risk-based optimization of inspection for piping and pressure … · 2010. 1. 12. · risk-based...

12
Risk-Based Optimization of Inspection for Piping and Pressure Vessels Based on Quantitative Models of NDE Reliability D. STRAUB, TU München, Germany Abstract. Risk-based inspection (RBI) strategies have become a standard in the oil and gas industry, following the API 580 procedure. The models of inspection quality used in this procedure, however, are only of a semi-quantitative nature and describe the quality in terms of five qualitative categories of inspection effectiveness. In this contribution, a novel procedure for life-cycle optimization of process systems and pipelines will be presented, which is based on fully quantitative inspection models, including POD, PFA and sizing error models. The proposed procedure is based on the Bayesian decision analysis, combined with a recently developed model for reliability updating following inspections in process systems [1]. The procedure will, for the first time, enable a fully quantitative optimization of inspection efforts in process systems according to the minimum expected cost criterion and will facilitate demonstrating compliance of the selected inspection strategy with prescribed risk acceptance criteria. 4th European-American Workshop on Reliability of NDE - Fr.2.A.1 1 www.ndt.net/index.php?id=8355

Upload: others

Post on 24-Dec-2020

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Risk-Based Optimization of Inspection for Piping and Pressure … · 2010. 1. 12. · Risk-Based Optimization of Inspection for Piping and Pressure Vessels Based on Quantitative Models

Risk-Based Optimization of Inspection for Piping and Pressure Vessels Based on

Quantitative Models of NDE Reliability

D. STRAUB, TU München, Germany

Abstract. Risk-based inspection (RBI) strategies have become a standard in the oil and gas industry, following the API 580 procedure. The models of inspection quality used in this procedure, however, are only of a semi-quantitative nature and describe the quality in terms of five qualitative categories of inspection effectiveness. In this contribution, a novel procedure for life-cycle optimization of process systems and pipelines will be presented, which is based on fully quantitative inspection models, including POD, PFA and sizing error models. The proposed procedure is based on the Bayesian decision analysis, combined with a recently developed model for reliability updating following inspections in process systems [1]. The procedure will, for the first time, enable a fully quantitative optimization of inspection efforts in process systems according to the minimum expected cost criterion and will facilitate demonstrating compliance of the selected inspection strategy with prescribed risk acceptance criteria.

4th European-American Workshop on Reliability of NDE - Fr.2.A.1

1

ww

w.ndt.net/index.php?id=

8355

Page 2: Risk-Based Optimization of Inspection for Piping and Pressure … · 2010. 1. 12. · Risk-Based Optimization of Inspection for Piping and Pressure Vessels Based on Quantitative Models

Risk-based optimization of inspection for piping and pressure vessels based on quantitative models of NDE reliability

Prof. Dr. Daniel StraubEngineering Risk Analysis Group, TU München

[ NDE Reliability - BAM Berlin, June 2009 ]

2

Utilize NDE Reliability Models for Planning of Inspections

• Risk-based inspection(RBI)

• API 580 approach is semi-quantitative –how to utilize NDE reliability models?

• Fully quantitative physically -based approaches are desirable

IV

IV

IV

IV

IV

IV

IV

III

III

III

II

II

II

I

II

1 2 3 4

1

2

3

4

Incr

easi

ngpr

obab

ility

Increasing consequences

2

Page 3: Risk-Based Optimization of Inspection for Piping and Pressure … · 2010. 1. 12. · Risk-Based Optimization of Inspection for Piping and Pressure Vessels Based on Quantitative Models

3

Quantify the effect of inspections : Bayes

� � � � � �|f Z L f� �� � � �

� � � �Pr |L Z� � �

• NDE model when no defect is found: PoD• NDE model when defect is found: measurement accuracy

4

Effect of inspections

• Computation of life-cycle cost requires computation of reliability following inspections (assuming no-detection):

Straub D., Faber M.H. (2006). Computer-Aided Civil and Infrastructure Engineering, 21(3), pp. 179-192.

3

Page 4: Risk-Based Optimization of Inspection for Piping and Pressure … · 2010. 1. 12. · Risk-Based Optimization of Inspection for Piping and Pressure Vessels Based on Quantitative Models

5

Optimization of life-cycle cost

Straub D., Faber M.H. (2004). J. Offshore Mechanics and Arctic Engineering, 126(3), pp. 265-271.

