TABLE OF CONTENT

CONTENTS PAGE NO.

1. Purpose 2

2. Scope 2

3. Personnel Required 2

4. Reference 2

5. Responsibilities 2

6. Equipment Requirements 2

7. Probe Deployment Considerations 2

7.1 Cleaning
7.2 General Considerations - All Welds
7.3 Finding the plane of crack
7.4 Cracking into Parent Plate
7.5 Through cracking
7.6 Circumferential Welds - No Ground Out Regions
7.7 Inspecting Ground Out Areas
7.8 Inspection during grinding
7.9 Inspecting Flat Plate Welded Supports/Stiffeners

8. Material Features
8.1 Coating thickness
8.2 Magnetic State
8.3 Surface Grinding or Work Hardening
8.4 Seam Welds

9. Reporting Requirements
9.1 General information
9.2 Scanning Data
9.3 Detailed Record of Indications/Anomalies

10. Interpretation of ACFM signals

10.1 General Method
10.2 Decision Tree for Identification of a crack
11. Crack Sizing

12. Length sizing with ACFM

1. Purpose

To establish the testing procedure for the Alternating Current Field Measurement (ACFM) to detect the crack
in uncoated tubular welded connections, structures, ground-out regions, and structures containing plate to
plate welds or stiffeners.

2. Scope

This procedure has been developed to be carried out ACFM inspection for coated and bare surface both.
It has been developed for the inspection of uncoated tubular welded connections, structures, ground-out
regions, and structures containing plate to plate welds or stiffeners.

3. Personnel Required

3.1 Experienced operator holding an ASNT ACFM Level – II certificate based on Oceancare written practice
shall be carried out test, interpret the test result and compile the r report.

3.2 Whenever remote probe operators are used, continuous audio communication is required to enable the
ACFM operator to lead the inspection and for the probe operator to be able to report back on any local
factors influencing the inspection.

4. REFERENCES
Reference Description
ISO9712 2005 Non-destructive Testing
ASNTSNT-TC-1A- 2006 Recommended Practice No. SNT-TC-1A - Non-Destructive
Testing
2006 Edition

5. RESPONSIBILITIES

5.1 The Technical / Operations Manager shall be responsible for ensuring that the necessary resources are
made available for the requirements of this procedure to be carried out.

5.2 ASNT NDT Level – III shall be responsible for ensuring that the requirements of this procedure are fully
implemented at all times.

5.3 All employees involved in the implementation of this procedure shall be responsible for adherence to
the requirements stated herein.

5.4 The Company HSSE Officer shall ensure that all operations are carried out in a safe manner.

6. Equipment Requirements

The following equipment list is for a particular job but does highlight essential and recommended
equipment that will be needed as a minimum.

 ACFM main equipment

 240V to 110V transformer (optional depending on local supply)
 Topside Unit power cable
 Communications cable
 ACFM Probes – Weld probe plus other types selected according to geometry and use
 Marker Pencils / chalk (Box) or magnetic arrows
 Ruler / Measuring Tape
  Function Check Block of the same material type as that to be inspected.
 Laptop portable PC
 Spare Computer Battery and Charger
 Oceancare ACFM Report Sheets

7. Probe Deployment Considerations

7.1 Cleaning

The surface must be cleaned sufficiently to allow smooth probe travel and to allow features such as
grinds or seam welds to be seen. It should be noted that cleaning to bright metal is not required.

The system operator shall confirm that the surface condition is acceptable prior to carrying out the
inspection, using information supplied by the diver and the divers video camera.

7.2 General Considerations - All Welds

The following notes on probe deployment describe how to inspect welded components for fatigue
cracks where it is assumed that defects will closely follow the weld line. The technique relies on
recognition of the signal from a probe scan along the length of a defect, so an ACFM probe is always
scanned along a line parallel to the weld. For this reason, defects that lie at an angle of more than about
25o to the weld may not be detected.

Standard weld probes should be used for all welds where access allows. A pencil probe should only be
used for inspecting ground out regions or other geometries where a weld probe cannot gain access.

