Online Inspection System Inclusion Detection System IDS

Whitepaper
Online Inspection System
Inclusion Detection System (IDS)
Internal Inclusions, Shell and
Optically Undetectable Surface Defects

CONTENTS
06
Measurement Results
01
Introduction
04
Development of
Inclusion Detection System
4.1 Sensor Technology
6.1 Laboratory Measurements
6.2 Operational Measurements
6.3 Calibration Coil
18 - 21
22 - 26
02
07 - 10
27
4.2 Structure of the Sensor Block
02
Task
4.3 Structure of the
Overall System
11 - 13
07
Summary
03 - 04
14 - 15
28 - 29
03
User Benefits
05
Technical
Production Design
05 - 06
16 - 17
2
3

01
Introduction
02
Task
The exact control of production processes in the manufacture of strip steel
and flat products is crucial for the production of high-quality products
with homogeneous properties over their complete length. For verification,
it is therefore necessary to determine various material parameters over
the complete length and not just selectively at the ends. Non-contact and
non-destructive testing methods are required in the running process.
In steel production and the subsequent casting process, inclusions get
into the melt during the metallurgical process. Some of these non-metallic
particles remain in the steel products produced. In the subsequent rolling
processes, these internal defects are not rolled out of the steel, but rather
take on an elongated shape due to the deformation of the material.
If these internal inclusions and shells do not appear as open defects on
the surface of the material, they cannot be detected with conventional
optical measuring systems.
The demands placed on strip steels as a
material are growing continuously due to
the ongoing development of the material
and processing methods. This also applies
to the degree of purity. In the cold strip
sector, even the smallest purity deficiencies
or inclusions have negative effects.
The relative magnetic permeability of such
inclusions is significantly lower than that
of the surrounding material. Based on this
physical fact, detection by magnetic flux
leakage is possible.
In this process, the tested material is
magnetised, whereby a homogeneous
magnetic field also results on its surface
if its structure is homogeneous. If there are
local areas in the material with significantly
lower permeability, caused by, for example,
non-metallic inclusions, cracks and indentations
on the surface, the magnetic
In concrete terms, material defects in the form of internal inclusions and
shell defects in manufacturing processes with high degrees of deformation
– such as deep drawing – lead to material fractures, an increased reject
and complaint rate, considerable disruptions to the production process,
and even cost-intensive damage to the tools.
Figure 1: Schematic illustration of the magnetic field profile in the case of an internal inclusion
02
INTRODUCTION
TASK
03

esistance increases at this point. A part of the magnetic flux is
thus forced to the surface of the material. The difference between
the relative permeability of the material and the surrounding air
results in pronounced magnetic refraction. The extent of the
stray field emerging from the material surface is therefore significantly
greater than the causal defect, which enables its detection.
Accordingly, the task included the technical and mechanical
development of a new on-line measuring system based on the
principle of magnetic leakage flux, which reliably detects and
visualises these internal inclusions and shell defects already
formed during the metallurgical process.
03
User Benefits
In accordance with these requirements,
the Inclusion Detection System (IDS) from
IMS Messsysteme GmbH, which is unique
on the market, detects these defects continuously
during the ongoing production process
without contact and non-destructively
for strip thicknesses of 0.1 to 1 mm. This
system is therefore particularly interesting for
manufacturers of tin plates and thin sheets
metal who, for example, have to guarantee
the highest quality standards for the automotive
or food industries.
In contrast to an optical surface inspection
system, the Inclusion Detection System
(IDS) – as already described – is also able
to detect defects lying completely within
the material for which there are no discernible
signs on the material surface.
However, the technology of magnetic
leakage flux applied here offers a further,
convincing advantage:
Surface defects such as cracks or roll
impressions that are too small in size or
insufficient in contrast for detection by
optical surface inspection are also magnetically
visible due to material stresses or
interruptions in the microstructure.
Some surface defects, such as shells and
scrap marks, often have significantly larger
extensions than are optically detectable on
the material surface. In contrast to an optical
surface inspection, the Inclusion Detection
System (IDS) detects the full extent of such
defects, as outlined in the following figure:
04 TASK
USER BENEFITS 05

