(I)

CrossMark

Available online at www.sciencedirect.com

ScienceDirect

Procedía Engineering 130 (2015) 1441 - 1459

Procedía Engineering

www.elsevier.com/locate/procedia

14th International Conference on Pressure Vessel Technology

Serviceability Assessment for Safe Operation ofHydroprocessing Reactor by FFS (HPIS Z101-2 Level 2 Assessment)

A. Yasutomia'*? S. Sannomiyaa, Y. Shimakia

aMuroran Plant The Japan Steel Works, Ltd. 4 Chatsu-machi Muroran, Hokkaido, 051-8505, Japan

Abstract

The Japan Steel Works, Ltd. (JSW) developed the Fitness-For-Service (FFS) assessment procedure of 21/4Cr-lMo steel heavy wall hydroprocessing reactors derived by the material properties of hydrogen embrittlement cracking of old generation 21/4Cr-lMo steel (60's steel), where the combining effect of temper embrittlement and hydrogen embrittlement defined by threshold fracture toughness value for a hydrogen environment, Kih and material fracture toughness measured in the hydrogen charging environment, Kic-h , and the in-service crack growth represented by hydrogen assisted cracking rate model, da/dt and/or dc/dt were provided based on JSW studies. This procedure evaluates the present integrity given a current state of damage and the projected remaining life of reactor by the applicable FFS Codes.

©2015 The Authors.PublishedbyElsevierLtd. Thisis an open access article under the CC BY-NC-ND license

(http://creativecommons.Org/licenses/by-nc-nd/4.0/).

Peer-review under responsibility of the organizing committee of ICPVT-14

Keywords: FFS ofhydroprocessing reactor; Material degradation; Hydrogen embrittlement cracking; Failure Assessment Diagram; remaining life

1. Introduction

The hydroprocessing reactor which had been in service for 26 years since 1964 was replaced because of the detections of crack-like flaws in the In-Service-Inspection (ISI). The assessment results applying Z101-1 Assessment Procedure for Crack-Like Flaws in Pressure Equipment-Level 1 published in 2001 and revised in 2008 by the High Pressure Institute of Japan (HPIS) to them were not acceptable for the continued operation based on their material studies for FFS. The re-evaluation applying HPIS Z101-2-Level 2 published in 2011 to the not

* Corresponding author. Tel.: +81-143-22-0155; fax: +81-143-22-1439. E-mail address: akitada_yasutomi@jsw.co.jp

1877-7058 © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Peer-review under responsibility of the organizing committee of ICPVT-14

doi:10.1016/j.proeng.2015.12.313

acceptable crack-like flaw in above Level 1 and furthermore the JSW studies of FFS for Level 3 are introduced in this paper.

2. Equipment specification

The detail of reactor specification is as follows [1].

• Internal Diameter: 2,000 mm

• Wall-Thickness: 150 mm with Type 310/308 Weld Overlay (WOL) of 4 mm

• Material: Annealed 2.25Cr-lMo Steel Plates, SA387-Gr.22, Cl.l

• Specified Yield Strength: 205 MPa

• Specified Tensile Strength: 415 MPa

• PWHT: Performed

• Service Duration: 26year since 1964

• DesignPressure:10.9MPa

• Design Temperature: 420 0C

• Process Environment: Hydrogen Service

The detail of seismic (bending) moment calculation for reactor is assumed as follows in accordance with High Pressure Gas Safety Institute of Japan (KHK) [1].

• soil classification: Soil Class 2

• normalized response spectrum: 2.8

• damping factor of reactor: 0.03

• correction factor: 1.18

• pi= importance category factor: pi = 0.5

• P2 = seismic zone factor: p2 = 0.8

• P3 = site amplification factor: P3 = 2

• P4 = horizontal response magnification factor (Static Coefficient Method): P4 = 2.66

• P5= horizontal response magnification factor (Modified Static Coefficient Method): P5 = 3.304

• KH = horizontal seismic coefficient of design base earthquake at ground surface

• : KH = 0.15^kpip2p3 = 0.1200, where 1.0 for Level 1 Earthquake

• KSH = design static horizontal seismic coefficient: KSH = P4KH = 0.3192

• KMH = design modified static horizontal seismic coefficient: KMH = P5KH = 0.3965

The first natural period of reactor, T1 is obtained by the following Eq. from the reactor configuration [2].

The moments due to earthquake are derived by the following Eq. substituting the simplified evaluation equation based on static coefficient method, p= and the simplified evaluation equation based on modified static coefficient method, p= fis, respectively [2].

The equation modified by the initial condition is as follows substituting the modified simplified evaluation equation based on static coefficient method, p= for T1 < 0.9s andp= fis for T1 = 0.9s [2].

M .WT .x2 +1M,

o 7. 1 a '

— ■ WTh

2h T 5

The comparative result obtained from above Eqs. is shown in Fig. 1.

Fig. 1. Comparative seismic (bending) moment as per simplified equations [1].

3. Material data

In order to examine the degree of material degradation during 26 years service, several test block taken from the retired vessels were subject to chemical analysis and mechanical examinations. The result of chemical analysis and tensile test for the reactors are summarized in Table 1 and 2, respectively. J-factor of the retired vessels varied between 60-200 in contrast with current typical level of 40-80 in new vintage 21/4Cr-lMo steel. The impurity levels, the J-factor and X-bar for the weld deposits were significantly higher than those ofthe base plates [3].

