You are on page 1of 53

MFL and STARS technology

Improving Inspection Through Technology


MFL and STARS technology
 Aim
 To explain the theory of the technology utilised by the scanner,
how it generates percentage loss values and identify what
factors can affect the MFL signal.
 Learning Outcomes
 Describe the MFL principle and STARS principle.
 Discuss the purpose of the calibration and reference plate.
 Establish a percentage value and surface origin from the
calibration trace and detected MFL and values.
 Identify a suitable Suggested Minimum OperatioSTARS ‘u’nal
Threshold level based on the Signal to Noise Ratio of the
calibration trace.
 List the factors that affect MFL detection and discover how
each can alter the systems results.
2
Magnetic Flux Theory
 Magnetic Fields
 To understand how magnets interact with each other the
concepts of a “Magnetic Field” is used.
 The idea of a magnetic field is based on the patterns made by
ferrous particles when they are placed in a magnetic field.
 These patterns are called “Magnetographs”.
 Magnetic fields are thought
to consist of “Lines of Flux”.
 A bar magnet can be
described as a “Dipole”.
NN SS

3
Magnetic Flux Theory
 Properties of Lines of Flux
 They flow from a North Pole to a South Pole outside the
magnet.
 They flow from a South Pole to a North Pole inside the
magnet.
 Most densely packed at the poles.
 They form closed loops.
 Take path of least resistance.
 They never cross.
 Like poles repel.
 Opposite poles attract.

4
Magnetic Domain Theory
 A magnetic domain is region in which the magnetic fields
of atoms are grouped together and aligned.
 In an un-magnetised object, the magnetic domains are
pointing in different directions.
 Each domain comprises of 1015 to 1020 atoms – typically
there are several million domains in each individual grain
of the material.
N
N

S
N S
N N S
S N
S

N
S

N
S
S
N N N

S
S

Magnetised Bar
Un-Magnetised

5
Magnetic Domain Theory
 When
If the Magnetising
the domainsForce
are partially
is increased
aligned
andbythe
andomains
outside
become fully aligned
“Magnetising Force” athe
ferromagnetic
bar is said tomaterial
be magnetically
will
saturated.
become magnetised.
 If the Magnetising Force is increased and the domains
become fully aligned the bar is said to be magnetically
saturated. MAGNETISING FORCE
MAGNETISING FORCE

S N
N
N

N
S N
N
S S
S
N
N
S N S N
N S S
N
N
S S
S

N
S

N
S N
S
N
S N S S
N S S
N
N N N

S
S

Saturated Magnetised
Un-Magnetised
Saturated BarBar
Magnetised

6
Magnetic Flux Leakage Principle
 Location of the magnetic bridge and MFL sensor array.
 The system is detecting a change in the quantity of flux
leakage between floor plate and MFL sensors.

Air gap
(Measuring flux leakage)

Bridge Permanent Magnets

Pole piece Pole piece


MFL sensors
Floor plate

7
Magnetic Flux Leakage Principle
 Scanner magnets creates a localised magnetic field that
fills or “Saturates” the full thickness of the plate.
 Corrosion causes the magnetic field to “leak” out of the
plate.
 Hall effect sensors located in the MFL sensor head detect
the leakage field and resulting signals are processed to
show an estimate percentage loss (EPL).

8
Magnetic Flux Leakage Principle
 The leakage field is primarily affected by the volume loss
of the indication.
 The larger the indication, the greater the leaking field the
so larger the outputted MFL signal.
 The raw MFL view is represented as a dark then light
rendered image.

DIRECTION OF SCANNING

40% EPL 60% EPL 60% EPL


9
Magnetic Flux Leakage Principle
 A major limitation of MFL is that it cannot discriminate
the surface origin of indications.
 Consider the response from a know feature that has a
50% wall loss and is 5mm long.
0.15
0.15
Defect from top (By)
Defect from top (By)
Defect from bottom (By)
Defect from bottom (By)
0.1
0.1

M agnetic flux density (T)


M agnetic flux density (T)

0.05
0.05

0
0

-0.05

 Difficult to discriminate
-0.05

-0.1

as there is a 7% difference -0.1

in the peak to peak. 0 10 20 30 40 50


0 10
60
Scan distance (mm)
20
70
30
80
40
90 100
50
Scan distance
60
(mm)
70 80 90 100

 Another technique is required for surface discrimination.


