Magnetic Analysis of Products

Ya. G. Dorfman and K. V. Grigorov, Leningrad.

In the manufacture of steel and iron products it is often very important to know the degree of homogeneity of these products and the presence in them of various kinds of flaws (cracks, blowholes, inclusions, etc.).

For this purpose, the so-called magnetic analysis has recently begun to be applied. In many cases this method has undoubted advantages in comparison with X-ray and other methods.

Among such advantages should be counted its comparative speed, cheapness, and simplicity.

The purpose of the present article is to clarify both the essence of this method and the possibility of its practical application.

The possibility of magnetic inspection of products is based on the fact that between the magnetic properties of the specimen under investigation and its other physical properties there exists a close and, moreover, unambiguous connection.

A specimen may be inhomogeneous in its chemical composition, in its geometrical form, or in the degree of continuity of the material; finally, different parts of the specimen, while chemically identical, may possess different physical properties depending on local elastic stresses, reorientation of crystallites, etc. Detection of these inhomogeneities is possible by magnetic means if they manifest themselves in the magnetic properties.

§ 1. The Influence of Nonuniformity on the Magnetic Field of a Specimen.

Let us consider, in the roughest outline,¹ the distortion of the field of a cylindrical specimen when a nonuniformity is present in it.

Let a specimen in the form of a very long cylindrical rod \(AB\) (Fig. 1) be magnetized by a uniform magnetic field of strength \(H\). The direction of the field is parallel to the axis of the specimen. Denote the cross-sectional area of the specimen by \(S\), and its permeability at the given value of the field by \(\mu\). In that case the specimen will be penetrated by a magnetic flux equal to:

Fig. 1.

\[ \Phi = \mu H S. \tag{1} \]

If the specimen is perfectly homogeneous along its entire length, then the magnetic flux \(\Phi\) is also the same along the entire length of the specimen (neglecting the influence of the ends).

But let us now suppose that inside the specimen there is a sufficiently sharply bounded region \(D\), differing in its magnetic properties from the rest of the mass of the specimen.

For simplicity, let us assume that this inclusion also has the form of a thin cylinder, whose length, however, is less than the length of the specimen. The axis of this cylinder is parallel to the axis of the rod \(AB\).

Let the cross-sectional area of this inclusion be \(S_1\), and its permeability at the same value of the field \(H\)

¹ A rigorous treatment of this case presents considerable difficulties.

let it be \(\mu_1\). Obviously, \(S_1<S\). Under these conditions the inclusion \(D\) will be penetrated by a flux \(\Phi_1\), equal to:

\[ \Phi_1=\mu_1 H S_1. \tag{2} \]

The total resultant flux at that place in the specimen where the inclusion \(D\) is located will be equal to:

\[ \Phi'=\mu H(S-S_1)+\mu_1 H S_1=\mu H S+(\mu_1-\mu)H S_1 \]

or, in other words:

\[ \Phi'=\Phi+\Delta_1\Phi, \tag{3} \]

where

\[ \Delta_1\Phi=(\mu_1-\mu)H S_1. \tag{4} \]

Thus, to the flux \(\Phi\) penetrating the specimen in its homogeneous part, there is added a new magnetic flux \(\Delta_1\Phi\), caused by the given inclusion.

The lines of flux \(\Delta_1\Phi\) may close either entirely inside the specimen \(AB\), or partly inside and partly outside it.

Closing inside the specimen, this force flux creates an additional magnetizing field. Along with this, the permeability of the specimen \(AB\) also changes. As a result of all this, the magnetic flux in the specimen \(AB\) near the inclusion \(D\) changes. This new, secondary change in the magnetic flux is opposite in sign to the change \(\Delta_1\Phi\).

If this secondary change (let us call it \(\Delta_2\Phi\)) is equal to \(\Delta_1\Phi\), then the total magnetic flux penetrating the whole specimen \(AB\), together with the inclusion \(D\), will not change. If, however, \(\Delta_2\Phi\ne\Delta_1\Phi\), then the magnetic flux \(\Phi_D\), penetrating the specimen \(AB\) at the location of the inclusion \(D\), will differ from the flux \(\Phi\) by some amount depending both on \(\Delta_1\Phi\) and on the magnetizing field \(H\) and the permeability of the specimen \(\mu\). In the case where \(\Delta_1\Phi>0\), i.e. \(\mu_1>\mu\), the flux \(\Phi_D\) near the inclusion is greater than the flux \(\Phi\); if, however, \(\mu_1<\mu\) and \(\Delta_1\Phi<0\), then \(\Phi_D<\Phi\). The difference \(\Phi-\Phi_D=\Delta\Phi\) may thus be either positive or negative. In some cases it may be equal to zero. Finally, it is possible to imagine such cases as well, when this difference is exactly equal to \(\Delta_1\Phi\).