Table 1. Typical result of check analysis for 21/4Cr-lMo reactor steels (wt.%) [3].

Base Plate Weld Deposit

Heat 1 Heat 2 Heat 3 Heat 1 Heat 2 Heat 3

c 0.16 0.13 0.13 0.06 0.06 0.08

Si 0.27 0.30 0.20 0.52 0.53 0.41

Mn 0.45 0.41 0.37 0.77 0.82 0.77

P 0.006 0.006 0.006 0.008 0.008 0.007

s 0.023 0.015 0.013 0.022 0.014 0.015

Ni 0.16 0.05 0.09 0.11 0.10 0.12

Cr 2.47 2.21 2.17 2.48 2.34 2.36

Cu 0.21 0.05 0.05 0.14 0.28 0.20

Mo 0.96 1.04 0.95 1.04 0.96 0.95

As 0.012 0.008 0.008 0.007 0.009 0.010

Sn 0.022 0.003 0.004 0.006 0.007 0.007

Sb 0.0030 0.0006 0.0009 0.0011 0.0019 0.0016

J-factor 202 64 57 181 203 166

X-bar 17.5 8.3 8.9 11.7 12.7 11.6

Table 2. Typical result of tensile test for 21/4Cr-lMo from retired reactor vessels [3].

0.2% Yield Strength (MPa) Tensile Strength (MPa) Elongation % Reduction of Area %

Specification of ASTM A387 Gr.D Mm. 207 414/586 Min. 20 Min. 50

Base Plate 193/312 421 /522 26.5 /39.0 52.7 / 73.0

Weld Deposit 288/488 481/614 21.2/35.2 50.4 / 67.3

Weld Joint 221 /328 443 / 518 21.0/24.8 64.5 / 68.6

In order to determine the degree of temper embrittlement associated with long term service, de-embrittlement heat treatment at 610°C for 3hours was conducted on these samples. Typical results of V-notch Charpy impact test before and after the de-embrittlement heat treatment are illustrated in Fig.2. Degraded impact toughness due to temper embrittlement were detected only on the weld deposits, while no significant temper embrittlement was observed in the base plates [3].

300 250 _ 200 150 ° 100 50 0

-SO -60 -40 -20 0 20 40 60 80 100 120 140 Test Temparature, °C

Fig. 2. Typical degree of temper embrittlement observed in base plate and weld deposit materials [3].

In order to study the possibility of hydrogen assisted crack growth in the reactors, several constant displacement tests were performed using compact 1T-CT type tension specimens for both of the base plate and the girth weld deposits. The 1T-CT specimens with side grooves were fatigue precracked, then subject to Ni plating to retard the diffusing-out of hydrogen during the short time constant displacement test. To simulate the condition of steady state operation, the CT specimens were exposed in an autoclave to high temperature and high pressure hydrogen for 48hours, rapidly cooled down to room temperature, and then subjected to the constant displacement test for 24hours. The stress intensity factor, KI0 of 62MPaVm was initially applied for each specimen, and then, load line displacement was held to be constant. During the test period of 24hours, load drop due to the crack extension was continuously monitored through load cell. After the test, specimen was artificially broken in liquid nitrogen and size ofthe extended crack from pre-crack front was measured using SEM [3].

Results obtained by the constant displacement test are summarized in Fig.3. For the base plate, no crack extensions could be detected in spite of the fact that hydrogen concentrations were up to 8ppm. The higher resistance to hydrogen assisted crack growth is due to the fact that the hardness value of the base plate is sufficient low. The experimental test result obtained in the base plate specimens can explain why all of the inner surface cracks in the Type 310/308 stainless steel weld overlay terminated at the interfaces between the weld overlay and the base plate. However, there was a sharp contrast for weld deposit specimens where crack growth was clearly

observed in the hydrogen concentration range of 3 to 8 ppm. The morphologies of crack surfaces observed in the weld deposit specimens using SEM were closely related to the morphologies of the actual extended cracks observed in the reactor girth weld seam. These experimental test results, simulating a hydrogen assisted crack growth, support the theory that the step-like cracks in the girth weldments were extended from the sharp tip of original hot tear crack through a mechanism ofhydrogen embrittlement [3].

-Base Metal (Hv : 150)

-Weld Metal (Hv : 185)

Constant Displacement Test

Initial Kj= 62MPaVm

Holding Time = 24hours

Hydrogen Charged Conditions

3ppm : 420°C. 9.81MPa o

6ppm : 450°C. 24.5MPa

Sppin : 500°C. 24.5MPa

123456789 10

Hydrogen Content, ppm

Fig. 3. Hydrogen assisted crack growth in hydrogen charged 1T-CT specimens taken from retired vessel [3].

4. Characterization of defect

In the reactors, there were two types of indications that were of concern. These were: PT indications at the weld overlay surface as shown in paragraph 4.1. and UT indication in the reactor girth seam as shown in paragraph 4.2. [3].