6

For Fatigue Deterioration in Structures …

• RBI based on physical models of deterioration and quantitative inspection models

• Successfully applied in practice: iPlan software (www.matrisk.com)

4

Page 5: Risk-Based Optimization of Inspection for Piping and Pressure … · 2010. 1. 12. · Risk-Based Optimization of Inspection for Piping and Pressure Vessels Based on Quantitative Models

7

We do not know, whether the observed is the largest defect in the element

8

Effect of inspections in process systems & pipingwith localized defects

Straub D. (submitted). Structural Safety.

Missed defect is dominating

Identified defect is dominating

5

Page 6: Risk-Based Optimization of Inspection for Piping and Pressure … · 2010. 1. 12. · Risk-Based Optimization of Inspection for Piping and Pressure Vessels Based on Quantitative Models

9

Quantitative RBI for process and pipeline

i. Determine consequences of failureii. Establish probabilistic deterioration modeliii. Select different inspection strategies (inspection times &

methods) with NDE reliability modelsiv. At each inspection, consider two outcomes:

– Largest identified defect does not become critical– Largest identified defect becomes critical

and compute corresponding probabilitiesv. Determine acceptability and optimality of each strategy

(Bayesian decision analysis)

10

Example: Process equipment (pipe) subject to CO2 corrosion

6

Page 7: Risk-Based Optimization of Inspection for Piping and Pressure … · 2010. 1. 12. · Risk-Based Optimization of Inspection for Piping and Pressure Vessels Based on Quantitative Models

11

1) Consequence assessment(personel, public, economical, ecological)

TV-5001 TV-5002 TV-5003 TV-5004

TV-5005 TV-5006 TV-5007 TV-5008

TV-5009 TV-5010

TV-5011

TV-5012

TV-5014

TV-5016

TV-2001TV-2003TV-2005

TV-2002TV-2004TV-2006

RIO SECO

ALMACENAMIENTO

TRATAMIENTO DEEFLUENTES

PATIO DETUBERIAS

LAGUNETA CHENAQUE

ZONA ADMINISTRATIVA PREESTABILIZACIONDE CRUDO MARINO

H-25

H-26

H-06 H-07 H-08

H-23 H-24

H-100

H-112

H-129

H-131

H-202

H-204A

H-204

H-208

H-211

H-406

B-26 H-101

H-114

H-107

B-30 M-119

B-27 H-102

H-108

H-113

H-117

B-31 M-120

H-109

H-115

H-118

H-108A

B-28H-103

H-116

H-110

H-121 B-32

H-104 B-29

H-106

H-105

H-124

H-124A

H-127

B-76 B-77H-200

B-34

M-26

M-28

M-30

B-35

B-78H-201 H-218

M-31

M-29

M-27

H-132

H-128

H-133 B-36 H-134 B-37

H-130

H-126

M-32

M-34

M-36

M-38

M-39

B-40

M-41

M-43

M-45

M-35

B-35

M-37

M-39

M-40

M-42

M-44

H-205

H-204

M-206

M-47

M-48

B-49

B-50

B-47A

H-137

H-140

H-136

H-139

H-143 H-144H-142 B-38

H-138

H-135

H-203

H-207H-206

H-220

H-222B-81

H-219

H-205

B-80 H-221AB-19 H-221

H-212

H-209

H-215H-214

H-213

H-210

M-51M-216 B-82 M-52

M-60

B-86

M-58

M-62

M-59

M-61

M-63

B-53 M-54 B-85A

M-85

H-401A

B-126

H-401 H-402

H-405 H-404

H-403H-400

H-411A

H-411

H-412

M-02A

B-05

M-01A

B-04

H-408A

H-408M-409

M-93

M-94

B-409A

B-01

B-02B-03

H-147

H-146

H-123A H-123 H-125

H-148

M-55 M-56 M-57 M-88

M-151

M-152

M-153

M-154 M-156

M-155

M-158

M-157

M-160

B-176

M-159

B-177

M-151

B-182 M-162

M-165

M-172

M-178

M-184

M-151

M-166

M-173

M-163 M-164

M-169M-168

M-177M-176

M-182M-189

M-185

M-179 M-180

M-175M-174

M-183

B-181

B-179

B-180

H-122A H-122B B-33

TORNO LARGO

2000 m.