Safely detectable area
Conditionally detectable area
Not detectable area
Optical Surface Inspection System
Inclusion Detection System (IDS)
using magnetic flux leakage principle
04
Development of an
Inclusion Detection System (IDS)
4.1 Sensor Technology
Based on this, the optical measurements of surface defects on
sample sheets presented in the remainder of the white paper are
assumed to be representative of internal and surface defects.
When operating an Inclusion Detection System (IDS) and a
conventional surface inspection system together on (non-ferromagnetic)
coated material, the IDS enables reliable discrimination
between coating and substrate defects because, unlike an SIS,
it only displays defects in the ferromagnetic substrate.
The flux leakage detection system
developed for the detection of internal
inclusions offers the functionality of a
complete flux leakage test during ongoing
production operation. Electromagnets
are used to magnetise the material. Their
power can be adapted to the properties,
structure and geometry of the material
being inspected and can be switched off
for maintenance and cleaning purposes.
The stray fields are detected by means of
GMR sensors. The sensors used are GMR
differential sensors (gradiometers). They
consist of 4 individual GMR sensors that
are connected in the form of a Wheatstone
measuring bridge. Two of these sensors
each are linked spatially as shown in Figure
2. A differential signal is formed in dependence
on the difference in magnetic field
strength between the two sensitive areas.
The sensitive areas of the sensors used
have a centre-to-centre distance of 1 mm.
V0
Exit 2
Exit 1
0
Figure 2: GMR differential sensor
06 USER BENEFITS
DEVELOPMENT OF AN INCLUSION DETECTION SYSTEM | SENSOR TECHNOLOGY 07

The use of gradiometers enables a significantly
higher amplification compared
to absolute sensors (magnetometers) as
external fields have no influence on the
sensor signal. This means that particularly
small local magnetic field inhomogeneities
can be detected.
In the IDS measuring system, the material
is magnetised transversely to rolling direction.
This direction of magnetisation was chosen
on the basis of laboratory measurements
with artificial defects. These defects were
defined as through holes of different
diameters as well as grooves with a length
of 1 mm, a width of 100 µm and variable
depth. The internal inclusions occurring in
cold-rolled strip normally take on elongated
shapes due to the strong deformation.
Grooves, therefore, correspond most
closely to the defects that actually occur.
For holes, which correspond to compact
defects, approximately equivalent results
were obtained with all magnetisation
directions. For grooves, significantly better
signal-to-noise ratios were obtained with
magnetisation transverse to the rolling
direction. This can easily be explained by
the larger defect cross-section in this direction.
Figure 3 and 4 show the signal-to-noise
ratios for a through hole with a diameter
of 100 µm and a groove with a depth of
25 µm, a width of 100 µm and a length
of 1 mm.
- Magnetization
- Magnetization
Figure 4: SNR of a 100 µm through hole at different magnetisation directions and measuring distances
Magnetisation at a 45° angle to rolling magnetic field homogeneity, the yoke width
direction is not a compromise, as this was chosen so that a 48 mm wide area is
worsens the signal-to-noise ratio for measured in each case.
compact defects compared to parallel
magnetisation, without improving the Consequently, multiple magnets are necessary
in order to cover the different material
signal-to-noise ratio for elongated defects.
widths. Since no measurement is possible
With magnetisation transverse to the rolling in the area of the pole shoes of the
direction, the maximum yoke width of the magnets, the sensor modules are arranged
magnet is limited. Due to a suitable in two rows for continuous coverage.
Figure 3: SNR of a groove 25 µm deep, 100 µm wide and 1 mm long at different magnetisation directions and measuring distances
08
DEVELOPMENT OF AN INCLUSION DETECTION SYSTEM | SENSOR TECHNOLOGY
DEVELOPMENT OF AN INCLUSION DETECTION SYSTEM | SENSOR TECHNOLOGY
09

