1

February 2008

M. Dadfarnia, M. Martin, P. Sofronis, I. M. Robertson, D. D. JohnsonUniversity of Illinois at Urbana-Champaign

In collaboration withB. Somerday

Sandia National Laboratories

Materials Innovations in an Emerging Hydrogen EconomyAmerican Ceramic SocietyFlorida, February 26, 2008

Fracture Toughness Assessment of Hydrogen Pipelines

2

February 2008

Hydrogen-Induced Crack Propagation

a b c

d e f

0 s 17s 21s

29s 32 s 39s

Static crack in vacuum. Hydrogen gas introduced

thinning thinning

maincrack

maincrack

Micro-crackformed

Crack linkage

No load increase is needed for the crack to grow

We do not understand the relationship between macroscopic parameters(e.g. applied load and pressure) and the operating microscopic

degradation mechanism

3

February 2008

Hydrogen Embrittlement Mechanisms

Several candidate mechanisms have evolved over the years each of which is supported by a set of experimental observations and strong personal views

Viable mechanisms of embrittlementStress induced hydride formation and cleavage

Metals with stable hydrides (Group Vb metals, Ti, Mg, Zr and their alloys)Supported by experimental observations

Hydrogen enhanced localized plasticity (HELP)Increased dislocation mobility, failure by plastic deformation mechanismsSupported by experimental observations

Hydrogen induced decohesionDirect evidence is lackingSupported by First Principles Calculations (DFT)

Degradation is often due to the synergistic action of mechanisms

4

February 2008

Embrittlement and Phenomenology

Fractographic evidence suggests that low strength steels under static loading fail by

Hydrogen-assisted transgranular fracture induced by void or microcrack initiation through decohesion at internal interface (precipitate/inclusion or phase boundaries) ahead of a crack or notch accompanied by shear localization (HELP) leading to the linking of the void/microcrack with the tip of the crackFracture is controlled by yield strength level and microstructure

Our contention, which needs to be verified through experiment, is that embrittlement

Under static load is a result of the synergistic action of the HELP and decohesion mechanismsUnder cyclic load can be intergranular (extremely dangerous mode of failure)

5

February 2008

Fracture Mechanics Approach to Design of Steel Pipelines Transporting Hydrogen

H2 gas

Hydrogen diffusion

H2 gas

Objective: Determine stress,deformation, and hydrogen concentration fields in the neighborhood of an axial crack in a steel pipeline

H2-Pressure of 15MPa

To characterize embrittlement we need to understandthe interaction of hydrogen with the elastoplasticdeformation of the material at a crack tip

6

February 2008

Fracture Mechanics Approach to Design of PipelinesActual-Pipeline Solution vs Laboratory-Specimen Solution

Is there a similarity between the full-field

(pipeline) solution and that at laboratory

specimens?

H2 gas

Subcritical crack growth experiments with

WOL specimen carried out at Sandia

If yes, we conjecture that parameters which characterize fracture in the laboratory specimencan be used to characterize fracture in the pipeline

Tranferrability

2IKr

σπ

=r

Crack tip σ

If characterizes fracture in the specimen,can it be used to characterize fracture inthe pipeline in the presence of hydrogen?

IK

7

February 2008

Diffusing hydrogen resides atNormal Interstitial Lattice Sites (NILS)Trapping Sites

Microstructural heterogeneities such as dislocations, grain boundaries, inclusions, voids, interfaces, impurity atom clusters

Diffusing hydrogen interacts with stresses and strainsHydrogen dilates the lattice and thus interacts with hydrostatic stress

Moves from regions under compression toward regions under tension, e.gahead of a crack tip

Hydrogen enhances dislocation mobility, thus it facilitates plastic flow

As hydrogen diffuses stresses and strains change. At the same time local stresses and strains affect the diffusion paths. So the problem is coupled

Hydrogen Transport Analysis

dislocations inclusions

Grain boundaries

TC

LC

Crack tip0σ >

8

February 2008

Cracked Pipeline: Problem Statement

( ) 0LC t =

( )LC t K f= ×

0 100-200 200-100

( ) 0LC t =

( )LC t f∝

( ) 0J t =

( ) 0LC t =

( )LC t f∝

( ) 0J t =

Hydrogen gas at pressure P

15 MPaHydrogen

gas

Hydrogen transport

15 MPa

time 1 sec

2.0 hrs

P

t

dimensions are in mm

outer diameter:thickness:

crack depth:initial CTOD:

40.64 cmh = 9.52 mma = 1.9 mmb0= 1.5 μma / h = 0.2

K : SolubilityJ : Hydrogen fluxP : Pressuref : Fugacity

9

February 2008

Materials CharacterizationMicrostructural characterization: Optical, SEM, and TEM studies

Existing pipeline steel samples provided by Air Liquide and Air Products.New micro-alloyed steels (new microstructures) provided by Oregon Steel Mills through DGS Metallurgical Solutions, Inc.

Establish the diffusion characteristics of existing and new pipeline steel microstructures

Determine uniaxial tension macroscopic flow characteristics in the presence of hydrogen

Carry out fracture testing: Collaboration with Sandia, LivermoreFracture surfaces, particle, dislocation, and grain boundary characterization

API/C Mn Si Cu Ni V Nb Cr Ti

GradeA X70 0.08 1.53 0.28 0.01 0.00 0.050 0.061 0.01 0.014B X70/80 0.05 1.52 0.12 0.23 0.14 0.001 0.092 0.25 0.012C X70/80 0.04 1.61 0.14 0.22 0.12 0.000 0.096 0.42 0.015D X52/60 0.03 1.14 0.18 0.24 0.14 0.001 0.084 0.16 0.014

Typical natural gas pipeline steel

Ferrite/acicular ferriteFerrite/acicular ferrite

Ferrite/low level of pearlite

10

February 2008

Optical Analysis of New “Steel C” Microstructure

Average grain size :35 μm

3% pearlite

Demonstrated to be good in the presence of H2S sour service natural gasapplications

API Grade C Mn Si Cu Ni V Nb Cr Ti

X70/80 0.04 1.61 0.14 0.22 0.12 0.000 0.096 0.42 0.015

Ferrite/acicular ferrite

11

February 2008

SEM analysis of New “Steel C” Microstructure

Al rich particle, most likely a sulfide

12

February 2008

TEM analysis of New “Steel C” Microstructure

a) EDS spectrum from particleb) Bright field TEM image of typical rectangular particlec) EDS spectrum from matrix

EDS analysis of fine precipitate inside ferrite grain suggests that precipitate is composed of Ti and Nb

(window detector: C, N, O not detected)

13

February 2008

TEM analysis of Air Liquide Steel Microstructure

Large intergranular particles (cementite)

Small intragranular particles (carbides with Nb and Ti)

14

February 2008

TEM analysis of Air Products Steel Microstructure

Pearlite colonies.

Left: cementite plate arrangement

Right: cross-section of platelets

15

February 2008

0.00E+00

2.00E+14

4.00E+14

6.00E+14

8.00E+14

1.00E+15

1.20E+15

0 200 400 600 800 1000 1200

time (s)

flux

(ato

ms/

m^2

s)

0.00E+00

2.00E+14

4.00E+14

6.00E+14

8.00E+14

1.00E+15

1.20E+15

0 5 10 15 20 25 30 35 40

time (s)

flux

(ato

ms/

m^2

s)

0.0E+00

5.0E+17

1.0E+18

1.5E+18

2.0E+18

0 20 40 60 80 100 120 140

time (s)

Inte

gral

of f

lux

(ato

ms/

m^2

)

0.00E+00

5.00E+10

1.00E+11

1.50E+11

2.00E+11

2.50E+11

3.00E+11

3.50E+11

4.00E+11

4.50E+11

0 10 20 30 40 50 60 70

Sqrt Pressure Pa^1/2

Stea

dy S

tate

flux

* th

ickn

ess

(ato

ms/

m s

)

•Oregon Steel Mills sample: thickness •room temperature

Ultrahigh vacuum (10-9 torr)Hydrogen pressure (10 torr)

Hydrogen Permeation Measurements

2

6Teff

LtD

=6.8sTt =

Steady state:

12 H atoms6.26 10MPa.m.s

Φ = ×

4.7torr 627PaP = =

J

J L∞

120micronsL =

Jdt∫

Time lag

Permeabilityat room temperature

J∞

0.0E+00

5.0E+16

1.0E+17

1.5E+17

2.0E+17

2.5E+17

3.0E+17

0 5 10 15 20 25 30

time (s)

