Principles of Nondestructive Evaluation

1
Principles of Nondestructive Evaluation
Lecture 39/20/99

• Shreekanth Mandayam
• Graduate / Senior Elective
• 0909-504-01/0909-413-01
• Fall 1999
• http//engineering.rowan.edu/shreek/fall99/nde/

2
Plan
9/27/99

• Magnetic Flux Leakage (MFL) NDE
• Principle
• Governing equations
• Practice
• Class projects
• Paper formats
• Open discussion
• Task assignment

3
Direct Current Magnetization
Scanner
Hall Probe
Specimen
Current Lead
4
Magnetic Flux Leakage Signals
600
400
300
500
200
400
100
0
300
-100
200
-200
100
-300
-400
0
Axial Component of Flux Density
Radial Component of Flux Density
5
MFL Image from a Rectangular Slot
6
Magnetic Flux Leakage (MFL) Detection of Defects
Specimens
Magnetic Images
7
Maxwells Equations
8
Electromagnetic NDE Methods
9
Static Phenomena Magnetic Flux Leakage
10
Static Phenomena MFL (contd.)
Elliptic partial differential equation
11
NDE Processes
Inverse Problem Difficulty
Forward Problem Difficulty
Informational Entropy
HIGH
LOW
HIGH
Elliptic Processes
Parabolic Processes
Hyperbolic Processes
LOW
LOW
HIGH
12
Gas Transmission Pipeline Inspection

• 280,000 miles
• 24 - 36 inch dia.

SCC
Weld
Valve
T-section
Sleeve
Corrosion
13
Gas Pipeline Incidents in the US
Mechanical damage is the single largest source of
gas pipeline related incidents.
14
Permanent Magnet
Data Acquisition and Storage
Hall-effect Sensors
15
Gas Pipeline Inspection
Defect
Pipewall
sensor
Magnetic Flux Leakage (MFL)
The Pig
Data Acquisition
Drive Section
Sensors
Brushes
16
Defect Characterization

• Artificial Neural Networks
• Multidimensional mapping from
• MFL signal to defect profile

MFL Signal
Defect Profile
mapping
PIPE
a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11
a12 T
0 0 0 50 50 50 40 40 50 0 0 0 T
17
Defect Characterization
MFL Signals
Defect Profiles
18
Typical Results
MFL Signal
Predicted Profile
1-D Scan of Predicted Profile
19
Radial Basis Function Neural Network
input layer
Input processed signal
Output defect profile vector
N
20
(No Transcript)
21
Governing Equation
Probe Velocity
Permeability
Remanent Magnetism
Stress
Sensor Location
22
Effect of Pipe Grade
Family of B-H Curves
Fixed Defect
Fixed B-H Curve
1400
Effect of Defect Depth
1200
1000
800
600
400
200
0
Depth in of pipe-wall thickness
1
3
5
7
9
11
13
15
17
19
21
23
position
23
Invariance Transformation

• Identify at least two distinct test signals
• Synergistically combine to isolate unique
  defect signature

Features from Tangential Component of Flux
Density ( Bz) Pz (d, l, w, t)
Invariance Transformation Function h (d, l, w)
Parameter-Invariant Defect Signature
Features from Normal Component of Flux Density (
Br) Pr (d, l, w, t)
24
Typical Results Pipe-wall Thickness
Wall thickness
1/2 3/8 5/16 1/2 3/8
5/16
25
Compensation Results
26
Experimental Set-Up
27
MFL Scans
Defect Depth
Pipe Grade
X-42 X-52 X-65
X-70
0.06 0.17 0.25
Line Scans
28
Compensation Results