Figure 5 shows the dependence of the
signal-to-noise ratio on the distance
between the sensor and strip surface.
Starting from a minimum air gap of 200
µm, the ratio decreases almost linearly
by about 1 dB / 100 µm additional air gap as
the distance increases. In practice, an air gap
of 700 µm is sufficient for this application and
resolution.
04
4.2 Structure of the Sensor Block
One magnet and the sensor line inside
each magnet were combined to form a
compact sensor module. This guarantees
easy maintenance, repair and scalability
of the measuring system.
quick-release clamps in conjunction with
fitting screws. Mechanical adjustment is
not necessary thanks to the precise manufacture
and assembly of the individual
module components.
Figure 6 shows an opened module with its
key components. A module contains the
sensors, the amplification and filtering of
the sensor signals, AD converters as well
as the control of the electromagnet and
stabilised voltage supplies. The sensor
modules are installed quickly and reproducibly
on the measuring system using
The GMR sensors are encapsulated into a
sensor block within a protective and stabilising
aluminium frame. This hinders adhesion
of dirt and provides protection against
mechanical damage. The sensor modules
comply with protection class IP 64, which
enables direct use in harsh environments.
Figure 5: Signal-to-noise ratio of grooves and holes in dependence on the measuring distance
Figure 6: Sensor module
10 DEVELOPMENT OF AN INCLUSION DETECTION SYSTEM | SENSOR TECHNOLOGY
DEVELOPMENT OF AN INCLUSION DETECTION SYSTEM | STRUCTURE OF THE SENSOR BLOCK 11

Figure 7: Sensor block open
Figure 8: Sensor block encapsulated
Figure 10: Signal amplitude of a 1 mm x 100 µm x 30 µm groove depending on
its position.
For compact small defects, for example holes, the stray field is optimal for detection
with this sensor arrangement.
Figure 9: Replacement of sensor block
Compared to smaller sensors with higher spatial resolution, the sensors used are more
sensitive. This means that particularly weak signals of small defects can be detected
better. The defects detected by a sensor are determined via the signal amplitude.
The modular design ensures easy replacement
of the sensor block without
adjustment work.
A sensor module has an outer width of
95 mm, which allows two sensor module
rows to cover the material without gaps.
Each sensor module contains 48 GMR
differential sensors in the centre of the
magnet. The distance between the sensors
transversely to rolling direction is 1 mm.
This is advantageous as the size of the
stray fields of the smallest defects is also
above 1 mm. Figure 10 shows the signal
amplitude of a 1 mm long, 100 µm wide
and 30 µ deep groove.
Figure 11: Signal amplitude of different defects depending on their volume
12 DEVELOPMENT OF AN INCLUSION DETECTION SYSTEM | STRUCTURE OF THE SENSOR BLOCK
DEVELOPMENT OF AN INCLUSION DETECTION SYSTEM | STRUCTURE OF THE SENSOR BLOCK
13

04
4.3 Structure of the Overall System
The overall system is structured hierarchically. The individual levels work on a task basis
and are interconnected via fast network technology. The sensor signals are converted
from analogue to digital with a sampling rate of up to 187.5 kHz at a resolution of 15 bits.
Length-dependent scanning takes place with a constant longitudinal resolution (rolling
direction) of 0.1 mm.
The digitised sensor signals from up to eight sensor modules are fed to a common
Gig-E hub and converted to the Gig-E camera standard.
Figure 12: Hardware structure
The Gig-E hubs are connected to a camera computer. This camera computer has
the following tasks:
1. Signal pre-processing
2. Detection of defects
3. Feature computation
4. Classification
5. Control and adjustment of the sensor modules
The database server is superimposed on this camera computer. The database server
stores the defect images and contains the production and training database. Visualisation
of the defects as well as the connection to the customer database is effected via the
database server.
Figure 13: Signal processing
Figure 13 shows the signal processing with the detection, classification and
post-processor functions.
14 DEVELOPMENT OF AN INCLUSION DETECTION SYSTEM | STRUCTURE OF THE OVERALL SYSTEM
DEVELOPMENT OF AN INCLUSION DETECTION SYSTEM | STRUCTURE OF THE OVERALL SYSTEM
15

