root mean square error functions

-Judgements about hardening law

-Proposed guidelines: based upon commonly

-accepted practice

-Proposed validation scheme: benchmark, metrics, and acceptance measuresPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 4

Chapter 2: Phase 2b Round Robin StudyPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 5Overview Public Meeting to Discuss Draft NUREG

-2228 and xFEM 6Deep Hole DrillingContourVector, Ltd.Hill Engineering, LLCChapter 2: Phase 2b Round Robin StudyMeasurements Public Meeting to Discuss Draft NUREG

-2228 and xFEM 7Axial Residual StressHoop Residual StressChapter 2: Phase 2b Round Robin StudyMeasurement Results Public Meeting to Discuss Draft NUREG

-2228 and xFEM 8Isotropic HardeningNonlinear Kinematic HardeningChapter 2: Phase 2b Round Robin StudyModeling Results Chapter 3: Uncertainty Quantification Methodology

  • Uncertainty quantification scheme was developed at Sandia National Laboratory
  • The prime information source is a publicly

-available Sandia report (ML16301A055

)*Draft NUREG

-2228 only summarizes the details of the method

  • Sandia implemented the method via the R programming languagePublic Meeting to Discuss Draft NUREG

-2228 and xFEM 9Sandia Report Chapter 3: Uncertainty Quantification Methodology

  • Recognize WRS as functional data, -i.e., the stress magnitude depends upon the location

-Apply established mathematical methods for functional data

-See references in Sandia Report and draft NUREG

  • Outlier screening

-Three round robin submissions screened out of the uncertainty quantification process

-Justification: modeling guidance was not followed

-One hoop stress profile that passed initial screening will be discussed further (slide 14)

  • Data smoothing

-Cubic spline routine (see Sandia report)

-Data smoothing for WRS may be worth consideration by the industry

-More sophisticated than a least

-squares polynomial fitPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 10Analysis Concepts=Axial Weld Residual Stress [MPa]

Chapter 3: Uncertainty Quantification Methodology

  • Amplitude and phase variability

-Both forms of variability are present in the Phase 2b dataset-Need methods to quantify these forms of uncertainty

  • Registration

-Aligns the stress profiles

-Quantifies phase variability

  • Functional Principal Components Analysis

-Quantifies amplitude variability

  • Bootstrap sampling from the statistical model

-Amplitude and phase variability in the samples reflects that of the original dataset

-Enables determination of confidence bounds and tolerance bounds

  • No assumptions on distribution typePublic Meeting to Discuss Draft NUREG

-2228 and xFEM 11Analysis Concepts Chapter 3: Uncertainty Quantification Methodology

  • Bootstrapping applied to entire prediction dataset, the hole drilling measurement dataset, and the contour measurement hoop stress data
  • Path data was extracted from the contour measurement axial stress data

-Extracted 500 stress profiles

-Confidence bounds and tolerance bounds calculated from the path data, not bootstrappingPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 12Analysis Concepts Chapter 3: Uncertainty Quantification MethodologyPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 13Example ResultsConfidence Bounds on the Mean 95-95 Tolerance Bounds Chapter 3: Uncertainty Quantification MethodologyPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 14Outlier Hoop Stress Prediction?

Chapter 4: WRS Impact on Flaw Growth CalculationsPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 15Flaw Growth Study Chapter 4: WRS Impact on Flaw Growth CalculationsPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 16Evaluation Period

  • Uncertainty decreases under shortened evaluation period (20 years vs. 60 years)

Chapter 4: WRS Impact on Flaw Growth CalculationsPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 17Area under the Curve

  • Simplified method to understand flaw growth sensitivity to residual stress through

-wall trends

  • Identified salient features of the residual stress curve that may have impacts on flaw growth

-ID stress magnitude

-Stress magnitude at initial flaw depth

-Location through the wall thickness where compressive zone begins

-Slope of the curve in the compressive zone Chapter 5: Validation Procedure and Finite Element GuidelinesPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 18Measurement

-Model Comparisons

  • Difference in means and root mean square error functions
  • Sample a number of measurement stress profiles
  • Sample a number of prediction stress profiles
  • Calculate the means of the two samples
  • Calculate the difference in the two means
  • Calculate square root sum of squares of the difference, averaged over the wall thickness Chapter 5: Validation Procedure and Finite Element GuidelinesPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 19Measurement