The recommended scanning speed is about 10mm per second. The standard probe scans a width of
approximately 20mm. Scans should always be made along both weld toes and, if wider than 20mm, the
weld cap should also be covered by making a number of passes sufficient to cover the weld cap width
taking into account the coverage of the probe.

7.3 Finding the Plane of a Crack

If a defect indication is found during a weld toe scan, repeat the scan 10mm away from the toe in the
relevant plate and in the first interbead root on the weld cap to confirm the crack identification and to
identify whether the crack indication is due to interbead cracking. If the indication in the parent plate is
of similar amplitude to that of the weld toe, it is likely to be due to Seam Welds, Surface Grinding or
Work Hardening. Note that a seam weld may have been ground off and so not visible. Conduct another
scan 20mm from the weld toe to confirm this observation. If the indication during the bead scan is
larger than that observed during the weld toe scan, the defect must be in a weld cap. Repeat in the
adjacent roots until the defect indication is greatest. The root giving the greatest signal is the root
containing the defect.

Note that interbead cracking often jumps between beads, which will result in a sudden drop in the
signal from one bead coinciding with a sudden rise in the adjacent bead.

7.4 Cracking into Parent Plate

In some geometry, especially tubular intersections, long defects that start growing along a weld toe can
deviate into the parent plate. This will show itself as a sudden drop in Bx with no corresponding Bz peak
or trough. In this case, scans should be made on the parent plate, along the expected defect line, to find
the defect's end.

7.5 Through Cracking

 Where the crack depth calculated is greater than the plate thickness this indicates that cracking may be
completely through the weld and so the diver should be asked to look for further evidence of this. Such
evidence may be visual crack opening or, if the back face is accessible (e.g. on flat plates), crack-like
indications there.

7.6 Circumferential Welds - No Ground Out Regions

Where the length of weld to be inspected is greater than 400mm the weld should be marked up into
smaller lengths which overlap. (Longer scans tend to be more difficult for the diver to perform,
therefore increasing the chance of probe lift off etc.). The measurements from a datum to the four
cardinal clock positions should be recorded. The location of any features that affect probe movement or
probe signal should be reported (e.g. the position of a seam weld that joins the weld under inspection,
or the presence of weld spatter).

When a crack is located it should be re-scanned in more detail by scanning the defect area plus 30mm
before and after each end at a slower scan rate. The crack tips should be located and marked and
surface breaking crack length reported to the ACFM operator. The circumferential distance from a
datum (e.g. 12 o'clock position) to one end of the crack should also be noted.

If the weld is to be ground out to remove the defect the estimated depth using ACFM should be the
depth to which the first grind is made. Before grinding takes place the defect plus 30mm either side
should be scanned with a Pencil probe for later comparison. After grinding the grind should be re-
inspected with the same Pencil probe to ensure the defect has been removed (the pencil probe type
selected for the pre-grind scan should therefore be chosen to fit into the subsequent ground area).

7.7 Inspecting Ground Out Areas

Welds that have been ground to remove defects require the use of a Pencil probe. A pencil probe is
necessarily easier to tilt or rock than a weld probe and so extra care must be taken during a scan to
ensure that this does not happen. The first scan with the probe should be with the probe axis
perpendicular to the root of the grind. The weld should be scanned such that at least 30mm either end
of the ground out region is scanned. If a defect signal is found two further scans should be performed to
determine whereabouts around the grind the defect is located:

a) Probe axis at +30o angle from the normal to the root of the grind.
b) Probe axis at -30o angle from the normal to the root of the grind.

The maximum signal perturbation should be used to determine crack depth and should also indicate the
location of the defect in the ground out channel.

The ground out region of the weld should be marked at suitable intervals (e.g. 20mm) which should be
called to the ACFM operator as the probe scan passes each mark. A measurement of length from a
suitable datum (e.g. 12 o'clock position) to the weld grind marks should be made and also reported to
the ACFM operator.

Where the geometry of the grind start or finish is sharp this may obscure small defects. The ACFM
operator will advise whether further grinding is required to smooth off the edges of the grind to give
good probe movement and signals at the grind start.