Inte

gral

of f

lux

(ato

ms/

m^2

)

16

February 2008

Material: X70/80 acicular ferrite microstructure

Dislocation trapping modeling

0 . 0 0 . 2 0 . 4 0 . 6 0 . 8 1 . 0 1 . 2 1 . 4 1 . 6

2 0

2 1

2 2

2 3

2 4

E x p e r i m e n t a l r e s u l t s [ 8 1 ]

Lo

g(N T)

ε p

pε 0 0 .2 0 .4 0 .6 0 .8 1 .0 1 .2 1 .4 1 .6

24 23 22 21 20

( )TLog N 60.0 KJ/molBW =2

TNaρ

=

0 0.150 15

const. 0.15.

p p

p

γρ ε ερ

ε

⎧ + ≤⎪= ⎨⎪ >⎩

10 2 16 20 10 10,m mρ γ− −= =

20.2 KJ/molBW =

Kumnick and Johnsontrapping model

Stress-strain

Plastic Strain

Stre

ss(M

Pa)

0 0.1 0.2 0.30

200

400

600

800

00

1np

Yεσ σε

⎛ ⎞= +⎜ ⎟

⎝ ⎠

0 595 MPaσ =

0.059n =

pε

, pσ ε

Experiment Model

8 21.271 10 m /sD −= ×Lattice diffusion coefficient

22 30 2.65932 10 H atom / mC = ×

183

H atoms6.54696 10m Pa

K = ×

21 30 2.084 10 H atom / mC = ×

15 MPa P =

1 atm P =

C K f= exp P df PRT

⎛ ⎞= ⎜ ⎟⎝ ⎠

3= 15.84 cm /mold

17

February 2008

Lattice Hydrogen Concentration at Steady State

-202 -200 -198 -196 -1940

2

4

6

8

2.52.42.32.22.121.91.81.71.61.51.41.31.21.110.90.80.70.60.50.40.30.20.1

0

LCC

195 200 2050

5

10

15

20

25

30

10.90.80.70.60.50.40.30.20.1

0

LCC

Kumnick and Johnson trapping model

22 30 2.65932 10 H atom / mC = × 15 MPa P =

Time to steady-state: 2.0 hrs68secsst =

18

February 2008

Evolution of Hydrogen Concentration at NILS

1.17 μmb =

0 2 4 6 8 10 12 14 16 18 200

0.5

1

1.5

2

2.5

1.0sec

0.25sec

0.5sec0.75sec

25sec10sec

0.1sec

3sec

R / b

68sec

at steady state (2hrs) R

b

Rb

0

LCC

68 secsst

19

February 2008

Trapped Hydrogen Concentration at Steady State

-202 -200 -198 -196 -1940

2

4

6

8

6.56.2565.755.55.2554.754.54.2543.753.53.2532.752.52.2521.751.51.2510.750.50.25

0

TCC

195 200 2050

5

10

15

20

25

30

0.10.090.080.070.060.050.040.030.020.01

0

TCC

Kumnick and Johnson trapping model

22 30 2.65932 10 H atom / mC = × 15 MPa P =

20

February 2008

Fracture Mechanics ParametersFrom the Full Pipeline to the Laboratory Specimen

0 0.1 0.2 0.3 0.4

0

20

40

60

80

-0.35

-0.33

-0.31

-0.29

-0.27

-0.25

KI

T/σ0

IK0

Tσ

ah

Crack depth/pipe thickness

h

a H2 pressure15 MPa

Tr

/ 2IK rσ π=

21

February 2008

Full Field (pipeline) vs Boundary Layer Solution (laboratory specimen)