10
STARS Principle
 Consider the magnetic bridge and MFL sensor array.
 STARS sensor array are located below the rear pole piece.
 The sensors measure the rate of change of the magnetic
field under the pole.
 This rate is dependant on the field disturbance caused by
changes to the surface.
Bridge Permanent Magnets

Pole piece Pole piece


MFL sensors
STARS sensors

11
STARS Principle
 Bottom side indications have no effect on STARS sensors
because magnetic field under the pole does not change.
 Top side indications cause a disturbance in the magnetic
field so the rate of change is measured by STARS sensors.
 STARS technology has been developed by Silverwing.
Surface Topology Air-gap Reluctance Sensors

POLE PIECE POLE PIECE


STARS STARS
SENSORS SENSORS

12
STARS Principle
 The raw STARS view of the data is represented as a dark
then light rendered image. Similar to that of the MFL.
 If there is a depression on the surface (corrosion pit for
example), the image is black then white.

DIRECTION OF SCANNING

 If there is a protrusion on the surface (weld spatter), the


image is then white to black.
DIRECTION OF SCANNING

13
MFL Reference plate
 To accurately size any discontinuities the system must
first be calibrated on the reference plate.
 The reference plate have machined pits to represent
different values of estimated percentage loss (EPL).
 These pits represent volumetric loss and are created
using a 22mm diameter ball end cutter.
 Available reference plate thicknesses:
 6mm – CP30 [1/4inch – CP72]
 8mm – CP28 [5/16inch – CP74]
 10mm – CP29 [3/8inch – CP75]

 12mm – CP31 [1/2inch – CP76] 22mm diameter ball


end cutter
14
MFL Reference plate
 Defect dimensions vary depending upon plate thickness.
 Depth is based on percentage loss of plate thickness.
 Diameter is governed by the tool diameter at the depth.
 Volume is based on depth, tool diameter and plate thickness.
20% 40% 60% 80%

Plate
Thickness

20% defect 40% defect 60% defect 80% defect


Plate
Thickness
Depth Diameter Volume Depth Diameter Volume Depth Diameter Volume Depth Diameter Volume

6mm 1.2mm 10.0mm 20.81mm3 2.4mm 13.7mm 109.48mm3 3.6mm 16.3mm 282.97mm3 4.8mm 18.2mm 542.87mm3

8mm 1.6mm 11.5mm 41.95mm3 3.2mm 15.5mm 215mm3 4.8mm 18.0mm 535.63mm3 6.4mm 20.0mm 1012.28mm3

10mm 2.0mm 12.5mm 70.16mm3 4.0mm 17.0mm 360.24mm3 6.0mm 19.5mm 876.50mm3 8.0mm 21.0mm 1574.99mm3

12mm 2.4mm 13.7mm 109.48mm3 4.8mm 18.15mm 541.06mm3 7.2mm 20.65mm 1290.67mm3 9.6mm 21.8mm 2229.37mm3

15
MFL Reference plate
 Estimate Percentage Loss (EPL) values are generated
based on the defects within the calibration plate.
 Reference plates must not contain pitting, weld repairs,
deep scratches, through holes or any other defects.
 Incorrect shaped defects in the reference plate can
prevent accurate calibration and sizing.

Straight edges from Burs around cut Incorrect tooling Drill bit used
drill bit used
16
Purpose of the calibration
 The MFL calibration trace links the amount of magnetic
leakage signal to a defect of a known EPL depth and
volume found in the reference plate.
Amount of leaking magnetic field

Defect Volume Estimated Percentage Loss depth

 The magnetic leakage signal is identified for defects


located on the Bottom side and Top side of the plate.
 The STARS calibration trace links the rate of change of the
magnetic field disturbance for top side defects of known
EPL depth and volume found in the reference plate.
17
Calibration trace and the unit ‘u’ value
 Two calibration traces display the MFL and STARS data
found for the 20%, 40%, 60% and 80% defects.
 The calibration trace displays the data as a unit ‘u’ value.
 Y-Axis represents a unit ‘u’ value, ranging from 0 to 256.