2500 m.

Nivel Fatal deRadiación

Nivel de Daños Serios deRadiación

O SECO CO

TORRM-57

1

O LLL

006

M-30

B-35

B-78 H-218

H-206

B-81H-221A

H-210

M-

H-118

H-1166

H-121 B-32

M-29

M-27

H-122A

B-34

M-26

M-28

04

H-106

M-32

B-33

TV-50

B-47A

MA

164

69

77

M-154

M 155

M-52

TV-2001

H-203H-220

M-38

M-39

B-40

M-39

M-40

M-47M-4

M-48

B-49

B-50

H-126

2

M-34

M-36

M-35

B-35

M-37

H-205

H-204

M-206M-2

122BH

RNO RTORNRTTORNOO

RIO

H-123A

M-153

M-156

M-158

M-157M

M-160M

B-176

M-159

B-177

M-151

B-182 M-162

M-165 M-166

O SEC

TV-2002

H-213

M-60

M-62

B-53 M-54 B-85A

M-88

M-152

B-86

C

H-207

H-22

M-31

TV-5

H-1288

TV-501

H-123 H-125

H-148

30

B-4

M-41

M-43

M-45

M-

M-44

6

-58

TV-5011

M-61

M-63M-85

H-147

H-146

008

010

B-37

H-1

H-137

H-140

H-144

V-5007 TV-50

TV-5009 TV-50

H-1322

H-133 B-36 H-134

H-136

H-139

H-143H-142 B-38

H-138

H-135

M-59

7

22

Source: COMIMSA

12

2) Probabilistic deterioration model

Here: DeWaards-Milliams model:

� �10 2

2

5.8 1710 0.67 log10 o COT fCOR

� �

� �2 2

0.0031 1.410 o oP TCO COf P �

2 2COCO oP n P

� �2M COS t X R t

7

Page 8: Risk-Based Optimization of Inspection for Piping and Pressure … · 2010. 1. 12. · Risk-Based Optimization of Inspection for Piping and Pressure Vessels Based on Quantitative Models

13

3) Select inspection techniques (NDE reliability)

• Ultrasonic scanning technology:

0

0.25

0.5

0.75

1

0 2 4 6 8 10

Corrosion defect depth s [mm]

Prob

abili

ty o

f det

ectio

nPoD

Mean PoD

95% confidencebounds

14

4) Event tree model for all inspection strategies

• At each inspection, verify whether largest identified defect becomescritical (i.e. determines the probability of failure)

• If so, perform a mitigation action• If not, continue with the regular inspection schedule

8

Page 9: Risk-Based Optimization of Inspection for Piping and Pressure … · 2010. 1. 12. · Risk-Based Optimization of Inspection for Piping and Pressure Vessels Based on Quantitative Models

15

How to determine the critical size at an inspection

• The defect size that would dominate the probability offailure before the next inspection

16

How to determine the critical size at an inspection

• The defect size that would dominate the probability offailure before the next inspection

9

Page 10: Risk-Based Optimization of Inspection for Piping and Pressure … · 2010. 1. 12. · Risk-Based Optimization of Inspection for Piping and Pressure Vessels Based on Quantitative Models

17

How to determine the critical size at an inspection

• The defect size that would dominate the probability offailure before the next inspection

18

How to determine the critical size at an inspection

• The defect size that would dominate the probability offailure before the next inspection

10

Page 11: Risk-Based Optimization of Inspection for Piping and Pressure … · 2010. 1. 12. · Risk-Based Optimization of Inspection for Piping and Pressure Vessels Based on Quantitative Models

19

4) Event tree model for all inspection strategies

• As long as no critical defect (sM<sD) is found:

20

5) Determine expected cost for each strategy

11

Page 12: Risk-Based Optimization of Inspection for Piping and Pressure … · 2010. 1. 12. · Risk-Based Optimization of Inspection for Piping and Pressure Vessels Based on Quantitative Models

21

5b) Verify acceptability

Compare the maximum probability of failure with given criteria

22

Conclusion

• In planning inspections over the life-cycle of the structure, we must anticipate all possible inspection outcomes

• Presented method provides a consistent, computationally feasible approach

• Physically-based deterioration models and fully quantitative NDE reliability models are the future

12