05
Technical Production Design
Product data:
Strip thickness
0.1 – max. 1 mm
Strip width
customizable
Strip speed
max. 1,000 m/min. at full resolution
Measuring distance > 0.5 mm
Measuring accuracy:
Detectable defect size Hole: 70 µm diameter; groove: 10 µm deep, 100 µm wide,
(substitute defect) 1,000 µm long in strip 250 µm thick.
The minimum defect size must be determined depending
on the application.
Distance influence 1 dB / 100 µm
Reproducibility >98%
The measurement takes place on a deflection roller with two rows of sensor modules.
The sensor modules can be adjusted to different material thicknesses with the help of
servo motors.
To ensure mechanical stability, the sensor
module rows and their carriers are kept at
a constant temperature by means of water.
The water cooling also serves to dissipate
the waste heat of the sensor modules.
The measuring system also has a pneumatic
drive to move the sensor rows on and off
the strip surface completely. The drive is
activated automatically for quick retraction
of the measuring system from the strip
surface if there is a risk of collision with the
material being measured.
The distance of the sensor modules to the
strip surface is monitored permanently by
three capacitive distance sensors per sensor
module row. As an additional safety device,
the measuring system has an optical
wrinkle detector. This is a laser light barrier
which is installed in the strip run 10-20 m
in front of the measuring point. In the case
of wrinkles, the sensor rows are swivelled
away.
The measuring system can be moved to
park position outside the system. In this
position, the sensors are adjusted automatically
and maintenance work can be
carried out while the line is running. During
the adjustment, all sensors are standardised
to a defined magnetic sensitivity and defective
sensors are detected.
Figure 14: Inclusion Detection System in operation
16 TECHNICAL PRODUCTION DESIGN
TECHNICAL PRODUCTION DESIGN
17

06
Surface defects on the opposite side of the material to the measuring system can also be
detected. Figure 17 shows the raw signal of grooves measuring 1 mm x 100 µm x 25 µm
in 200 µm thick steel strip. The first groove is on the side of the material facing the measuring
system, the second on the opposite side.
Measurement Results
6.1 Laboratory Measurements
The following figures show the raw signals of artificial defects: a hole with a diameter of
100 µm (Figure 15) and a surface groove measuring 1 mm x 100 µm x 30 µm in 200 µm
thick steel strip at a measuring distance of 500 µm and a speed of 500 m/min.
Figure 18: Groove signal profile, top and bottom strip side
Signal curve of a 100 µm
through hole
Signal curve of a 30 µm deep,
100 µm wide and 1 mm long groove
As can be seen in Figure 19, the speed has only little influence on the signal-to-noise
ratio of defect signals.
Signal to noise ratio at variable material speed,
100 μm through hole
Signal curve of a 10 µm deep,
100 µm wide and 1 mm long groove
Speed
Figure 19: Signal profile at different speeds
Figures 15 – 17:
Signal profiles of a 100 μm hole and 10 μm and 30 μm deep,
100 μm wide and 1 mm long groove
18 MEASUREMENT RESULTS | LABORATORY MEASUREMENTS
MEASUREMENT RESULTS | LABORATORY MEASUREMENTS
19