0 2 4 6 8 101.5

2

2.5

3 Rb

y

Modified Boundary Layer solution

Elastoplastic full-field solution at 15 MPaP =0

14.38 MPa m/ -0.292IK

T σ=

=

Modified Boundary Layer solution0

14.38 MPa m/ 0IK

T σ=

=

0T =0

yyσσ

/ 0.05a h =R

b 1.17 μmb =

Neglecting the T -stress in the MBL formulationfails to predict the true stress

22

February 2008

Crack-Tip Fields Scale with KI and T-stressIndependence from Crack Depth

R / b0 2 4 6 8 101

1.5

2

2.5

3

0

0.5

1

1.5

2

2.5

CL / C0

σkk / 3σ0

Crack depth a = 0.476 mm

Crack depth a = 1.9 mm

b03

kkσσ

0

LCC

h

a

T

0

34.12 MPa m/ 0.316IK

T σ== −

/ 0.2a h =

0

14.38MPa m/ 0.292IK

T σ== −

/ 0.05a h =

Rb

2IK Rσ π=

1.17 μmb =

7.13 μmb =

ConstraintFracture Mechanics

23

February 2008

0 1 2 3 4 50

0.5

1

1.5

2

2.5

ξ = 1, No softening

ζ

ξ = 0.955, Softening

Tr = σkk / σYTr

h

a

no hydrogen

ς

a / h = 0.05

0

H

H

ςς

ς=

=

b R

with hydrogen

Hydrogen Accelerates Void Growth

Rb

( )0

0

( ) ln exp /p

p pvv kk Y

R a dR

ες ε σ σ ε

⎛ ⎞= =⎜ ⎟

⎝ ⎠∫

Void growth parameter

triaxialitykk

Y

Tr σσ

= =

24

February 2008

WOL Specimen for Subcritical Crack GrowthFinite Element Mesh

W

a

2mV

0V

Crack tip

H

1.090 2.240 2.745H W B′′ ′′ ′′= = =

B

Applied displacement

: Crack mouth opening displacementmV

25

February 2008

WOL Specimen (X-100) loaded to KI=158 MPa√m

Pl. strain0.10.090.080.070.060.050.040.030.020.010

Plastic zone

W

a

2H

/ 0.5608a W =

2.180 2.240H W′′ ′′= =

2mV

1.204mmmV =0V

Plasticity is confined to the crack tip under K-dominance

26

February 2008

Crack Arrest in WOL Specimen : KI - dominance

FEM/ 0.9408a W =1.204mmmV =

63.8 MPa mIK =

57.5 MPa mIK =ASTM

FEM (Plastic)

16008 N/mJ =

62.2 MPa mIK =

21IJ EKν

=−

KI dominance when crack stops

27

February 2008

Long Term Objective: Multiscale Fracture Approach

3u

33Σ maxσ

Γ Dissipated energy

(c) Traction - separation law(b) Axisymmetricunit cell model(a) Crack tip

fracture process zone

TriaxialityHydrogen concentration

(e) Cohesive elements characterized bya traction-separation law based on the unit cell model

1 11,u Σ

3 33,u Σ

at time=0initialLc

(d) Cohesive element

Adjacent finite element

, LT c

Δa/D0

J/(σ

0D0)

0 2 4 6 8 100

2

4

6

8

10

12

With hydrogen softening in(1) cohesive zone and matrix(2) cohesive zone only

No hydrogen

28

February 2008

Conclusions and Future WorkAttempted to characterize the hydrogen concentration and stress fields in a pipeline in terms of KI and T-stress (J-T fracture locus - constraint fracture mechanics)

Model depends on assumptions (e.g. trapping according to Kumnick and Johnson model, reversible traps, etc) that need to be explored through microstructural characterization and permeation measurementsSelf similarity and no explicit dependence on crack depthTransferability of results from laboratory specimensIf void growth is the mechanism of failure, hydrogen enhances void growth through softening-induced straining

Developed cohesive element technology to simulate decohesion- or ductile-driven processes for crack propagation

Simulated J-R curve

29

February 2008

Coupling fracture mechanisms and microstructuralanalysis with hydrogen transport, thermodynamics of decohesion, and plastic flow localization to understand

Interaction of time scales (loading rate, diffusion rate, adsorption rateCrack initiationCrack propagationDevise fracture criteria with predicting capabilities

Possibly a JIC-T locus

Fracture mechanics/mechanism-based approach to design

As opposed to the SMYS approach

Conclusions and Future Work

30

February 2008

We have years of experience and extensive knowledge of all aspects of hydrogen embrittlement. We have a tremendous collection of analysis tools. We can tame the problem

Where We Go From Here

Support by theU.S. Department of Energy is

Gratefully Acknowledged