4.1. Weld Overlay Cracking [3]

Figure 4 illustrates the typical stainless weld overlay surface cracking with a maximum length of 12mm and a maximum depth of 6mm observed at cross section of tray support ring attachment area of the reactor [4]. All of the cracks were generated through the double-layer weld overlay along dendritic columnar grain boundaries and perfectly terminated at the interface between austenitic welds and 21/4Cr-lMo base plate. At the crack front in the base plate, small sized corrosion pits filled with corrosion products identified mainly as FeS were observed [3]. It was reported that results ofcheck analysis and metallurgical examination revealed no delta ferrite or sigma phases in the first layer and second layer weld overlays. Hence, it was concluded that the exact cause of the large number of weld overlay cracking is attributed mainly to hot cracking due to the fully austenitic structure during fabrication [5]. The summery of detection result of weld overlay cracking in the tray support ring attachment area cross section is as follows.

• Component geometry : Cracks in weld overlayed plate

• Crack geometry : General solution

• Detected crack size : Max.6mm depth and Max. 12mm length

• Crack loading : Mechanical load condition due to the difference of elastic modulus and thermal load condition due to the difference ofthermal expansion coefficient between the weld overlay and base material

• Cause of degradation : Corrosion (There were corrosion products at the tip of cracks due to process fluid.) [6]

• Cause of crack growth : None (Cracks perfectly terminated at the interface between austenitic welds and 2V4&-lMo base plate.)

4.2. Flaw in 21/Cr-1Mo Steel Girth Weldment [3]

Figure 5 illustrates the typical etched structure of a circumferential girth weld containing the flaw with a prolonged length of 90mm parallel to the weld direction and a maximum height of 21mm embedded from the internal surface at 32mm depth. This flaw was detected and sized by field ultrasonic examinations. A higher-magnification photomontage of the crack is shown in Fig.6 [3]. A typical feature of hot cracking during the welding process was observed at the defect's origin. The defect appears to initiate by grain boundary hot tearing with many MnS inclusions. These features clearly indicate that the origin of the defect is hot cracking during welding. However, totally different features of crack surfaces are observed in the step-like crack extension. The step-like crack in the weldments propagating through the columnar weld deposit indicates hydrogen assisted crack growth. It is believed that each step observed in the extended crack corresponds to a shutdown cycle that the reactor has undergone during 26years service. [3]. The summery ofdetection result offlaw in 21/4Cr-lMo steel girth weldment is as follows.

• Component geometry : Girth weld (Cylinder)

• Crack geometry : Embedded crack, circumferential direction

• Detected crack size : Max.21mm depth in thickness direction and Max.90mm length in girth weld direction located at 32mm depth from internal surface

• Crack loading : Internal pressure loading, Design seismic (bending) moment, and Weld residual stress

• Cause ofdegradation : Temper embrittlement and Hydrogen embrittlement

• Cause of crack growth : Hydrogen assisted crack growth

Fig. 4. Tray support ring attachment area cross section of weld overlay cracking [3].

Fig. 5. Typical etched structure of girth weldment containing a flaw [3].

Fig. 6. Photomontage of flaw observed in girth weldment [3].

5. Modeling of crack-like flaw

These crack-like flaws are modeled by converting to elliptical shape cracks shown as follows.

5.1. Weld Overlay Cracking [4]

• Surface crack in thickness of 150 mm, Semielliptical shape

• ae = Crack depth, ae = 6 mm

< Allowable flaw depth (= 6.8 mm) derived by Level 1 Assessment Screening Curve [4]

• le = Crack length, le = 2ce =12 mm

• ae/le = Aspect ratio, ae/le = 0.5

This weld overlay cladding is judged to be acceptable per Level 1 Screening Criteria and it is not considered that there is the potential for it to propagate to the base plate by hydrogen assisted crack growth in service, where that is judged by the criterion of KI (= 20.7 MPa Vm) < KIH (= 22 MPa). Then, it is possible to return to service in the condition of MPT > MAT (= 34°C) without the appropriate repair of it and/or remediate or replace of the component containing it [4].

5.2. Flaw in 21/4Cr-1Mo Steel Girth Weldment [4]

• Cylinder - Embedded crack, Circumferential direction, Elliptical shape

• 2ae = Crackdepth, 2ae = 21mm

< Allowable flaw depth (= 8.6mm) derived by Level 1 Assessment Screening Curve [4]

• le = Crack length, le = 2ce = 90mm

The flaw in 21/4Cr-lMo steel girth weldment is judged to be unacceptable per Level 1 Screening Criteria, then it is necessary to perform Level 2 assessment to make the decision to repair welding and/or rerating dependent on the present integrity of above 21/4Cr-lMo steel girth weldment containing the flaw given a current state of damage and the projected remaining life, where these assessment procedures are based on the Failure Assessment Diagram (FAD) method. If the results of Level 2 assessment indicate that this reactor is not suitable for the current operating conditions, the repair welding or replace is required, except that a reduced maximum allowable working pressure and/or coincident temperature to rerate the design conditions of it is found [1]. The Level 2 assessment performed for this flaw in 21/4Cr-lMo steel girth weldment is as follows containing the determination of stress distribution normal to crack face, reference stress and stress intensity factor, the estimate of material fracture toughness, and FAD analysis.

6. Determination of stress distribution

The design internal pressure loading and seismic (bending) moment loading conditions as primary stress and the residual stress for circumferential weld in cylindrical shell as secondary stress are considered, respectively, to determine the stress distribution normal to the crack face modeled by above procedure.

