Peter Mudge, BSc, CEng, MIM, MWeldI, MInstNDT, is Head of the NDT Research Department.
A recent research programme carried out by The Welding Institute on behalf of a group of sponsors studied the factors controlling detection of defects in welds by magnetic particle inspection (MPI). Limiting conditions were quantified so that recommendations could be made for improved testing practice. Some of the findings are presented in this month's cover story.
The factors influencing the performance attainable by MPI are considerably more complex than is apparent from the straightforward way in which the tests can be carried out. To perform reliable and consistent MPI, it is essential that the test conditions, e.g. surface preparation and magnetic medium, are appropriate and that inspection parameters are correctly selected. When designing a test, it is important to consider the limitations which may be imposed, either inherently by the nature of the testpiece or arising from the prevailing test conditions.
Optimum test conditions are inevitably a compromise between attaining sufficient sensitivity to detect all the flaws of interest and the test being over-sensitive, resulting in a large number of 'false calls' from geometrical and surface features. Nevertheless, the following general statements apply:
- For a given magnetic medium (i.e. ink or powder type, manufacture, concentration, etc) there will be a threshold value of leakage field required to attract the particles to the local leakage site and hold them there. Consequently, for a given applied field strength or more correctly, flux density, in a particular test there will be a minimum size of flaw which will produce a discernible indication.
- For a given workpiece surface finish there will be a minimum flaw size which will produce indications that can be distinguished from those produced by the background, i.e. from roughness, weld ripple, weld toe geometry, and so on.
- One of the above factors will limit the sensitivity achievable for the detection of flaws in a magnetic particle test. However, the detection of flaws by MPI is statistical in nature, and the fact that the size of a flaw exceeds the minimum threshold imposed by 1 or 2 above does not guarantee that the resulting indication will be reported. Detection is size dependent and, in general, the likelihood of detection increases with flaw size.
These statements were the subject of the investigation reported here.
Test parameters
To detect flaws using MPI certain minimum conditions have to be met. These are:
- Generation of sufficient leakage field at the flaw in the magnetised zone to cause magnetic particles to migrate to the flaw from the surrounding area.
- Sufficient magnetic particles must be present in the vicinity of the flaw to provide a visible indication under the influence of the leakage field.
- Lighting conditions must be appropriate to allow the indication so formed to be seen.
- Interpretation by the operator is required to distinguish indications from flaws and those from the background, and hence avoid false calls.
The parameters which affect the generation of an adequate level of leakage field from a flaw are:
- - Applied field strength;
- - Relative permeability of the material and local variations of it;
- - Direction of magnetisation relative to the flaw;
- - Workpiece geometry in the vicinity of the flaw;
- - Size, shape and position of the flaw.
Mechanism of detection by MPI
Magnetic particles will migrate to a defect under the influence of a local field gradient, at or just above the surface, produced by the tangential component of the leakage field at the flaw. The force to move them is provided by their own magnetic response to this field. The nature of this response is complex and beyond the scope of this article. It is the source of the leakage field and the effects of changes in its magnitude which have been studied, assuming the characteristics of the magnetic particles to be a constant for each case.
Flux density
Leakage fields are produced as a result of the inability of flawed regions to sustain the same level of magnetisation as the surrounding material. To ensure continuity of the lines of magnetic flux some of the field must leak out of the surface so producing leakage field. The magnitude of a leakage field is determined by the size of the discontinuity, but is principally dependent on the flux density, B, which exists in that region. B is related to the applied field, H, by the expression:
B = µoµrH . . . [1]
where:
H = magnetic field strength, ampere/metre (A/m);
B = magnetic flux density, tesla (T);
µo = permeability of free space, 4Π x 10-7 H/m;
µr = relative permeability, (dimensionless).
The relationship between B and H is non-linear, and can be represented by a hysteresis loop (Fig.1) for a typical steel. This also indicates that µr is not constant with H, so not only is the value of µr dependent on the specific properties of the material under test (µo being a physical constant) but it varies significantly with the strength of the applied field (Fig.2).