7.8 Inspection during Grinding

If a detected defect is to be removed by grinding, a scan of the defect using a pencil probe must be
made prior to any grinding. This provides a baseline for comparison, removing the effects of variations
 in sensitivity between different probes. The whole of the grinding process should be monitored with the
same probe.

Between grinding passes, the defect should be rescanned. Particular note should be made of the
position of the defect in the grind profile, to ensure that the grind follows the defect even if the defect
curves away from the expected direction.

The ACFM operator should note changes in the ACFM signal between grinding passes, such as changes
in length, or splitting into two or more separate cracks. An apparent increase in length or depth may
indicate a sub-surface or branching crack.

7.9 Inspecting Flat Plate Welded Supports/Stiffeners

Full geometry detail is required by the ACFM operator before scanning can commence. As with tubular
components, the standard weld probe should always be used where geometry allows and other probe
types only used where necessary.

The use of a uniform input field, which allows defect sizing, means that the standard weld probe picks
up strong signals from sharp geometry changes such as plate edges, corners and rat-holes. This edge
effect occurs when a probe is used within a certain distance of the edge, the distance being roughly
equal to the width of the main part of the probe body. Thus a standard weld probe will pick-up edge
signals when the mid-point of the probe is within about 50mm of the edge. A pencil probe also
experiences edge effects, but to a much lesser degree. Mini/micro pencil probes can be used up to
about 10mm from an edge.

The presence of an edge does not preclude the use of standard probes but signal interpretation is not so
easy. The effect of an edge is to superimpose a general slope in probe readings that will increase in
gradient as the plate edge is approached. For detection, this has the effect of separating the signal from
either end of a defect on the butterfly plot so that the loop formed is not closed. Where cracks are
located within 50mm of a plate edge the depth of the defect should be calculated using an estimate of
what the background Bx would have been at the centre of the defect, as shown below.

Fig. Bx background and minimum values to be used near a plate edge

If there is not too much curvature in the signal, this background Bx is best estimated as an average of
the Bx values either side of the defect.

Instead of clock position marking, the weld to be inspected needs to be marked off in suitable linear
intervals (maximum 100mm).

8. Material Features
 8.1 Coating Thickness
With the ACFM technique, it is possible to detect defects through 5mm or more of non-conducting
coating. The signal strength obtained is obviously smaller than for an inspection with no coating but
sizing estimates can still be obtained by entering an estimate of the coating thickness in the QFMu sizing
procedure using the box labelled "Extra lift off". The look up tables presently implemented do not
support extra lift-offs greater than 3 to 5mm (dependent on probe type) - refer to TSC for sizing through
thicker coatings. Note that, because of the use of a unidirectional input field, the ACFM signals are
relatively insensitive to lift-off and so the thickness need only be estimated to the nearest millimetre.
For the same reason, the ACFM technique is relatively insensitive to changes in coating thickness (or
changes in lift off) and in any case, only the Bx signal is affected so no confusion with a possible defect
signal can occur. It should be borne in mind that the reduction in Bx and especially Bz amplitudes due to
coatings will result in smaller butterfly loops than otherwise expected for a given defect size. To
compensate for this, a new probe file should be made through a non-conducting layer of the same
thickness on a standard function check sample.

8.2 Magnetic State

It must be ensured that the surface being inspected is in an unmagnetised state. Therefore the
procedure followed with any previous magnetic technique deployed (particularly MPI inspection) must
include demagnetisation of the surface. This is because areas of remnant magnetisation, particularly
where the leg of an MPI yoke was sited, can produce loops in the butterfly plot which may be confused
with a defect signal. If magnetisation of the structure is unavoidable (e.g. d.c. magnets used to attach an
ROV), the magnetised area must be at least 500 mm away from the weld to be inspected

8.3 Surface Grinding or Work Hardening

It has been noted in a number of instances, especially on Duplex, that areas of surface grinding or work
hardening etc. which produce localised changes in material properties can result in strong signals in
both Bx and Bz which may be confused with a defect signal. Areas of grinding should be reported by the
probe operator at the time of scanning and recorded by the ACFM operator.