The meridional membrane stress, am and the global bending stress, agb due to bending moment,M in cylinder are provided as follows, respectively [1].

V - Rf

* R4q - Rf

where, p is internal pressure, Ro is outside radius of cylinder, Rj is inside radius of cylinder, and F is net-section axial force acting on cylinder.

To derive the stress intensity factor, the stress distribution normal to the crack face due to the design seismic moment loading is considered to be acting on the thickness direction represented by Eq. (6) as contrasted with the action of the constant membrane stress distribution due to the design internal pressure loading. It is transformed to the third order polynomial stress distribution by curve-fitting represented by Eq. (7). Then, the membrane, am and bending, ab stress distribution due to the design seismic moment derived by Eq. (7) are represented by Eq. (8) and Eq. (9), respectively.

a(x ) = M (Rj + x ) (6)

a( x ) = a0 + oi f J ) + a2 (J 1 + f y 1 (7)

where, x is radial local coordinate originating at the internal surface, 0 =x =.Ro-Ri , / is geometrical moment of inertia, I = n(Ro4-Ri4)/4 .

The residual stress distribution due to the circumferential welds in cylindrical shell is provided as shown in Fig. 7 [1]. It is shown that the vertical axis is magnitude perpendicular to welds and horizontal axis is extent of the residual stresses at the weldedjoint.

PWHT: Performed

Width of Weld Metal : W, = 59mm (Outside) W2 = 30mm (Inside)

Location of Crack

Magnitude of Residual Stress

: 0.2^ =0.2«+69) = 54.8 MPa

Outside: residual stress extent

(Non-PWHT) (Non-PWHT)

(Effect of PWHT) = 54.8 MPa

(Effect of PWHT) = 54.8 MPa

(Non-PWHT)

1.0(7™°" or 1.0o'"f'""

(Non-PWHT)

'IPj/„) ipgjdg ■ residual stress extent

Fig. 7. Residual stress distribution for circumferential welds in cylindrical shell (Stress distribution perpendicular to weld) [1].

7. Determination ofreference stress

The reference stress solution represented by the following eqs. for cylinder - embedded crack, circumferential direction elliptical shape, internal pressure with net-section axial force and bending moment is applied to the stress distributions derived by the design internal pressure loading and seismic (bending) moment loading and the residual stress distribution for circumferential welds in cylindrical shell in above calculations for the flaw in 21/4Cr-lMo steel girth weldment, respectively [1]. The results ofreference stress calculations are shown in Table 3.

aref ~ Am cam + Ab ca;

where, provided by the following parameters.

\{km + Q ¥m 2 ydl-jei

, +(km - 1Ü--

n 2-r V 2 n

16 jl - (1 -rf

\¡fm - arccos

8 (km + l)|l "(I-rf

-sin 6

8(km +l){l-(1 -4}cos¥b +[(2-2^e2 -yrf-(2-2re2 + yrf -8(km -l){l-(l-rf^sinO

(2 - 2re2 + yrf + 8(km - l)il -(l -rf}-(2 - 2re2 - ytf

for ¥m ^0-'-.

¥m =V—.

for ¥m <0-\

¥ = 2yg 1 -,e2 _km-1

Yb km + 1 2 -X km + 1 I 2

for Vb >0--

- n . n n

¥b =Q~ — for ¥b <0~-

d, d-, t 2a . nc el , e2 = , t = d[ + d2, r = —, y ~ —, u --

t' ' f 1 ^ Ro' t' 4 (R + d)

Table 3. Results ofreference stress calculations [1].

Crack Loading

Primary (Internal Pressure & Seismic Moment) 1.0046 1.0023 34.94 MPa 43.14 MPa 78.34 MPa

Secondary (Residual Stress) 54.80 MPa OMPa 55.05 MPa

8. Determination of stress intensity factor

The stress intensity factor solution represented by the following eqs. for plate - embedded crack, elliptical shape, through-wall membrane and bending stress is applied to the stress distributions derived by the design internal pressure loading and seismic (bending) moment loading and the residual stress distribution for circumferential welds in cylindrical shell in above calculations for the flaw in 21/4Cr-lMo steel girth weldment, respectively [1]. The results ofstress intensity factor calculations are shown in Table 4.

Ki =(Mm*me + Mb^be ^^

The local membrane and bending stress components acting on the crack face are given by the following eqs., respectively.

, + Ob I 1

&be =&b ^ YJ (13)

The membrane and bending correction factor is given by the following eqs., respectively.

Mm = H0Ufw (14)

/ \l-5 / \Z5

ll-l - 0.091891-

Mb =-|0. 5 + 0. 2591|-J - 0. 09189 |-J | Ufjpsiny (15)

The flaw shape parameter, Q and plastic zone correction factor, qy are defined by the following eqs., respectively. Q = 1.0 +1.4641 - qy for C < i.o (16)

Q = 1.0 +1.4641 - qy for C > i.0 (17)

q = 11 Mm°me + Mb°be

6 I (Jmean

where, provided by the following parameters.

H0 = — sin2 ip{H90 (1 + sin + H270 (1 - sin $>)} + H0 cos2 p

cos ç + sin Ç ^

for - < 1.0

sin2 <p + cos2 4" for L0< - < 2.0

fß = f270 + f90 _ f270f sin v

Table 4. Results of stress intensity factor calculations [1].