The following figures show the signal characteristics of grooves with a width of 100 µm
and a depth of 25 µm and 50 µm respectively, as well as a hole with a diameter of 100 µm
in 800 µm. As can be seen from the 50 µm groove, the sensitivity for defects located on
the underside of the material is reduced at this material thickness, but larger defects can
still be detected on both sides.
The following figures show examples of cold strip defects visible visually without the aid
of a magnetic particle inspection, with the associated IDS signals:
Figure 20:
Signal characteristic of a
25 µm deep and 100 µm
wide groove on the upper
side of the material
Figure 24:
Small shell, material depth 700 µm
Figure 21:
Signal characteristic of
a 50 µm deep and 100
µm wide groove on the
upper side of the
material
Figure 25:
Signal characteristic, small shell
Figure 22:
Signal characteristic
of a 50 µm deep and
100 µm wide groove on
the under side of the
material
Figure 26:
Rolling and coating defect,
material depth 660 µm
Figure 23:
Signal characteristic of
a 100 µm hole
Figure 27:
Signal characteristic, rolling
and coating defect
20 MEASUREMENT RESULTS | LABORATORY MEASUREMENTS MEASUREMENT RESULTS | LABORATORY MEASUREMENTS 21

06
6.2 Operational Measurements
The internal inclusions measured are evaluated as follows:
1. Based on defect size (amplitude/volume)
2. Based on classification
e.g. shells, cracks, overlaps, M-defects, scratches, indentations
Figure 29: Large defects such as scratches and wrinkles
Some inclusions detected were located precisely by means of magnetic
particle inspection and examined subsequently by making 3 sections each
transverse to the rolling direction.
Figure 28: Compact defects, various representations
Figure 30: IDS image of defect 1 Figure 31: Magnetic particle image of defect 1
22
MEASUREMENT RESULTS | OPERATIONAL MEASUREMENTS
MEASUREMENT RESULTS | OPERATIONAL MEASUREMENTS
23

Figure 37: 1st section, defect 2 Figure 38: 2nd section, defect 2 Figure 39: 3rd section, defect 2
Figure 32: 1st section, defect 1 Figure 33: 2nd section, defect 1 Figure 34: 3rd section, defect 1
Figure 35: IDS image of defect 2 Figure 36: Magnetic particle image of defect 2
Figure 40: IDS image of defect 3
Figure 41: Magnetic particle image of defect 3
24 MEASUREMENT RESULTS | OPERATIONAL MEASUREMENTS
MEASUREMENT RESULTS | OPERATIONAL MEASUREMENTS
25

06
6.3 Calibration Coil
The complete measuring system, including the classification
performance, is checked cyclically by means of a
calibration coil. The calibration coil should contain as
many defects as possible. Cold strip from the production
of transitional slabs is suitable for this. A reproducibility
of > 98% was achieved during the cyclical check.
Figure 42: 1st Section, defect 3
Figure 43: 2nd section, defect 3 Figure 44: 3rd section, defect 3
26 MEASUREMENT RESULTS | OPERATIONAL MEASUREMENTS
MEASUREMENT RESULTS | CALIBRATION COIL
27

07
Summary
Using high-quality measuring electronics and
advanced image processing, a high-resolution
measuring system for internal defects
and external material damage has been
developed.
The IDS measuring system is used for complete
evaluation of purity. The maximum strip
thickness at full sensitivity is 1 mm and is
scalable to any strip width. The measuring
system can be adapted to customer specifications.
The image processing with feature computation
and classification used distinguishes the
defects based on their size and type. The
classification is adapted to the respective
material and the customer specification.
The IDS measuring system avoids the delivery
of defective material and ensures perfect
product quality for the end customer. In addition,
the measurement results are used to optimise
the pre-material stages. By improving
quality and output, resources are conserved
and costs reduced.
28 SUMMARY
SUMMARY 29

IMS Messsysteme GmbH
Dieselstraße 55
42579 Heiligenhaus | Germany
Phone: +49 (0) 2056 / 975 – 0
Fax: +49 (0) 2056 / 975 – 140
Mail: info@ims-gmbh.de
www.ims-gmbh.de | www.ims-experts.com | www.imsocial.info