The latter is possible, for example, when the specimen \(AB\) has been magnetized to saturation. In this case, as the magnetizing field is increased, the magnetization in the specimen does not change, and consequently the flux \(\Phi\) will change precisely by the amount \(\Delta_1\Phi\), for \(\Delta_2\Phi = 0\). One may think that the same will also occur in other cases when the permeability changes only slightly with a change of the field \(H\), for example when \(\mu\) is equal to a maximum.

The final change of flux, or the difference

\[ \Delta\Phi = \Phi - \Phi_D \]

is localized near the inclusion \(D\).

The lines of force of the flux \(\Delta\Phi\) are lines which do not close inside the specimen; consequently, they can close only outside it.

Thus, if \(\Phi_D \ne \Phi\), then in the space surrounding the specimen \(AB\), near the location of the inhomogeneity \(D\), lines of force of the flux \(\Delta\Phi\) appear. The magnetic field present in the space outside the specimen (in our case, for a very long cylinder, this is simply the magnetizing field \(H\)) and homogeneous in those places where the specimen is homogeneous ceases to be homogeneous in the place where the specimen has the inclusion \(D\). To the field \(H\) there is added the field \(\Delta\Phi\).

Thus, as a result of the presence in the specimen \(AB\) of a region with a magnetic permeability different from the permeability of the whole specimen, generally speaking, the field changes both inside the specimen itself and outside it. This change is localized near the place where the inhomogeneity is located.

The complexity of the dependence of \(\mu\) and \(\mu_1\) on the field (depending on the material of the specimen and of the inclusion) does not allow us to compute theoretically in advance the relation between \(\Delta\Phi\) and the character and dimensions of the inhomogeneity.

Let us see how this change can be detected. We shall denote the beginning of the inhomogeneous part of the external field by \(a\) (Fig. 1), measuring along the length of the specimen from \(A\) to \(B\), and the end by \(b\). In the remaining places along the whole specimen, i.e. from \(A\) to \(a\) and from \(b\) to \(B\), we shall suppose the external field and the magnetization of the specimen to be perfectly homogeneous.

Place on the specimen a coil \(K\) closely fitting it, so that the axes of the specimen and of the coil coincide.

We shall now move this coil along the specimen from \(A\) to \(B\) with uniform velocity. While the coil is moving along the homogeneous part of the specimen, between \(A\) and \(a\), it is at all times pierced by one and the same flux \(\Phi\). But near \(a\) this flux changes and becomes equal to \(\Phi_D\). It remains so over some length between \(a\) and \(b\), depending on the length of the inhomogeneity \(D\). The coil is again pierced by one and the same flux \(\Phi_D\). Further, when this coil approaches the end of the inhomogeneity, i.e. the point \(b\), the magnetic flux again changes and returns back to \(\Phi\). Thus the magnetic flux piercing the coil changes in magnitude twice, and these changes are opposite in sign. As a result of each such change, a certain electromotive force is induced in the coil, and the quantity of electricity passing through the coil and the galvanometer is:

\[ \vartheta=\frac{n\Delta\Phi}{R}, \tag{5} \]

where \(n\) is the number of turns of the coil, \(R\) is the resistance of the coil and of the galvanometer closing it.

Converting to practical units, we have:

\[ \Delta\Phi=\frac{R}{n}\,\vartheta\cdot 10^8. \tag{6} \]

Let us see how accurately a change in magnetic flux can be detected.

Let the number of turns of the coil be \(n=200\). The resistance of the circuit is \(R=500\,\Omega\), and the sensitivity of the ballistic galvanometer is \(10^{-8}\ \mathrm{C/mm}\). Substituting in (6), we have:

\[ \Delta\Phi=\frac{500}{200}a, \tag{7} \]

where \(a\) is the deflection of the galvanometer in mm of the scale.

Thus, when \(\Phi\) changes by several maxwells, the presence of inhomogeneity can be detected by a sensitive galvanometer.

