Compliance with specified material characteristics is an important part of quality
control when manufacturing high-quality industrial products. Even today, however, quality
control is still only based on random samples in a large number of cases. The relatively
large test effort required and the sometimes inevitable destruction of the test specimen
frequently makes it impossible to carry out 100% tests using classical methods, such as
determining the case hardening depth of surface-hardened components and the mechanical
strength of sheet steel.
The magneto-inductive method, on the other hand, is ideally suitable for automatic 100%
testing. Such characteristics as the surface hardness and case hardening depth can be
verified without destruction using this method. This not only ensures greater reliability
with regard to the quality of the individual products, but also guarantees that defects
can be identified and remedied during the production process since the test is integrated
into the production line. When testing hardened components, the test unit can be linked
to a control computer performing SPC analyses and thus permitting automatic control of
the hardening furnace.
Compared with classical materials testing, however, a new approach will be necessary
when using the magneto-inductive method, for it is not an absolute-value method and must
be calibrated with the aid of parameters which may have to be ascertained destructively.
Physical principles of magneto-inductive testing. The magneto-inductive
test method is based on an electromagnetic field built up by a field coil and modified
by the presence of a test piece with conductivity and permeability μ. This makes the
method suitable for all technological variables correlating with μ. The instrumentation
comprises either a field coil and a sensor coil embracing the test specimen or several
such coils of smaller diameter which, together with a highly permeable core, form what
is known as a probe (Fig. 1).
Figure 1. Two basic arrangment for magneto-inductive testing
In the first case, the specimen forms the core of this transformer-type arrangement; the
interaction between field and specimen is most effective when the inner volume of the
field coil has been filled completely. In the second case, the probe is placed on the
surface of the specimen in order to carry out the test. The field strengths produced in
the material in this way are smaller than those produced by the embracing coils. However,
the test area is pinpointed more precisely with this method and these probes can also be
used to test large parts, as well as parts with complex geometry.
Permeability characteristics are extremely informative parameters for a variety of
material states due, for instance, to hardening processes, particularly for materials
with a relative permeability of well over 1. The properties of a hysteresis loop create
additional effects in the sensor coil signal, its amplitude depending on the maximum
magnetization while the existence of coercive field strength alters the signal phase
relative to the exciting phase and the non-linearity of the hysteresis generates
harmonics 3f, 5f, 7f, etc.
The maximum information available can only be derived from the sensor signal if all
these effects are included in the analysis. In addition, however, the form of the
hysteresis depends on the test frequency f and on the field amplitude
in a manner characteristic of the material state. This makes it clear that considerably
more information about the material properties can be obtained by testing with several
frequencies and amplitudes. The result is a multi-parameter test. Since different test
frequencies also penetrate into the material to different depths, the test frequency
can be used to achieve a certain degree of depth selection. Integral and superficial
properties can be identified simultaneously by appropriately analyzing the results
obtained with several test frequencies.
Calibration based on representative parts with known properties is an essential
prerequisite for magneto-inductive testing, since the precise shape and unambiguous
relationship between the measuring signals and the required technological characteristic
are dependent on the momentary test task and therefore not immediately known. The
relationships are determined empirically using statistical methods (Fig. 2). Only the
relationships between the physical variables μ,... (and the
geometry within limits) and the magneto-inductive signal are predicted theoretically.
The effect of primary influencing variables on μ and ... on the one hand
and on technlogical characteristics on the other, however, cannot be predicted
quantitatively. For this reason, the relationship between technological
characteristics and the measuring signal can only be determined empirically in
the test task concerned. The sequence
of work for every test task is therefore as follows: setting the test parameters -
calibration - testing.
Figure 2. Schematic illustration of the relationships prevailing in magneto-inductive
Properties of a computer-controlled test system. The Magnatest is a
computer-controlled modular test unit with one or more test channels containing the
actual test electronics and a computer unit as controller.
The capabilities offered by this system include the following:
- 100% on-line testing of material identity and hardness
- Case-hardening depth, strength and many other material properties
- Integration into CAQ systems
- Quantitative determination of test values for statistical process control
- Storage of test settings and results.
Two different evaluation methods have been implemented in the Magnatest. Group analysis
is the classical method for sorting test specimens into two or more classes.
The second evaluation method is based on a regression algorithm and is particularly
suitable for test tasks involving quantifiable properties, i.e. properties which can be
described numerically. The fact that regression analysis permits greater differentiation
of the test results can be used to great advantage to control the production parameters.
Such a control loop, in which SPC methods may be applied, can only fulfill its intended
purpose - preventing production of reject parts - if it can also identify minute
deviations from the product quality.
This system can also be used for multi-parameter testing, i.e. combined testing with
several test settings. The system helps the user to find the optimum test parameters.
For single-parameter tests, up to 24 settings can initially be specified and the system
then calculates the reliability, the selectivity of group analysis and regression
correlation for each setting. The user can then select the required test setting. Up to
24 (individual) settings are also specified for multi-parameter tests. The combination
of test parameters yielding the greatest reliability for the test task is then calculated
from the 24 specified settings by a program running on a separate PC linked to the test
Applications. The magneto-inductive method can basically be used to
test all material properties associated with changes in conductivity or permeability,
such as alloy fluctuations, strength, core hardness, surface hardness, case-hardening
depth, heat treatment, residual austenite content, soft-spottiness, cementite accumulation
and surface decarburization.
However, each specific test task must be preceded by a test phase in which setting and
optimization aids are used to determine whether the test property is reflected clearly
enough in the magneto-inductive signal. Only a small number of parameters may vary in
addition to the property under investigation if the test is to prove successful.
Testing of the case-hardening depth is a typical application in which electromagnetic
methods are effectively the only non-destructive methods available. The mean deviation
for both methods equals 0.08 mm. Three sets of parameters were combined without optimizing
the selection. The deviation was reduced to 0.054 mm, by using an optimized test setting.
This shows that the reliability of this method can be improved
"economically", i.e. without increasing the number of measured quantities
and the associated disadvantages, such as longer test times.
On-line testing of sheet steel is a promising application that is currently still in the
trial stage. Four sensors are installed side by side so that the tensile strength of the
sheet can be tested over the full width as it is transported. Maintaining a constant
(small) distance between sheet and sensor is a factor of critical importance for tests on
moving sheets. The mechanically ascertained strength values have been plotted against the
results of magneto-inductive tests for a sample series. Two values were obtained for each
specimen in the magneto-inductive tests, for they permit testing of both the upper and the
lower sides. Any differences between the two sides of the sheet due to the rolling
process can be identified in this way.
Modern magneto-inductive test units can be used to determine technological material
characteristics thanks to the implementation of complex algorithms for evaluation. The
magneto-inductive method is predestined for use in automatic non-destructive 100% tests
integrated into the production line.
When using this method, it is important not to forget that it is not an absolute-value
method and that a great deal depends on the care taken during calibration. If this point
is noted, the accuracy and variability achieved with the new evaluation capabilities make
magneto-inductive testing an ideal complement or, substitute for the random tests based
on classical destructive or semi-destructive methods.