-Model Comparisons

  • Difference in means and confidence bounds on the difference
  • "Good agreement" implies that tolerance bounds encompass zero throughout entire wall thickness
  • Areas where confidence bounds do not encompass zero indicate disagreementNonlinear kinematic hardening predictionsHole drilling measurement Chapter 5: Validation Procedure and Finite Element GuidelinesPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 20Measurement

-Model Comparisons

  • Best comparison

-Average hardening predictions

-Contour measurementAverage hardening predictionsContour measurement Chapter 5: Validation Procedure and Finite Element GuidelinesPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 21Measurement

-Model Comparisons Chapter 5: Validation Procedure and Finite Element GuidelinesPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 22Hardening Law Recommendations

  • Uncertainties are large
  • Nonlinear kinematic appears to be the least accurate overall
  • The averaging approach is the recommended hardening law Chapter 5: Validation Procedure and Finite Element GuidelinesPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 23Modeling Guidelines

  • Weld bead geometry definition

-Modeling precise bead shape is unnecessary

-Weld beads may be approximated as trapezoids

-Total number of beads and layers and beads per layer should accurately reflect the real configuration

  • Bead and process sequence

-Analyst should model actual fabrication sequence-Analyst should explicitly model the application of the butter and postweldheat treatment

  • Heat input model tuning

-Analyst should confirm that weld bead and surrounding material reaches melting temperature

  • Structural and thermal boundary conditions

-Boundary conditions should represent the physical situation and can differ from application

-to-application

-For a typical dissimilar metal butt weld, axial displacement in one node away from the weld should be constrained

-Modeled pipe length should be at least 4 times the inner diameter to avoid edge effects

-Heat convection at the surface can be neglected

  • Material properties

-Average of isotropic and nonlinear kinematic is the recommended approach

-Temperature

-dependent material properties provided in NUREG for reference

  • Element selection and mesh

-Quadrilateral, linear elements

-Avoid triangular elements

-Fine mesh in weld Chapter 5: Validation Procedure and Finite Element GuidelinesPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 24Overview of Validation Approach

  • Three ingredients required for a validation approach

-Benchmark-Set of metrics

-Acceptance measures

  • Section 5.4 of the draft NUREG develops and justifies NRC's recommendations
  • Development of metrics and acceptance measures is tied to flaw growth argument*Analyst creates a finite element model of the Phase 2b mockup, calculates the metrics, and compares the results to the acceptance measures Chapter 5: Validation Procedure and Finite Element GuidelinesPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 25Benchmark*Four choices of benchmark

-DHD measurement data

-Contour measurement data

-Average of DHD and contour results

-Mean of the Phase 2b models (assuming a given hardening approach)

  • Benchmark based upon measurement data

-Perhaps the natural choice

-Complicating factors: not a pure measurement, independent techniques may not agree, differences are significant from a flaw growth perspective

-Averaging all the measurements may not be rigorously correct

-Not possible to say which measurement is more correct

  • NRC recommended mean of the average hardening predictions

-A qualitative sense that the measurements and models agree

-The "mean" is derived from the uncertainty quantification scheme Chapter 5: Validation Procedure and Finite Element GuidelinesPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 26Proposed MetricsDerivative estimated by slope of secant line

-average sense of agreement

-related to slopes and concavity

-related to importance of the inner surface k k-1k+1 h 2 Chapter 5: Validation Procedure and Finite Element GuidelinesPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 27Proposed Acceptance MeasuresAxial Stress Acceptance MeasuresHoop Stress Acceptance Measures

  • See P. Raynaud's presentation for additional background on metrics and acceptance measures Chapter 5: Validation Procedure and Finite Element GuidelinesPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 28Scope and Applicability

  • Representativeness

-Excavate and weld repair given as an example

-The general principles in the draft NUREG could be applied to developing a validation approach for applications that are not well represented by the Phase 2b mockup

  • Welding process

-High heat inputs and/or large weld beads may not be represented by Phase 2b mockup

  • Hardening law

-Hardening law recommendations should be more

-or-less universal

-Additional study of the mixed hardening law may be worthwhile ClosingPublic Meeting to Discuss Draft NUREG

-2228 and xFEM 29*Summarized the contents of the draft NUREG

-2228-Phase 2b round robin

-Uncertainty quantification method

-WRS impacts on flaw growth calculations

-Proposed guidelines and validation scheme

  • P. Raynaud will discuss the metrics and acceptance measures in detail
  • NRC has initiated the formal public comment process on this NUREG
  • This meeting: seeking initial feedback and discussion on the NUREG, in order to facilitate the public comment process