Such a quantity of electricity will pass through the galvanometer twice: once when the coil \(K\) passes through \(a\), and a second time in the opposite direction when the coil passes through \(b\). Let our inclusion \(D\) be a cavity whose permeability is approximately equal to unity.

Let, for example, the cross-section of the specimen be \(1\ \mathrm{cm}^2\), the induction about \(25{,}000\), \(\Phi = 25{,}000\); then we can detect the presence of a cavity whose transverse cross-section is about \(0.1\ \mathrm{mm}^2\).

The length \(l\) of such a cavity or blowhole, if the ratio of \(l\) to \(S_1\) is greater than 50, does not affect the magnitude of the change in flux. This change (in absolute magnitude) depends only on the change in the transverse cross-section of the specimen. If the specimen has not been magnetized to saturation, the phenomenon will already be more complicated, and it is impossible to calculate the change in flux so simply.

Further, we have so far assumed that our inclusion has the form of a cylinder parallel to the axis of the specimen \(AB\) and to the direction of the field \(H\). In reality, the inhomogeneity may have any form. For detecting an inhomogeneity it is important only that it cause a change in flux.

Having explained how it is possible to detect inhomogeneities in specimens, we shall turn to the consideration of certain experimental investigations. At the same time the possibility of the practical application of the method of magnetic analysis will become clear.

§ 2. Inhomogeneities in cylindrical specimens.

As we have already mentioned above, inclusions and defects in a specimen may differ in their nature.

An inhomogeneity in a specimen may be either of a geometrical character, i.e., there may be a change in the transverse cross-section, or of a physical character—this may be a change in the structure of the specimen, in its physical properties. An example of this is furnished by Fig. 21. In this figure several curves are given, representing charac-

characteristics of cylindrical specimens in which there were various kinds of inhomogeneities. The diameter of the specimens was 12.7 mm. In the drawings the length of the specimens is plotted along the abscissa axis; the change in magnetic flux along the ordinate axis. The magnetization of the specimens was close to saturation.

In their initial state all the specimens were quite homogeneous. The characteristic of their magnetic properties was a straight line (Fig. 2, curve A).

Fig. 2.

Then the homogeneity of the specimens was disturbed. Individual specimens were subjected to the action of various factors. The change in homogeneity was local. As a result, the magnitude of the magnetic flux changed at the place where a given inhomogeneity was concentrated. In specimen B, a notch 3 mm deep was made. Here there was mainly a decrease in the area of the cross-section, as a result of which a rather sharp jump appeared in the characteristic, indicating a change in the magnetic flux at the location of the inhomogeneity.

The characteristics C and D refer to two other specimens. Here we see the same jumps as in the case of curve B. But their cause is different. Namely, specimen C was bent through an angle of 10° and then straightened again. Specimen D was compressed in one place in small vices.

As a result of this, residual stresses appeared in both specimens C and D, perhaps insignificant in magnitude

their magnitude, but which, nevertheless, affect the magnetic properties. And, what is most interesting, the change in the magnetic properties, so far as can be seen from these curves, is of the same order of magnitude as in the case of specimen \(B\), where the change in the cross-sectional area was quite considerable. Thus we already see that insignificant mechanical deformations may produce the same effect as a substantial change in the geometrical dimensions of the specimen.

Next, curves \(E\) and \(F\) refer to specimens whose homogeneity was disturbed thermally. Namely: specimen \(E\) was heated in a small flame, and then the heated place was slowly cooled in air, so that at this point the material was annealed. Specimen \(F\) was heated over a short length and hardened by rapid cooling.

In these cases as well, as may be seen from the curves presented, the changes in magnetic flux are, in magnitude, approximately the same as in the other specimens. Only in specimen \(E\) (the annealed one) does the change have the opposite sign from that in all the others.

Finally, curve \(G\) refers to a specimen that was cut in two; then the cut surfaces were well polished and pressed tightly against each other. At the place where the parts of the specimen were joined, owing to the resulting air gap, however negligible, the permeability of the specimen changed sharply. As a result, a very abrupt jump was obtained on the characteristic.

Thus, from this comparison we see that inhomogeneities differing in their nature may be reflected in exactly the same way in the magnetic properties of the specimen.

Not only the magnitude, but even the character of the change may be the same, as is evident from curves \(B\), \(C\), and \(D\). This circumstance, of course, introduces serious difficulties in interpreting the results of the analysis. Later we shall see by what means this difficulty can be overcome.