Relative permeability
The relative permeability of steels varies widely, being governed largely by composition, and both thermal and mechanical history. Permeabilities for constructional steels are not generally known because they are difficult to measure. BS 6072 (see reference) overcomes this problem by assuming a value of 240. A considerable number of measurements of permeability was made during this project using a specially developed miniature closed loop sample technique which enabled the permeability of parent material, weld metal and heat affected zone (HAZ) material to be determined for a variety of steels. These results are shown in histogram form in Fig.3 and 4. All lie above 240, which indicates that BS 6072 is conservative in this respect. The results in Fig.3 and 4 were taken at measurement conditions to give maximum permeability values and at 2400 A/m respectively. The overall trends in the values obtained follow closely the behaviour of the example shown in Fig.2.
Limiting conditions for detection of surface features
Surface roughness,
µm Ra | Flux density level B, T |
| Black ink | Fluorescent ink |
| AC* | DC | AC* | DC |
| 0.4 | > 1.80 | 1.62 | 1.69 | 1.59 |
| 0.6 | > 1.80 | 1.55 | > 1.80 | 1.54 |
| 1.4 | 1.53 | 1.48 | 1.46 | 1.46 |
| 7.8 | 1.37 | 1.43 | 0.88 | 1.34 |
| *adjusted to compensate for skin effect |
Detection of defects
Minimum flux density
Practical measurements indicated that there was an approximately linear relationship between the depth of a fatigue crack and the local flux density required to produce a sufficiently large leakage field for it to be detected, for the range of cracks studied. The values were comparable for both AC and DC magnetisation, provided a correction was made for the skin effect with AC. The results are shown in Fig.5 and 6 for tests conducted with AC and DC magnetisation respectively. These also show that fluorescent inks are generally more sensitive. The figures indicate that by using a sufficiently high flux density, flaws of small depth can reliably be detected. However, sensitivity is limited by the propensity for high levels of magnetisation to reveal surface features of the workpiece. This is demonstrated by the results given in the Table, which are flux density levels that produced discernible indications from surfaces of four different degrees of roughness. Consequently, it was considered that a sensible recommendation for maximum flux density for MPI in general would be 1.1T.
Comparison with BS 6072 requirements
To make the above discussion relevant to testing practice the means of achieving the appropriate level of flux density has to be considered in terms of applied field, H, rather than flux density B. Further, it is necessary to put this into context with respect to current practice, as defined by BS 6072. Two points have to be examined; the value of µr for the material under test and how the distribution of the magnetic field between the poles of the magnet, or input points of the magnetising current, is defined.
BS 6072 states that a flux density equal to or exceeding 0.72T is required for inspection, and that the permeability of steel is 240 or above. Consequently, the field strength required for current flow magnetisation is 2400 A/m. Further, the standard assumes that the minimum level of flux density is achieved at the periphery of a circle inscribed between the pole pieces of a yoke, or between the prods if current flow is used.
For practical testing, the value of H is important because it can be measured or calculated as a test parameter. The higher the value of µr, the less the value of H required to achieve a given level of flux density in the workpiece. For example, if it is assumed that a steel has a relative permeability of 240, then the magnetising force required to give 1.1T is calculated by rearranging [I]:
In practice, however, the actual flux density achieved will be considerably higher than the minimum of 0.72T, even if a field strength of 2400 A/m is used because:
- The relative permeability of the steel will nearly always exceed 240. A value of approximately 366 is required to give a flux density of 1.1T at the periphery of the circle at a magnetisation of 2400 A/m. However, most values measured exceed 400 as shown in Fig.3 and 4, giving even higher flux densities, so that values of H less than 2400 A/m can be used effectively, provided that a level of B of 1.1T can be maintained.
- The flux density within the circle will be greater than that at the edge. If 1.1T is being exceeded on the circumference, flux density levels could approach magnetic saturation (approximately 2T) towards the centre.
This suggests that by following the recommended applied field strengths given in BS 6072, the proposed optimal value of flux density of 1.1T will be achieved at the periphery of the inspection zone in most cases. Nevertheless, following the recommendations of the standard can lead to overmagnetisation of the workpiece.