18
Calibration trace and the unit ‘u’ value
MFLi ‘u’ value
 This can be considered as the amount of magnetic leaking
field coming out from the top of the plate.
 The larger the defect the greater the leaking magnetic
field and so the higher the displayed MFLi ‘u’ value.
 A very low background MFLi ‘u’ value is displayed when
no indications are present within the plate.
higher MFLi ‘u’ value
lower MFLi ‘u’ value

19
Calibration trace and the unit ‘u’ value
Extracting the MFLi ‘u’ value
 Consider top side and bottom side traces separately.
‘u’ value 20% 40% 60% 80% Background
Bottom MFLi
Bottom MFLi 40
40 105
105 163
163 204
204 9
9
Top
Top MFLi
MFLi 56
56 109
109 149
149 173
173 10
10

u = 204 80%

u = 173 80%
u = 163 60%
u = 149 60%

u = 105 40% u = 109 40%

u = 56 20%
u = 40 20%

u = 9 Background u = 10 Background

BOTTOM TRACE TOP TRACE

20
Calibration trace and the unit ‘u’ value
STARS ‘u’ value
 This can be considered as the rate of change in the
magnetic field as it enters the top of the plate.
 The greater the change in the magnetic field as it enters
the top of the plate the higher the displayed STARS ‘u’
value.
 A very low background STARS ‘u’ value is displayed when
no indications are present on the top of the plate.

Background low STARS higher STARS


STARS ‘u’ value ‘u’ value ‘u’ value

21
Calibration trace and the unit ‘u’ value
Extracting the STARS ‘u’ value
 These are for top side defects only.
‘u’ value 20% 40% 60% 80% Background
STARS 47 104 120 142 8

u = 142 80%
u = 120 60%
u = 104 40%

u = 47 20%

u = 8 Background

22
Surface Origin and Percentage Loss
 MFL and STARS ‘u’ values are compared to the calibration
to determine percentage and surface origin.
 Estimated Percentage Loss is from MFL ‘u’ value data.
 Surface Origin is determined by comparing the STARS ‘u’ value
data to the MFL ‘u’ data.
MFL Severity

STARS Top
NormaliseMapdiffalong 227

MFL
1

0.9 Severity
0.8

0.7

STARS
0.6
Bottom
0.5

0.4
23
Surface Origin and Percentage Loss
 What percentage would an MFLi ‘u’ value of 174 show?
 Check to see what value the detected STARS ‘u’ is.
 If STARS ‘u’ is around 146: 81% Top
 If STARS ‘u’ is at the background level:65% Bottom

65% BOTTOM
u = 174 81% TOP
u = 146

u=4

24
Surface Origin and Percentage Loss
 What percentage would an MFLi ‘u’ value of 174 show?
 Check to see what value the detected STARS ‘u’ is.
 If STARS ‘u’ is around 146: 81% Top
 If STARS ‘u’ is at the background level:65% Bottom

Top EPL View Bottom EPL View

25
Surface Origin and Percentage Loss
 Always confirm the origin by checking MFL, MFLi and
STARS views and when possible visually inspect the floor.
 Example of incorrect surface origin.
 Top percentage view displays six indications.
 STARS view only displays four indications.
 The two highlighted indications should be on the bottom side.
Top EPL View STARS View

26
Dynamic Cursor
 Dynamic cursor is a software approximation tool for
estimating geometry based on the calibrated geometry.
 The dynamic cursor consists of two circles linked by a
mesh.
Outer Circle
Inner Circle
 The diameter of the cursor is determined by the MFL
calibration trace and changes to represent the expected
size of a percentage indication.

27
Dynamic Cursor
 The diameter is determined for each percentages on the
MFL calibration trace by looking at the width of the MFL
signal a predetermined level below its peak.
 Between each known percentages the diameter of the
80%
cursor changes linearly.
60%
 The mesh has a tolerance of
+/-20% of the diameter.
40%

+/-20% of
diameter 20%

28
Dynamic Cursor
 Typical examples of approximating defect geometry.
1. Indication lies between the inner and outer circle.
 Likely to be accurately sized to the calibrated indications.
2. Indication lies outside of the outer circle.
 Likely to be oversized and display a higher EPL than expected.
3. Indication lies within the inner circle.
 Likely to be undersized and display a lower EPL than expected.