Let us now dwell on the investigation of the influence of the dimensions and form of the inhomogeneity on the magnetic properties. Such

the investigation was carried out by Sanford1 on specimens of steel rifle barrels.

The method of investigation was outlined in general terms in the preceding paragraph. A specimen about 200 cm long was fastened vertically in such a way that its lower and upper ends were clamped by iron clamps. These latter were connected to one another by three vertical iron columns. In this way the magnetic circuit of the specimen was made closed. The magnetization of the specimen was produced by a solenoid about 40 cm long.

The solenoid, placed on the specimen, could, by means of a motor, move up and down along the specimen, magnetizing it successively over its entire length.

Inside the solenoid there was placed a system of indicating coils, which also moved together with it. This system consisted of three coils. They were completely identical with one another, both geometrically and in the number of turns. The length of each of these coils was of the order of several centimeters. On each of them 500 turns of thin wire were wound. Two of these indicating coils were connected to one another in opposition. Thus changes in the magnetic field caused by accidental fluctuations of the magnetizing current did not affect the galvanometer. These two coils were at a distance of 10 cm from one another and were placed in the middle of the solenoid. Let us now see what happens when an inhomogeneity is present in the specimen. So long as both coils moved along a homogeneous part of the specimen, no electromotive force was induced in them. On approaching that place of the specimen where there is some inhomogeneity, the magnetic field changes. The first to sense this change is the front coil, since, when it approaches the inhomogeneity, the rear one is still located in a homogeneous part of the specimen.

Thus an electromotive force is induced at first only in the front coil. The galvanometer indi—

causes a certain deflection; then, when the front coil has passed the nonuniform part and begins to move away from it, an electromotive force of the opposite direction is induced in it. Then the rear coil approaches the nonuniform part of the specimen. It now gives a deflection in the galvanometer, but of the opposite sign to the first deflection caused by the front coil. When the rear coil leaves the place where the nonuniformity is concentrated, a new impulse is obtained in the galvanometer, opposite to the preceding one.

Fig. 3.

Between these two coils there was a third, serving only for control. At the place of the nonuniformity it produced two successive deflections, opposite in sign.

The characteristic of each specimen was taken twice. The first time with a simple coil—the upper curves in the drawings shown; the second time with the astatized system—the lower curves in the same drawings.

The magnetization of the specimens was such that the induction amounted to about 15,000 gauss. The deflections of the galvanometer were recorded photographically.

To create nonuniformity in the specimens, thin strips of transformer iron were used. These strips were placed along the specimen and pressed against it. The results of the investigations are presented in Figs. 3–6. The position and shape of the iron strips are shown in the same

in the drawings, corresponding to how they were placed on the specimens. The lengths of these strips were as follows: in the case of Fig. 3—25 mm; Fig. 4—600 mm; Fig. 5—300 mm and Fig. 6—again 600 mm.

Fig. 4.

The width (in the specimens of Figs. 5 and 6, the width at the middle of the strip) of all the strips was 19 mm.

In these experiments, in addition to changing the geometrical dimensions of the specimen—increasing the cross-sectional area—

Fig. 5.

we also have a change in permeability, since the strips of transformer iron have a permeability different from that of the steel specimens.

From the drawings presented it is evident, above all, that the localization of the inhomogeneity is determined fairly accurately both by the simple coil and by the double one.

MAGNETIC ANALYSIS OF PRODUCTS

Even in those cases (Figs. 5 and 6) where the uniformities of the specimen change not abruptly but gradually, this change is easily detected. Further, from a comparison of the upper and lower curves, i.e. characteristics relating to one and the same kind of nonuniformity but obtained by different methods, it is evident that the simple, single coil proves to be more sensitive. This is especially clearly seen from the comparison of the upper and lower characteristics in Fig. 6.

The difference in the characteristics of simple and astatic coils is especially noticeable in Fig. 4, where below four deviations, completely separated from one another, are clearly visible. The same thing, only less sharply, is also seen in Fig. 6.

Fig. 6.

In both of these cases the nonuniformity along the length was much greater than the dimensions of the indicator coils (in the astatic system) and the distance between them. In the case when the length of the nonuniformity is small, the difference in the characteristics obtained by both methods is smaller, as is seen from Fig. 5, and it is already quite small in the case of the very small nonuniformity presented in Fig. 3.