- When large pole separations are used with either permanent magnets or yokes, the field can become concentrated around the pole pieces, resulting in a generally low level of magnetisation in the area to be examined.
Discussion
The British Standard compensates for the fact that the permeability of a given steel is rarely known by assuming a conservatively low value of µr, such that the value of B generated will always exceed the minimum, if the recommended H is applied. Similarly, the inspection zone is defined in such a way that the majority of inspections will be carried out in a region which is more highly magnetised than predicted, i.e. within the circle. The availability of more extensive information about the permeabilities of steels and thresholds for detection of flaws has enabled a more appropriate level of B to be defined and the value of H required to achieve it can be determined. The test conditions recommended in BS 6072 do not, in general, differ widely from those proposed, but the more quantitative approach should enable the occasional uncontrolled variations in test conditions, caused by assumptions inherent in the standard, to be avoided.
It can be seen from Fig.5 and 6 that specifying a flux density of 1.1T is likely to limit the detection of shallow flaws. Using current practice it is known that flaws less than the predicted threshold of between 0.75 and 1mm in depth can be found by MPI, although high local flux densities are required. A flux density of 1.1T was chosen as an optimum because above this value it becomes increasingly difficult to distinguish between indications from flaws and those produced at irregularities on the surface, on all but the best prepared surfaces, as shown from the results in the Table. If surface preparation is carried out to produce a smoothly contoured surface with a roughness better than 7.8µm Ra then higher levels of B can be used effectively. On the other hand, at high flux densities there can be a sufficient concentration of magnetic flux at geometrical discontinuities, even at blended weld toes, to produce a leakage field strong enough to cause an indication in the absence of a defect.
So-called 'spurious indications' can also occur even when a surface has been machined flat. This is thought to be caused by variations in permeability between parent plate, weld metal and the HAZ. Such variations can be seen by comparing the distribution of permeabilities for the different regions in Fig.3 and 4, those of the HAZ generally being lower in each case. If the permeability drops significantly across the HAZ, then a leakage flux will be formed, which can be strong enough to produce an indication.
More rigorous verification of test conditions for MPI is limited by lack of appropriate instrumentation. It is possible to estimate the input field strength when using current flow prods or looped conductors, but for magnetic flow methods this is not feasible. Hall effect probes can be used to measure field strength, but these are only sensitive to the flux just above the workpiece surface and not within it.
Similarly, segment and foil type indicators only provide information about the field just above the surface of the workpiece, and are exceedingly difficult to calibrate in terms of actual magnetisation level.
The above techniques all measure field strength and not flux density. Flux density is a much more difficult quantity to measure and commercial instruments to measure it are not widely available. Flux density can be inferred from measurements of field strength but knowledge of the material permeability is required. It is not generally possible to measure permeability outside the laboratory.
Summary
The results of a research programme have indicated a linear relationship between flux density and the minimum size of flaw which can be detected. From this an optimum flux density level for detection of defects by MPI of 1.1T has been proposed. In practice this does not differ widely from the value generally expected to be achieved when using the recommendations of BS 6072, when taking materials' magnetic properties into account. This is because measured relative permeabilities of the steels studied lay in the range of 400-1000 instead of around 240, as assumed by the standard.
Higher levels of flux density tend to be inappropriate because of the likelihood of delineation of surface features, except on specially prepared surfaces. Modelling of the magnetic field distribution also showed that at high magnetic flux density levels concentrations of leakage flux at geometry changes or variations in permeability can cause false indications on the absence of defects.
Quantitative determination of the value of test parameters for field testing is difficult owing to the lack of suitable instrumentation.
Acknowledgements
The author wishes to thank the sponsoring organisations below for their support during the research programme and for permission to publish this paper.
Sponsoring organisations: Babcock Power, British Coal, British Gas, British Petroleum, Britoil, Brown and Root (UK), Central Electricity Generating Board, Conoco (UK), Department of Energy, Framatome, GEC, NOVA, an Alberta Corporation, Oilfield Inspection Services, Shell (UK).
Reference
BS 6072 : 1981 'Method for magnetic particle flaw detection'. British Standards Institution, London.