0 Mm M„ Crack Loading fme.MPa tffe.MPa K, . MPaVm

0° 0.4860 0 Internal Pressure 34.94 0 2.90

Seismic Moment 41.25 0.39 3.42

Primary - - 6.32

Secondary (Residual Stress) 54.80 0 4.55

90° 1.0059 -0.5385 Internal Pressure 34.94 0 6.00

Seismic Moment 41.25 0.39 7.05

Primary - - 13.05

Secondary (Residual Stress) 54.80 0 9.41

270° 1.0060 0.5389 Internal Pressure 34.94 0 6.00

Seismic Moment 41.25 0.39 7.12

Primary - - 13.12

Secondary (Residual Stress) 54.80 0 9.41

9. Estimate of material fracture toughness

As shown in Table 1, the J-factors, Jte derived by the chemical analysis of specimens taken from the weldments in retired reactor are JTE = 165~203. It is reported that the material fracture toughness measured in the non-hydrogen charging environment, KIC is predicted to be bout 60MPaVm and the material fracture toughness measured in the hydrogen charging environment, Kic-h is predicted to be 30MPaVm for the 21/4Cr-lMo steel with jte = 200 [7]. Then, the value ofthe material fracture toughness used in the assessment, Kmat is given as follows.

Kmat = KIC-H ~ lOMPajm (18)

where, J-factors, Jte and x-bar, are given by Eqs.(19) and (20), respectively [4].

JTE - (%Mn + %Si) (%P + %Sn)x 104 (19)

X =(10 • %P + 5 • %Sb + 4 • %Sn + As) x 102 (20)

10. Failure Assessment Diagram (FAD) analysis

The acceptability of crack-like flaw is determined by the failure assessment diagram (FAD) criteria defined as follows [1].

K = l jJL in (21)

This equation is derived by the crack opening displacement ofDugdale Model represented by Eq.(22) [1].

where, Eq. (21) is derived by substituting the stress intensity factor, K = a(na)1/2 , the material fracture toughness, Kic = Eoodc , the critical crack opening displacement, 8 = Sc , and the reference stress, a = Oyef, to Eq.(22).

The load ratio, Ir is defined by the reference stress derived for primary loading in above paragraph using the safety factor ofXs = 1.5 for the yield strength providing plastic collapse limit, as represented by eqs.(23) and (24) [1].

°0 = 0 Xs

where, a™ is minimum specified yield strength.

The toughness ratio, Kr is defined by the stress intensity factors derived for primary and secondary loading, respectively in above paragraph using the safety factor oiXk = 1.5 for the calculation of stress intensity factor due to primary stress taking versatility and conservative solution into consideration [1].

XkKr + Ks

where, Kmat = Kic-h , and ^ is plasticity interaction factor.

The assessment point calculated as per Eqs. (23) through (25) is defined as (Ir, Kr), and plotted on the FAD. The FAD assessment point derived for the flaw in 21/4Cr-lMo steel girth weldment is shown in Table 5 and Fig. 8. As shown in Fig. 8, the assessment results as per Level 2 shows that the FAD assessment point is on the unacceptable region, then it requires the repair and/or remediate, or the replace ofcomponent.

Table 5. Calculated FAD assessment point for the flaw in 21/4Cr-lMo steel girth weldment [1].

—111111 Vys r- mat K Kr

205 MPa 30 MPa\'m 1.5 1.5 0.0142 0.573 0.984

JNACCEPT/ hBLE REGI ON

ACCE PTABLE R EGION

FAD As deflnee >sessment as ( Lr. Kr 'oint )

0 0.2 0.4 0.6 0.8 1 1.2

Fig. 8. Assessment results using FAD for the flaw in 21/4Cr-lMo steel girth weldment [1].

11. Proposed Fitness-For-Service assessment procedure of21/4Cr-lMo steel heavy wall hydroprocessing reactors

Since heavy wall hydroprocessing reactors in petroleum refineries operate in high temperature, high pressure hydrogen service, hydrogen dissolves from the wall surface and at shutdown. Those absorbed hydrogen can cause subcritical crack extension where the substantial hydrogen concentration is retained for longer period even after shutdown due to the slow migration and diffusing out process of those in thick wall. Furthermore, it is assumed that the increased susceptibility to hydrogen assisted cracking is considered when the steel is temper embrittled due to the high temperature exposure after long-term service. Therefore, it is of great interest to know the loss of toughness and cracking behavior by internal reversible hydrogen embrittlement mechanism (I.H.E.) in order to define the remaining life, operating procedure and startup/shutdown management. The general flowchart for Fitness-ForService assessment of 21/4Cr-lMo steel heavy wall hydroprocessing reactors is given in Fig. 9 [8]. This evaluation consists ofthe following procedure;

1) Judge the probability of crack propagation from the prediction of threshold fracture toughness value for a hydrogen environment, KIH and the calculation of stress intensity factor, KI for the crack-like flaw detected by NDE.

2) Estimate the material fracture toughness measured in the hydrogen charging environment, Kic-h and determine the adequacy for the current operating conditions including the start up/shutdown procedure applying FAD.

3) Estimate the hydrogen assisted crack growth rate model and perform the in-service crack growth analysis for the representative load histogram for continued operation.