The deviation of the characteristic from rectilinearity at its end or at its beginning is due to the influence of the ends of the specimen. It may also be added that the detection of local nonuniformities becomes possible in principle only when these nonuniformities are situated comparatively far from one another; otherwise the effects are superposed and confuse the picture.

Sanford1 investigated steel cables and wires. The method of investigation did not differ in essence from that,

which was used for studying gun barrels. The difference consisted only in the fact that in this investigation the magnetic circuit of the specimen was not closed, and the wire was simply drawn through a system consisting of a magnetizing coil and an indicating coil placed inside it.

The specimens studied had already been in service for some time, and no defects had been found in them with respect to their mechanical properties.

Here, however, there is no longer any question at all of any rectilinearity of the characteristic. Evidently, the specimen was so inhomogeneous that its magnetic—

Fig. 7.

properties changed from point to point. These changes could have been due not only to the internal, but also to the purely surface inhomogeneity of the specimen. Obviously, one cannot expect a wire that has been in service to the same extent to have a smooth, clean surface. Furthermore, it is quite probable that in such a wire there could have been various residual stresses, deformations, etc. All this was reflected in the magnetic characteristic.

In Fig. 7 two characteristics are given, obtained from one and the same specimen, but at different values of the magnetizing field. Namely: the upper curve was obtained with a magnetizing field of 20 gauss, the lower one—with a magnetizing field of 100 gauss. The properties of the specimen had previously been changed at several points. Specifically: at the point corresponding to \(A\) in the drawing, the wire had been slightly filed; at point \(B\) it had been bent at an angle

in \(90^\circ\) and then straightened again, and, finally, at point \(C\) the wire was heated by the flame of a burning match. Thus, on one and the same specimen we have three kinds of inhomogeneity known to us in advance. On the upper curve of Fig. 7 it is almost impossible to notice the presence of these artificial inhomogeneities.

On the lower curve, taken with a stronger magnetizing field, the initial inhomogeneities are imperceptible, and then the inhomogeneities artificially produced at points \(A\), \(B\), \(C\) stand out clearly. Let us now examine these curves more attentively. We shall notice that on the lower curve all three factors, so different in their nature, have affected the characteristic in exactly the same way.

The influence of local inhomogeneities on the upper curve is quite different. Here the effect of overstraining at \(A\) is quite imperceptible. Conversely, the influence of residual stresses produced by bending the wire, at point \(B\), is much greater than on the lower characteristic.

This can be understood on the basis of the fact that, for the steel used in the manufacture of such cables, the influence of mechanical stresses on the magnetic properties is much greater at low magnetizing fields than at high ones. Probably from this point of view one may also understand the fact that the lower characteristic is more nearly rectilinear than the upper one: in a strong field, inhomogeneities caused by elastic stresses did not affect the magnetic properties, and magnetic analysis could not detect them. Finally, the jump of the characteristic at point \(C\) on the upper characteristic is directed in the opposite direction from that on the lower curve.

This is explained by the fact that, in a weak field, the change in permeability caused by local annealing has a different sign than in a strong field. Thus, from comparison of these two curves one can see that the influence of inhomogeneity of the specimen on the magnetic characteristic depends to a very great extent on the magnitude of the magnetizing field. This means that, in general, the results of magnetic inspection can depend very strongly on the magnitude of the magnetizing field. On the other hand, examining a specimen at various—

personal fields, we can ascertain the nature of the inhomogeneities present in the specimen. However, for this it is necessary to know comprehensively the magnetic properties of the given material, their change under cold-working, under annealing, etc. The results of investigating the traces also indicate to us the direction in which the study of the specimen should be conducted if it is desired to establish the presence of inhomogeneities of some particular type. We shall return to this question later.

Several more examples may be given of the application of the method with a coil moving along the specimen. Namely, by exactly the same method as that described above, Burrows1 attempted to investigate railway rails. Again the magnetizing coil moved automatically along the specimen. Inside the magnetizing solenoid an indicator coil was placed. The coil was given such a shape that it closely embraced the rail. The magnetic flux was closed by another, auxiliary rail. This second rail was placed parallel to the first and was connected with it by means of thick steel strips, which were clamped to the ends of the rails.