If a defect signal is suspected in a region of grinding or possible work hardening, further scans should be
taken parallel to but away from the weld toe. The signal from a defect (especially Bz) will drop off
quickly away from the defect, so that a scan away from the weld toe will be much flatter. The rate at
which the signal drops off will depend on the length and depth of the crack, but a significant difference
in amplitude should be seen 20mm away if the signal is due to a defect. If there is no significant change
in signal amplitude 20mm away from the weld toe, the signal is likely to be due to the effects of the
grinding. A further scan 40mm away should be made to confirm the trend. To assist later signal
interpretation, marker lines should be added on the display at the start and end of the ground area.

If an edge effect probe is available, and geometry allows its deployment, a scan with this probe will
produce a much flatter signal from grinding marks than from a defect.

8.4 Seam Welds

Seam welds running across the line of scanning also produce strong signals in Bx and Bz which can
sometimes be confused with a defect signal. As for grinding marks above, if a defect is suspected at or
near a seam weld, further scans should be made away from but parallel to the suspected defect line to
confirm the defect. The signal from the seam weld will persist at roughly the same amplitude, whereas a
defect signal will exhibit much reduced amplitude. Again, an edge effect probe, if available, can be used
to emphasise any defect signal.

9. Reporting Requirements
 ACFM results are backed up onto floppy disks, or other media, which contain the raw inspection data.
These will only be useful to the client if backed up with data sheets filled in correctly. This section deals
with the information required and production of these data sheets.

The data report sheets generated by ACFM inspections will be specifically designed with the ACFM
system and current inspection requirements in mind. The essential information contained on an ACFM
data sheet will include:

9.1 General Information

Date
Operators Name
Probe Operator
Component number/name
File Number
9.2 Scanning Data
Filename
Page Number
Position on Weld (i.e. Chord toe, Brace toe, Weld cap)
Probe Number
Probe Direction
Clock or Tape Positions
Inspection Summary
9.3 Detailed Record of Indications/Anomalies
Filename
Page Number
Position on Weld
Start of Defect (Tape reference)
End of Defect (Tape reference)
Length of Defect (in millimetres)
Depth of Defect (in millimetres)
Remarks
Diagram/Drawing of component under inspection.

10. Interpretation of ACFM Signals

10.1 General Method

In general, a defect will product a characteristic signal on Bx and Bz signals and these combine to give
the “butterfly". The rule adopted is that the signal represents a defect when

1. Bz responds by a peak and trough, and

2. Bx responds by a "dip" from the mean value.

In the majority of situations for, isolated and relatively short cracks, this will give a butterfly plot which
moves to right or left (according to probe direction), then downward then returns to the original
position to close the loop. Figure The figure below shows the relationship between Bx, Bz and the
butterfly trace where "A" represents the first Bz peak, "B" represents the deepest point in Bx, and "C"
represents the Bz trough.

Lift off, permeability changes, probe rocking etc. can cause responses from Bx and/or Bz. Lift example,
causes the Bx signal to dip or peak (depending on material type) with little response from the Bz signal.
This would produce a closed loop confined to the vertical axis rather than the open loop produced by a
crack. A seam weld, on the other hand, usually causes a peak in Bx combined with Bz signals that result
in an open loop moving upwards from the starting point. Hence, for the signal to represent a crack, the
butterfly loop must move downwards. This rule is intended to eliminate false calls due to momentary lift
off, perhaps
as the probe runs over local weld imperfections. It therefore is intended for small signals. There are
exceptions to the general rules and these mainly apply to long defects (> 50mm).

For long cracks the Bz peak and trough may be some way apart. This means that at the centre of the
crack there is no Bz signal and we rely on Bx (i.e. the dip in Bx) to make the butterfly go down. If there is
a general drift on the Bx signal, i.e. if the top signal is not perfectly level, this may confuse the butterfly
by acting against the crack signal. This conflicting Bx signal means the butterfly rules no longer can be
relied on and it is necessary to use the Bx and Bz plots to look for tell-tale signals where there is a
response from Bz followed by a downward deviation, from the trend, on Bx. This is particularly
important for tight angles where the Bx signal trend rises toward the tight angle due to global geometry
effects.