4) Continue to increment the crack size by the numerical integration of above crack growth rate model and perform FAD analysis to define the limiting crack size.

5) Determine the period to reach above limiting crack size as remaining life of the reactor for continued operation.

In this study, the effect of absorbed hydrogen in steel on toughness degradation is evaluated by hydrogen charging the large 3.5T-CT specimen in high pressure, high temperature autoclave to reproduce the hydrogen embrittlement of thick wall reactor at shutdown. The hydrogen distribution inside of the specimen was confirmed almost uniform with around 3ppm concentration just after charging the specimen and the remained hydrogen in the specimen was found around 2ppm even after 2-week exposure in ambient air. Using those hydrogen charged 3.5T-CT specimens, long term crack growth behavior in ambient air were examined on temer embrittled old generation 2.25Cr-lMo steel (60's steel) with some quantity of tramp elements and less temper embrittled recent generation one (recent steel) with less quantity of tramp elements. As result, when hydrogen is introduced in the 60's steel, the susceptibility to fast fracture significantly increases with higher crack growth rate while the non-hydrogen charged one showed ductile fracture with upper shelf fracture toughness characteristic. On the other hand, hydrogen charged recent steel did not exhibit fast fracture but showed sub-critical slow crack growth. Moreover, it is found that the crack growth rate has a strong dependence on loading rate on both steels, which can be described in the form of K and dK/dt formula. Also, high temperature tests were conducted and results indicated that hydrogen embrittlement phenomena disappeared with increased temperature (150 °C) even high hydrogen concentration (~ 3.9ppm) detected in the specimen after slow rising load tests [8].

Cracked component geometry and operating conditions

Initial crack size -

Stress Intensity factor K

■j Update stress

K„ = {20 + 80 exp[0.0164 (T -138)]} (MPaVm,°C)

Valid for

• Temperature T : 20-150°C

• H2 concentration : <3ppm

Estimate K,C.H from: • Temperature ■ FATT

^Yes Crack increment

Judgment of propagation of crack

Prediction of amount of flaw growth

rai.-i.iat» A a II f Crack incremenl from da/dt I Aa

Determination of limiting flaw size using FAD i°On the FAD curve

Remaining life calculated by Integration of crack growth rate

Calculate Remnant life

t= Ida/CK"

Re ra ting ?

I Repair or replace

— = CK" x J6206 x f-___—1}

dt [ I.T+ 273 293 J J

C = 0.0373 xKA0.6067 n = 1.269 + 0.33 logK

Fig. 9. Proposed Fitness-For-Service assessment flow of60's 21/4Cr-lMo steel heavy wall hydroprocessing reactors [8].

12. Material properties for proposed Fitness-For-Service assessment procedure of21/4Cr-lMo steel heavy wall hydroprocessing reactors

Material properties in this study are listed in Table 6 and Table 7. The two steels selected for testing are full bainitic quenched and tempered 2.25Cr-lMo steels; one representing today's steel making practice (defined as "recent steel" in the rest of this study) and the other representing 1960's steel making practice (defined as "60's steel" in the rest of this study). The recent steel was made by recent steel making practice with high purity, less temper embrittlement susceptibility. The 60's steel was produced to simulate 1960's steel making practice by intentionally adding impurity elements such as Si, P and Sn to the level similar to the 1960's steel making practice, which was subject to severe temper embrittlement. The heats were subjected to quenching tempering, followed by post weld heat treatment for 8 hours and subsequent step cooling heat treatment [8].

The tentative drawings of the temperature dependence of Km, Kic-h and Kic assessment of hydrogen embrittlement cracking of 60's steel is shown in Fig. 10 which considers the combining effect of temper embrittlement and hydrogen embrittlement. In region I there is a possibility of critical flaw growth by both hydrogen effect and low temperature effect. In region II, there is a possibility to fast fracture due to the combining effect of temper embrittlement and hydrogen embrittlement. In region III, there is a possibility of slow hydrogen flaw growth. In this region, crack growth analysis might be available using K versus da/dt curve and fracture resistance curve (R-curve) of hydrogen charged material. In region IV, there is a small possibility of hydrogen crack growth [9]. The recent steel did not exhibit any fast fracture phenomenon above room temperature (Kih =42MPa V m) and Kih increased at 86 °C. Based on the conservative data in Fig. 10, the equation for evaluating temperature dependence of Kih for 60's steel can be expressed as follows [8].

Kih = 20 + 80exp[0.0164(T -138.6)] (MP^m, oC) (26)

Figure 11 and Fig.12 shows the relationship between crack growth rate da/dt and applied stress intensity factor K. The crack growth rate in between Kih and Kic-h shows plateau characteristics in relationship between da/dt and K. The crack growth rate tend to increase with increasing loading rate for both steel. The trend equations for da/dt are constructed taking those strong dependences on loading rate dK/dt and K into consideration [8]. For recent steel,

= CKn = 8.550E -11K-L338 XK«is+^fogii r2?x

dt \rt yl'}

For 60's steel,

= CKn = 0. 03733K0.6067 XKL269+0.(28)

Where Kih<K<Kic-h , dK/dt=0.0005~0.005 MPaVm/s, where da/dt in mm/s , K in MPaVm , dK/dt in MPaVm/s .