In the investigation, the rails were magnetized approximately to saturation. Before the investigation the rails had already been in service for a considerable time. During the investigation inhomogeneities were artificially produced in the rails. Some results of the investigation are given in Fig. 8.

In the upper half of the drawing, at the points denoted by the letter \(A\) with different subscripts, one can see the influence of defects of a geometrical character. Namely: holes of various diameters were drilled in the wall of the rails. The hole corresponding to \(A_1\) had a diameter of \(12.7\) mm; the one corresponding to \(A_2\), \(6.3\) mm in diameter. From the drawing it is immediately evident that the influence of the defect \(A_2\) on the characteristic is extremely negligible and, in comparison with fluctuations of the curve due to other, undetermined causes, is quite imperceptible. The influence of the hole at \(A_1\) is already more sharply expressed, but it too is of the same order as the fluctuations of the curve due to other, unknown

reasons. Further, on the same curve, to which \(A_1\) also belongs, we see a sharp jump at point \(B\).

This jump is likewise due to an artificial defect. Namely: the rail was cut transversely at the corresponding place. The width of the cut was \(1\ \mathrm{mm}\). The depth was such that, at the place of the cut, the cross-sectional area of the rail decreased by \(10\%\). This slit was then tightly filled with transformer iron, which has high magnetic permeability. Here we have, on the one hand, an increase in magnetic resistance owing to the air gap that appeared between the rail and the iron shim, since to some extent a gap remains here, despite the filling of the slot with iron, no matter how tightly the shim enters the cut. On the other hand, there is a decrease in resistance caused by the material of the shim itself. In this case it is impossible to separate these two factors. It is therefore difficult to say what, in fact, causes the jump in the characteristic. Further, at the very beginning of the characteristic, on the left, we see a sharp jump, the cause of which remained unexplained. In general, this entire characteristic here is extremely confused. It may be assumed that either the internal structure of the rail was non-uniform, or these constant deviations of the characteristic—

Fig. 8.

cracks are due to the influence of surface irregularities. However, other causes may also be present here. Thus, in the same Fig. 8, below, the characteristic of a rail is given which had been in service for a very long time. The periodic character of the curve immediately catches the eye. In the opinion of the author of the study, this periodicity is due to the influence of the rail fastenings in the sleepers. The magnitude of the period of the curve corresponds exactly to the distance between the sleepers on which the rail was laid. The peaks in the characteristic correspond to those parts of the rail that were located between the sleepers. The depressions correspond to the parts that were above the sleepers.

Incidentally, as can be seen, the top of peak \(C\) is cut off at a certain height. The rail at the place corresponding to this peak was carefully examined, and it turned out that this effect, in all probability, was caused by inclusions of finely crushed pieces of steel. These inclusions had a much greater mechanical hardness than the material of the rail itself.

On the basis of all this, the author draws the very plausible conclusion that, at the places where the rail was fastened to the sleepers, the rail had greater rigidity, and therefore its permeability at these places was less. This decrease in permeability is what caused the falling-off of the characteristics in those parts of the rail that were located above the sleepers.

Besides Burrows, rails were studied by Dudley1 in America and Suzuki2 in Japan. Dudley’s research results, even if they deserve interest, do so perhaps only from the point of view of their unreliability and unsatisfactoriness. The characteristic of a rail which he gives in his note is a continuous series of oscillations and jumps, on the basis of which it is utterly impossible to judge anything about the object of the investigation.

Suzuki’s method differs somewhat from the methods of Sanford and the others discussed above. In this

in this method the rail closed the magnetic circuit of a large magnet. The electromagnet, on small rollers, was placed vertically on the rail in such a way that its poles rested against the head of the rail. Between the poles and the surface of the rail head there was a gap of 0.5 mm.

Thanks to the rollers the electromagnet could move freely along the rail. Between the poles of the electromagnet an indicating coil was placed, which enclosed only the head of the rail. The shape of the indicating coil is shown in Fig. 9. In this way it was possible to examine rails laid on a railway track. For this purpose the electromagnet with the indicating coil was attached to a special trolley so that its flux was closed by the rail itself and, together with the trolley, moved along the rail. On the same trolley were mounted the recording instrument and the storage batteries feeding the electromagnet.

Fig. 9.

Strictly speaking, the chief interest of this work consists precisely in the described modification of the installation. The plotting of characteristics was carried out in a very primitive manner. They were traced by hand, by means of a special device, on a paper tape wound on a rotating drum; only the strongest deflections of the recording instrument were marked.

