METAL LOSS IN-LINE INSPECTION SURVEYS - PowerPoint PPT Presentation

1 / 22
About This Presentation
Title:

METAL LOSS IN-LINE INSPECTION SURVEYS

Description:

INSPECTION SURVEYS LIMITS AND ... 0.2t Case Study High Resolution Metal Loss Survey 36 ... 1 * Exceeding RSTRENG without 10% surge allowance OPS Position Needed on ... – PowerPoint PPT presentation

Number of Views:120
Avg rating:3.0/5.0
Slides: 23
Provided by: Deav7
Learn more at: https://pstrust.org
Category:

less

Transcript and Presenter's Notes

Title: METAL LOSS IN-LINE INSPECTION SURVEYS


1
METAL LOSS IN-LINEINSPECTION SURVEYS
  • LIMITS AND INACCURACIES

2
BACKGROUND
  • Began use in 1960s.
  • Initial inspection tools were low resolution, low
    power tools.
  • Suitable for finding where to look for corroded
    pipe.
  • Better resolution and power offered in 1980s.
  • High resolution tools began use in late 1980s.
  • Despite limitations and inaccuracies, pipelines
    began to calculate pressure limits based on
    in-line inspection grading in late 1990s/early
    2000s.

3
  • INACCURACIES IN PRESSURE
  • LIMIT CALCULATIONS

4
(No Transcript)
5
RUPTURES
6
LEAKS
7
High Resolution Inspection Tool Stated Limits
and Inaccuracies
  • API 1160
  • Inspection Companys Specs

8
API 1160 Limits onHigh Resolution Tools
  • Cannot detect or reliably detect
  • Narrow axial external corrosion
  • Cracks and crack like defects
  • Laminations and inclusions
  • Pipe mill anomalies
  • Can detect, but cannot identify or size
  • Dents, wrinkle bends and buckles
  • Gouges

9
High Resolution PigSpec. Limitations
  • Cannot detect or unknown accuracy
  • Surface areas smaller than t x t
  • Axially oriented areas, width less than 2t
  • Areas interacting with weld, smaller than 3t x 3t
  • Minimum reported depth
  • For t x t to 2t x 2t, 0.4t
  • For larger than 2t x 2t, 0.2t
  • For larger than 3t x 3t, 0.1t
  • For areas interacting with weld, larger than
    3t x 3t, 0.2t

10
High Resolution PigSpec. Accuracy
  • Width and length
  • Isolated pits, t
  • Complex shapes, 2t
  • Depth
  • For less than t x t, unknown
  • For t x t to 2t x 2t, 0.2t
  • For larger than 3t x 3t, 0.1t
  • For areas interacting weld, 3t x 3t and larger
  • 0.2t

11
Case Study High Resolution Metal Loss Survey
  • 36 miles of pipeline
  • 695 metal loss anomalies deeper than 15 were
    reported
  • Breakdown on graded wall loss
  • gt 50 t 1
  • 40 to 49 t 6
  • 30 to 39 t 31
  • 20 to 29 t 197
  • 15 to 19 t 460

12
Field Inspection Results
  • 200 anomalies were excavated
  • Detailed records produced on 11 pipe joints
    covering 62 anomalies
  • 90 of anomalies were deeper than graded
  • 58 of anomalies were longer than graded
  • 43 ungraded corrosion areas were found

13
Ungraded Corrosion Areas
  • 20 to 68 deep
  • 0.25 to 25 inches long
  • 0.25 to 10 inches wide
  • Two exceeded RSTRENG

14
Computer Display
15
Computer Display
16
Computer Display
17
Computer Display
18
Computer Display
19
Needed Post Field Inspection
Activities
  • Organize graded vs. found data
  • Regrade in-line inspection survey
  • New field inspection plan
  • Re-inspect and repair pipeline
  • Statistical analysis of areas not inspected
  • Apply future growth to anomalies

20
Statistical Analysis of Non-Inspected Areas
  • Bayes Theorem
  • First applied on TAPS
  • Also called probability of exceedance (POE)
    analysis
  • Probability that an anomaly will exceed a given
    integrity criteria

21
Non-Inspected Pipe POE Summary
  • POE Number of Pipe Joints
  • 20-29 5
  • 30-39 3
  • 40-49 4
  • 50-59 7
  • 60-69 2
  • 70-79 5
  • 80-89 3
  • 90-99 1
  • Exceeding RSTRENG without 10 surge
    allowance

22
OPS Position Needed on POE Non-Compliance Limits
  • Probability to Exceed Part 192 or Part 195
  • Per anomaly?
  • Per pipe joint?
  • Per pipeline section?
  • Per pipeline inspection survey?
  • Per pipeline system?
  • 1, 2, 5, 10, More?
Write a Comment
User Comments (0)
About PowerShow.com