1. Accurately Sized 2. Oversized Indication 3. Undersized Indication


Pitting Type Lake Type Pipe Type

29
Indication colour gradient
 Look at how the colour gradient changes to illustrate the
geometry of the indication.

 Compare this to the know geometry of the calibrated


indications from the verification scan.

30
Signal to Noise Ratio
 Signal-to-noise ratio is a measure that compares the level
of a signal to the level of background noise
 For MFL based inspections the background noise can
affect the detection capability of the MFL equipment.
 As the scanners sensitivity levels are increased so to is
the noise emanating from the inspection surface making
the identification of true defects more difficult.
 Factors that can affect the signal to noise ratio include
 Surface cleanliness.
 Plate and coating thickness.
 Sensor height.

31
Signal to Noise Ratio
 The MFL calibration trace can be used to identify the
signal and background levels.
 Typically an expectable signal to noise ratio is 3:1.

80%

60%

40%
20% Signal is above
the 3:1 ratio
20%

3:1 ratio is 45u


Noise level 15u

32
Signal to Noise Ratio
 Suggested Minimum Operational Threshold (SMOT).
 The signal to noise ratio affects the minimum discontinuity
percentage that can be reliably detected.
 Viewing the calibration trace to can help to determine a
Suggested Minimum Operating Threshold percentage level to
view the inspection data.
 The SMOT percentage level must
be set above the 3:1 ratio. SMOT of 40% since it is
above the 3:1 ratio

3:1 ratio is 45u


Noise level 15u

33
Signal to Noise Ratio
 Suggested Minimum Operational Threshold (SMOT).
 When viewing inspection data below the SMOT of 40% the noise
generates lots of spurious indications.
 Setting the SMOT to 40% helps clean up the inspection data by
removing the spurious indications.

SMOT of 40%

34
Factors that affect the MFL signal
 Important to understand how each of the following
factor affect the detection of the MFL signal and the
resulting percentage value.
 Magnetic Saturation.
 Corrosion Shape.
 Scanning Orientation.
 Surface Origin.
 Magnetic Lift-Off and Sensor Height.
 Tank Floor thickness change.
 Coating thickness change.
 Surface Condition.

35
Magnetic Saturation
 This is a key fundamental element of MFL technology.
 Factors that can affect the saturation level of a plate:
 Magnet Strength.
 Plate thickness.
 Coating thickness.
 Three main states of saturation levels:
 Perfectly Saturated.
 Over Saturated.
 Under Saturated.

36
Magnetic Saturation
 Perfect Saturation
 This is the ideal case.
 When a plate is fully saturated no flux leakage is evident until a
defect is present.
 The flux leakage is proportional to the volumetric loss of the
defect so allows defects to be sized correctly.

37
Magnetic Saturation
 Over Saturation
 This is when there is flux leakage outside of the plate even
when no defect is present.
 The flux leakage from a defect will be greater than its
volumetric loss.
 This can increase the likelihood of noise and spurious
indications as it causes defects to be oversized.

38
Magnetic Saturation
 Under Saturation
 This is the least desirable case since the plate cannot achieve
saturation.
 The flux leakage from a defect will be much smaller than its
volumetric loss.
 This can increase the likelihood of indications being missed as it
causes defects to be undersized.

39
Corrosion Shape
 MFL is considered as detecting defect volumetric loss.
 Volumetric loss does not always reflex as defect depth.
 The calibration links volumetric loss to a specific depth.
 Consider the following defects:
Volume = Length x Width x Depth
depth
Lake Type 6 = 2 x 3 x 1 depth
depth width
Conical pit Type 6 = 1 x 3 x 2 depth width
length width
Pipe Type 6 = 1 x 2 x 3 length
width
length
length
 Identical Volumes but at different Depths
 How are percentage values affected by defects with the
same volume but a different depth?
40
Corrosion Shape
 Lake Type
 Large diameter compared to their actual depth. LAKE TYPE
 Creates a large flux leakage signal.
 The percentage value is expected to be oversized.
 Conical Pit Type
 Conical in shape with rounding at deepest point.
CONICAL PIT
 Similar to the defect in the reference plate. TYPE
 The percentage value is expected to be +10% accurate.
 Pipe Type
 Small diameter compared to their actual depth. PIPE TYPE
 Creates a small flux leakage field.
 The percentage value is expected to be undersized.