In this way only large inhomogeneities, which over their area extended over the entire (or almost the entire) section of the rail head, or even of the whole rail, could be detected.

Suzuki believes that his method makes it possible to detect damage to rails in the track. This assertion seems to us very doubtful, since here inhomogeneities of different nature (which, consequently, had different effects on the strength of the rail) were marked on the characteristics in approximately the same way. Many of them could be completely harmless to the rail and, conversely,

it is quite possible that harmful defects remained unnoticed.

Let us now turn to the consideration of several other methods of magnetic analysis.

In all the methods considered, the indicator coils were arranged so that the planes of their turns were perpendicular to the axis of the specimen. These indicator coils took into account only that component of the magnetic field which was parallel to the axis of the specimen. But, after all, in the case of the presence of inhomogeneity in the specimen, that component of the field which is directed perpendicular to its axis also changes.

Fig. 10.

More precisely, if, in the case of a very long specimen, homogeneous along its entire length and uniformly magnetized parallel to its length, the normal component of the induction and of the external field is equal to zero, then, in the presence of inhomogeneity, this normal component of the field is no longer equal to zero. Thus, from the change in this normal component of the external field of the specimen one can judge the presence of inhomogeneity in the specimen.

In Fig. 10 are shown schematically the two arrangements, a and b, proposed by Voigt¹). Here \(A\) is the specimen under investigation, in the form either of: a) a straight cylinder, or b) a ring of circular cross-section. \(S\) is an indicator coil wound on an iron core. This coil is nothing other than a magnetic potentiometer.

¹) Voigt. See: Anwers Phys. Zeitschr. No. 24, p. 871, 1927.

Until the specimen is homogeneous, the intensity (its radial component) of the magnetic field at both ends of the potentiometer is the same (equal to zero). In the case, however, of the presence of an inhomogeneity, the situation will be approximately the same as in the case of an astatic system of coils: between the ends of the potentiometer there will appear a difference in the intensities of the magnetic field. When the potentiometer is moved along the specimen, this difference in intensities changes at the site of the inhomogeneity, and an electric

Fig. 11.

current is induced in the potentiometer, which is indicated by the recording instrument. It is, of course, possible to keep the potentiometer stationary and to move the specimen along its axis. It is also obvious that the potentiometer need not have an iron core, but then it is less sensitive. It is important only that the bases of the potentiometer be perpendicular to the normal component of the magnetic field.

In all the arrangements listed up to now the specimen is magnetized by a constant field. But one may also use an alternating field; then the indicator coils or the potentiometer will operate as the secondary winding of a transformer. It is necessary only that a homogeneous specimen not produce a current in this winding. Several schemes have been proposed that make it possible to investigate by means of an alternating field.

One such apparatus was designed by Neiffel­dom and Kuhnke1 and is shown schematically in Fig. 11. Here \(A\) is the specimen under investigation, which is magnetized by an alternating current passing through winding \(B\). The turns of the indicator coil \(S\) encircle the specimen. The surfaces of the turns of the indicator coil are parallel to the surface of the specimen and fit closely against the latter.

§ 3. Application of the magnetic method to the investigation of disks.

In the present section we shall become acquainted with a method of applying magnetic analysis when the object of investigation was steel turbine disks of uniform thickness. Such disks were investigated by Capp2.

The disk under investigation was rotated about a vertical axis in such a way that part of it passed between the poles of an electromagnet and thus closed the magnetic circuit.

Fig. 12.

The arrangement of the apparatus is presented in Fig. 12. The designations here are: \(A\)—the winding of the electromagnet poles; \(B\)—indicator coils, \(C\)—the disk under investigation, \(D\)—the yoke of the electromagnet.

The magnetic circuit thus consisted of the poles, the yoke, the air gaps between the poles and the disk, and, finally, that part of the disk which was located between the poles. The electromagnet was mounted on a special frame in such a way that it could be moved along the radius of the disk from the rim to the center and back. In this way separate annular sections of the disk were successively obtained. The disks had surfaces—

...were carefully treated. As long as the permeability of the ring passing between the poles remained constant (as the disk rotated), the total magnetic flux in the circuit did not change. When a zone of the disk containing an inhomogeneity entered the space between the poles, the resistance of the magnetic circuit changed, and therefore the magnetic flux penetrating the circuit also changed. In the indicator coils $B$ placed at the ends of the poles, a certain electromotive force was induced, which was also registered by the galvanometer connected to the coils. The same occurred when the inhomogeneity left the space between the poles (only with the opposite sign). Thus the galvanometer indicated the passage between the poles of one or another inclusion by a more or less sharp deflection. The disk under investigation is a conducting body rotating in a magnetic field; therefore Foucault currents are induced in the disk. In order to eliminate the action of these currents on the galvanometer, the indicator coils were made in the form of an astatic system.