Figure 13 shows crack growth curves of 60's steel obtained at elevated temperature, which was tested by slowest rising load, test condition (dK/dt = 0.0005 MPa V m/s). From this figure, it can be recognized that the crack growth rate decrease with increasing temperature and those at elevated temperature also showed plateau characteristics in relationship between da/dt and K. The average crack growth rates in plateau region are plotted against the reciprocal of test temperature in Fig. 14. Since it shows linear temperature dependence for 1/T, the crack growth rate at temperature T °C can be expressed by referring room temperature's da/dt in a following equation by conservative incrementof da/dt- 1/T plots,

Cc = £ X eXPI6206 X t^W-M H'- OC) (29)

assuming that the strain rate dependence of da/dt at temperature T °C is the same as the one at room temperature, the combined effect of strain rate and temperature can be unified in the above equation by substituting Eq. (28) to Eq. (29) and those analyzed value obtained were plotted altogether in Fig. 13 [8].

Table 6. Chemical composition of21/4Cr-lMo base metal (mass.%) [8].

Material C Si Mn P Ni Cr Cu Mo Al Sn J-factor

Recent Steel 0.14 0.08 0.55 0.005 0.21 2.42 0.06 1.08 0.019 0.010 95

60's Steel 0.14 0.25 0.56 0.014 0.16 2.43 0.07 1.05 0.025 0.022 292

Material

Table 7. Material properties of 21/4Cr-lMo base metal [8]. Tensile Properties at Room Temperature Impact Properties

0.2% Yield Strength (MPa)

Tensile Strength (MPa)

Elongation

Reduction

of Area

CVN_us

Fracture Toughness (MPaVm)

Recent Steel 60's Steel

475 504

614 648

28.8 28.5

82.4 80.1

335 233

532 490

700 600 500

S. 400

300 200 100

60's steel

■ A : KIC-H

KIH Threshold K at which measurable crack extension occurs

K|C-H K where crack growth rate shows rapid increase or fast fracture

Kic Fracture toughness without hydrogen charging condition

Critical

growth

-200 -100 0 100 200 300 Temperature / °C

Fig. 10. The tentative drawings of the temperature dependence of Kih, Kic-h , Kic assessment of hydrogen embrittlement cracking of60's steel [9]

Recent steel K>0

£ 1E-3 r t/j

3 1E4 ro ■o

1E-5 r

K=0.005 M Pa 7"m/s

K=0.0005

30 50 100 200

K(MPaVm)

¡6 13 -o

60's steel K>0

■ 0.01

0.005 MPa7"m/s

K=0.0005

- 1E-4

- 1E-5

30 50 100 200

K (MPaVm)

Fig. 11. Rising load rate effect on crack growth rate on recent steel [8]. Fig. 12. Rising load rate effect on crack growth rate on 60's steel [8].

----substitution

of eq.(28) to eq.(29)

(0.8ppm)

50 100 200

K CMPaV"m)

():H2content after lest ,ppm '1): 480°C, 20MPa hydrogen charged: Initial Hppm=6.0ppm

Fig. 13. Crack growth rates at elevated temperature [8].

T (°C)

20 50 86

0.01 r

K=0.005 MPaVm/s

0.0005 MPaV~m/s

1E-3 r

1 E-4 r

5.0 4.5 4.0 3.5 3.0 2.5 1000/T (K1)

Fig. 14. Temperature dependence of average crack growth rate at different rising load rate for 60's steel [8].

13. Prediction ofremaining life by hydrogen assisted cracking of21/4Cr-lMo steel heavy wall hydroprocessing reactors

The remaining life where the initial flaw sizes (a0, c0) found at the time of the inspection continue to increment their dimensions until the calculated FAD assessment point is on its envelope is established performing the following crack growth analysis by hydrogen assisted cracking. The analysis involves the numerical integration of the crack growth rate model represented by Eqs. (26) through (29).

a) STEP 1 - Compute the stress at the flaw based on the future operating conditions. In these calculations, all relevant operating conditions including normal operation, start-up, upset, and shutdown are considered.

b) STEP 2 - Determine an increment in crack growth based on the previous flaw size (to initialize the process, the previous flaw size is the initial flaw size), stress, estimated stress intensity, and the crack growth rate model. For surface and embedded flaws, the increment of crack growth will have a component in the depth and length dimension. For embedded flaws, the increment of crack growth may also include a component to model the flaw location in the wall thickness direction. The increment of crack growth is established based on the applied stress intensity associated with the component of the crack and the crack growth equation. For example, if a surface flaw is being evaluated, the crack depth is incremented based on the stress intensity factor at the deepest portion of the crack and the length is incremented based on the stress intensity factor at the surface. The flaw size to be used in the next step is the previous flaw size plus the increment of crack growth.

The crack growth equations are expressed as follows. The crack growth rate equations by substituting Eq. (27) for recent steel or equation (28) for 60's steel to Eq. (29) are shown as follows [8]. For the crack growth rate at the deepest portion,

f = C r* - I62«6 fe - ¿J} (30)

For the crack growth rate at the surface,

f = C T * - H - (31)

where the material parameters of crack growth rate model by hydrogen assisted cracking, C and n are shown as follows.