41
Corrosion Shape
 Through Hole
 Holes typically form at the deepest point of conical pits.

 Can also appear as very small pipe like in shape.

 Leakage fields for holes can be very low since the field takes the
path of least resistance and so ‘jumps’ across the gap.

 The percentage value for a hole is expected to be undersized.


 A detailed visual inspection of the floor is always recommended.
42
Scanning Orientation
 Corrosion shape and geometry can result in different flux
leakage signals when scanned in different directions.
 This results in different percentage values for the same
indication.
 An example can be seen when transverse scanning.

Different percentage
values detected for
the same indications

No transverse scan Transverse scan


43
Surface Origin
 MFL alone cannot identify the surface origin of defects.
 Bottom side defects generate a larger flux leakage field
compared to the same defect on the top side.

 Percentage loss of defects on top and bottom together


cannot be added together due to their volume.

 6mm reference plate, 40% volume is 110mm3, 80% is 543mm3


 The percentage value is expected to be undersized.
 Does depend upon geometry and top / bottom loss quantity.
44
Magnetic Lift-Off and Sensor Height
 Magnetic lift-off is the distance between the floor plate
and the magnets.
 Sensor height is the distance between the floor plate and
the MFL sensor head.
 Alterations in these distances can be due to a buckled
floor, sensor head movement, cleanliness of the floor or
a change in coating thickness.

Magnetic bridge distance


MFL sensor head distance

45
Magnetic Lift-Off and Sensor Height
 Magnetic lift-off
 Magnetic lift off changes the how much magnetic field enters
into the plate which affects the flux leakage field.
 Increase in magnetic lift-off
 The percentage value is expected to be undersized.
 Decrease in magnetic lift-off
 The percentage value is expected to be oversized.

46
Magnetic Lift-Off and Sensor Height
 Sensor height
 A change in sensor height alters the amount of flux leakage
passing though the MFL sensors.
 Increased sensor height
 The percentage value is expected to be undersized.
 Decreased sensor height
 The percentage value is expected to be oversized.

47
Plate thickness change
 It is acceptable to use a MFL reference plate up to 1mm
[0.04inch] thicker than tank floor plate.
 Do not recommended to use a thinner reference plate.
 Decrease in floor plate thickness
 The percentage value is expected to be oversized.
 Increase in floor plate thickness
 The percentage value is expected to be undersized.

Reference Plate
Thickness

48
Coating thickness change
 Nominal coating thickness greater than 0.5mm [0.02inch]
must be accounted for during calibration.
 Decrease in coating thickness
 The EPL value is expected to be oversized.
 Increase in coating thickness
 The EPL value is expected to be undersized.

Simulated Coating
thickness

49
Surface condition
 Tank floor cleaning is required before MFL inspection.
 Dirty or rough inspection surfaces will cause spurious
indications and generate incorrect percentage values.
 Ensure there is no scale, product residue, loose metal
and debris or any blasting cleaning media.

ap
G

Thick Scale Product Residue Rough inspection surface


50
Surface condition
 Mechanical defects will cause spurious indications and
generate incorrect percentage values.
 This include weldment, weld spatter, grinding marks,
dented plates, blistered or delaminated coatings.

Weld spatter Grinder marks Blistered coating


51
Practical Session
 Exercise
 Determine a Suggested Minimum Operational Threshold Level
for the following calibration traces:

52
Practical Session
 Reviewing the Learning Outcomes
 Describe the MFL principle and STARS principle.
 Discuss the purpose of the calibration and reference plate.
 Establish a percentage value and surface origin from the
calibration trace and detected MFL and STARS ‘u’ values.
 Identify a suitable Suggested Minimum Operational Threshold
level based on the Signal to Noise Ratio of the calibration trace.
 List the factors that affect MFL detection and discover how
each can alter the systems results.

53

You might also like