In theory it is possible for a weld to be cracked around the full circumference, thus resulting in no crack
ends. If the crack has a uniform depth, the Bx signals would be lower than expected, but in all other
aspects the signals could be similar to an uncracked connection. In this situation the presence of a signal
centred much lower on the butterfly plot than for other connections should be an indication that a full
circumference defect is present. This can be further investigated by comparing the Bx signals as the
probe is moved from parent plate to the weld toe area and then repeating the exercise on the other
weld toe.

In practice, cracks do not tend to grow this way in tubular connections. Experience has shown that full
circumferential cracking is normally associated with significant variations in crack depth around the
connection combined with crack branching. In this situation the crack branching provides Bz signals
resulting in butterfly loops in the normal way. Thus in practice detection of full circumferential cracks on
node connections is similar to normal cracks in that butterfly loops will normally be present together
with significant depth variation represented by dips in Bx and downward movement of the butterfly
trace.
Crack depth measurement is more complicated if full circumferential cracking is suspected because the
"length" and background Bx readings will not be easily established. Refer to TSC if assistance is needed.
 Figure .Schematic showing relationship between butterfly and chart recorder plots.

10.2 Decision Tree for Identification of a crack

At all times look at the Bx and Bz traces. If Bx has a dip then suspects a crack. If the butterfly makes
any significant loops, look at a broad area either side of the signal. This is particularly important if
the butterfly is moving up or down the screen. A butterfly moving up or down the screen with any
sort of looping is likely to be a long crack.

11. Crack Sizing

Crack sizing can be carried out either on-line (immediately after the necessary data has been collected) or offline
(by recalling stored data). During the detection stage, scans of Bx and Bz are recorded for all defects found. A first
estimate of length is also obtained from a manual measurement of the extent of the Bz signal.

This data is then used to obtain sizing estimates using the QFMu program as outlined below. For more details,
refer to the Level 1 Software User Manual.

1. Select a section of the Bx timebase plot to one side of the defect signal that is representative of the
background level. This should simultaneously be near the centre of the area filled by the general
offcrack background noise in the butterfly plot. This value is chosen as the Bx Background Level. N.B. If
the background level is significantly different either side of the defect, the average of the two values
should be used. This is normally necessary for defects in regions of changing geometry or near to a plate
edge (sees Inspecting Flat Plate Welded Supports/Stiffeners).
2. Select the minimum of the Bx timebase trace from the centre of the defect signal (or the bottom of the
butterfly loop). This value is chosen as the Bx Minimum Level.
 3. To calculate defect length and depth, the QFMu software requires three parameters - the two Bx values
selected above, and the length estimate obtained during the detection stage (the length in mm between
the Bz peak and trough indications at the inspection site).
4. Where the crack depth calculated is greater than the plate thickness this indicates that cracking may be
completely through the plate and so the probe operator should be asked to look for further evidence of
this. Such evidence may be visual crack opening or, if the back face is accessible (e.g. on flat plates),
crack-like indications on a scan made there.

Alternatively, because the currents follow the crack faces exactly in ferritic steel, depths apparently
greater than the wall thickness can arise from crack branching into the parent plate, or from a highly
curved crack path.

12. Length Sizing With ACFM

The method of length sizing involves determining the location of the Bz peak and trough signals on the
component. The distance between peak and trough is referred to as the Bz Length. The ACFM modelling
shows that the Bz peak/trough actually occur just inside the physical ends of the crack. This is confirmed
in practice when the crack ends are marked and compared visually. Thus the Bz lengths are always less
than the actual crack length. This is taken into account in the sizing algorithms so that the ACFM
calculated length is greater than the Bz length.

Figure. Length by ACFM

Because ACFM also considers the depth of the crack, ACFM will "concentrate" on those parts of a crack
with significant depth.

If two cracks are joined at the surface by a shallow surface crack, ACFM will focus on the two deep
parts, not the whole, and ACFM sizing procedures take this into account.