Forrecent steel [8],

C = 8.550E - 11K (32)

n = 3.616 + 0.&95logK (33)

For 60's steel [8],

C = 0.03733K a6067 (34)

n = 1.269 + 0.330 logK (35)

The crack depth-increment, A a , and half length-increment, Ac , for time increment, A t, are defined as follows, respectively;

Aa = (da/dt )-At (36)

Ac - (dc/dt)-At (37)

The current crack size is defined as follows;

a = a + Aa for the crack depth (38)

i = 2c + 2Ac for the crack surface (39)

where, da/dt and dc/dt in mm/s , K/="/2 and Ki*=0 in MPaVm, K in MPaVm/s , A t in s , and a, Aa, c and Ac in mm.

c) STEP 3 - Demonstrate that for the current crack size, the applied stress intensity factor is less than the critical stress intensity factor, KIC-H. If the assessment point for the current flaw size is outside of the FAD or the crack is recategorized as a through-wall crack, then go to the next step; otherwise, go to Step 2 and continue to grow the crack.

d) STEP 4 - Determine the time of for the current crack size to reach the critical flaw size. The component is acceptable for continued operation provided:

1) The time to reach the critical flaw size, including as appropriate in-service margin, is more than the required operating period.

2) The crack growth is monitored during shutdown by a validated technique.

3) The observed crack growth rate is below the value used in the remaining life prediction as determined by an inspections during shutdowns.

4) Upset conditions in loading or environmental severity are avoidable.

5) If the depth of the limiting flaw size is recategorized as a through-wall thickness crack, the conditions for an acceptable leak-before-break (LBB) criterion should be satisfied, e) STEP 5 - At the next inspection, establish the actual crack growth rate, and re-evaluate the new flaw conditions per procedures of this paragraph. Alternatively, repair or replace the component or apply effective mitigation measures.

14. Summary

This paper outlined the FFS by HPIS Z101-1-Level 1 and Z101-2-Level 2 assessments and the studies of material degradation of the hydroprocessing reactor which had been in service for 26 years since 1964 and replaced because of the detections of crack-like flaws in the ISI. Their results were not acceptable for the continued operation due to the flaw associated with hydrogen assisted crack growth in 21/4Cr-lMo steel girth weldment raising the material degradation characters by temper embrittlement and hydrogen embrittlement.

Based on these analyses, The JSW developed the FFS assessment procedure of 21/4Cr-lMo steel heavy wall hydroprocessing reactors derived by the following material properties of hydrogen embrittlement cracking studied on60's steel.

• Threshold fracture toughness for hydrogen environment, Kih applied to the judgement of crack propagation

• Fracture toughness measured in the hydrogen charging environment, Kic-h applied to the determination of limiting flaw size using FAD

• In-service crack growth represented by hydrogen assisted cracking rate model, da/dt and/or dc/dt applied to the prediction of amount of flaw growth and remaining life of reactor

This procedure recommends not only whether the repair of damage and/or remediate, or the replace of 2V4Cr-lMo steel heavy wall hydroprocessing reactors containing crack-like flaws is necessary but also the appropriate inspection interval in conjunction with the applicable in-service inspection code providing information for the inservice monitoring plan.

As the estimate of material degradation studied on old generation steels was shelved in the FFS Codes such as HPIS Z101-2 and API 579-1/ASME FFS-1, this procedure has been proposed for Level 3 assessment.

References

[1] High Pressure Institute of Japan (HPIS) Z101-2 Assessment Procedure for Crack-Like Flaw in Pressure Equipment-Level 2.

[2] S.Konosu and N.Mukaimachi, "The Simplified Evaluation Method of the Seismic Moment for Towers and Tanks", JHPI Vol.45 No.l 2007.

[3] A.J.Bagdasarian, E.L.Bereczy, T.Ishiguro, K.Kimura, T.Tahara, "Material Degradation and Hydrogen Assisted Crack Growth of Old Generation Hydrocracking Reactor Vessel Made of21/4Cr-lMo Steel".

[4] High Pressure Institute of Japan (HPIS) Z 101-1 Assessment Procedure for Crack-Like Flaw in Pressure Equipment-Level 1.

[5] A.J. Bagdasarian, E.L.Bereczy, T.Ishiguro, T.Ishizuka, T.Tahara, "Investigation of 26years Hydroprocessing Reactors - a Summary Report", Corrosion 93 NACE Annual Conference and Corrosion Show PaperNo.629.

[6] T.Tahara, L.P.Antalffy, R.Kayano, T.Kikuchi, "Chronological Review of Manufacturing Technologies and Considerations to Maintenance / Inspection for Heavy Wall Hydroprocessing Reactors".

[7] JPVRC, "Embrittlement of Pressure Vessel Steels in High Temperature, High Pressure Hydrogen Environment", Welding Resarch Council, Bullutin 305, p.9, 1985.

[8] Y.Wada, R.Kayano, T.Hasegawa, H.Inoue, "Hydrogen Embrittlement Testing of Aging Pressure Vessel Steels Using Large Thickness Specimen", Proceeding ICPVT-10; pp.543-552; (2003).

[9] Y.Wada, "Hydrogen Embrittlement of Low Alloy Steel for Pressure Vessels Based on a Large Size Specimen", Dept. of Nanomechanics, Graduate School ofEngineering, TOHOKU UNIV.; Doctor ofPhilosophy; (2007).