Fig. 13.

The shape of the coils and their connections are shown schematically in Fig. 13. The area of the middle coil (1) is exactly equal to the sum of the areas of the two outer coils (2). The number of turns in all three coils is the same.

The direction of the turns of the middle coil is opposite to the direction of the turns of the two outer coils. This system of coils was slipped onto the poles, in which corresponding slots had been made. Specifically: the base of the cylindrical pole (the one nearest the disk) was divided into three parts by two parallel secants so that the area of the middle part was equal to the sum of the areas of the two outer segments. The coils were put onto the poles in such a way that coil (1) enclosed the middle part of the bases of the poles...

s, while the external coils (2) encompassed the external segments of the pole pieces.

The action of the Foucault currents on the recording galvanometer is therefore reduced to zero. There were two such galvanometers in Kapp’s installation. One of them served for visual observation of sharp deflections. The other galvanometer had photographic recording. The characteristic curves were recorded in the form of circles, the radius of which, on the corresponding scale, was equal to the radius of the ring of the disk under investigation. The recording system was arranged so that the deflections of the galvanometer used for visual reading were 15–20 times greater than the deflections of the galvanometer used for photographic recording. The sensitivity of the method was such that, according to the author, small scratches on the surface of the disk could be marked on the diagram. Small cracks, almost indistinguishable to the eye, which sometimes appear along the edges of the disk as a result of its hardening, could also be marked directly by the deflection of the galvanometer. If the disk had not been machined sufficiently cleanly, then irregularities of the surface or changes in thickness, even very slight ones, showed up on the diagram.

Fig. 14.

When disks are cast, cavities and slag inclusions sometimes arise in them; later, when the disk is forged, these may turn into a narrow void, with a thickness of several tenths or even hundredths of a millimeter. Such voids, under direct observation, gave ...

a deflection of the galvanometer of the order of 10–15 cm, whereas the constant oscillations of the same galvanometer, owing to insufficiently careful machining of the disk—for example, nonuniformity in thickness—were of the order of 3–4 cm.

Slag inclusions in disks, consisting of very small grains whose linear dimensions are of the order of tenths and hundredths of a millimeter, are often concentrated within a limited space inside the disk, with a fairly uniform density, in peculiar colonies. Such colonies can also be recorded photographically on the diagram of the disk. They are revealed by the fact that the characteristic curves of the disk, which in the case of a completely homogeneous specimen form a series of concentric even circles (see Fig. 14), in the corresponding places acquire a wavy character and thereby differ sharply from the curves characterizing a pure specimen.

Fig. 15.

Figure 15 gives the characteristic of a disk containing such slag inclusions. The field of the electromagnet was sufficiently large, so that the region of the disk was brought almost to saturation. Owing to this, apparently, the inhomogeneities caused by local elastic stresses remained unnoticed. Several hundred disks were examined. Production inspection of the disks was carried out not photographically, but visually. In this procedure, small oscillations of the galvanometer, of the order of 3–4 cm on the scale, were not taken into account. Only those specimens in whose inspection deviations on the scale of about 10 cm and more were obtained were considered defective. These disks were then studied more

in detail, and there was not a single case in which the defect that caused the deflection of the galvanometer proved to be imaginary.

§ 4. Conclusion.

On the basis of the material presented, it may be seen that the sensitivity of the method can be brought to a very high degree, and the most insignificant defects of a specimen can be determined with sufficient accuracy. This property of the method is both its positive quality and its shortcoming, for, being sensitive to very insignificant flaws, it is at the same time equally sensitive to insignificant defects of the surface itself, to fluctuations in the thickness of the specimen, and so forth, which in many cases are of no essential significance from the point of view of the suitability of the object under investigation. In addition, slight fluctuations of the air gap between the specimen and the indicator, especially in methods with a magnetic potentiometer, can, in the investigation of disks, produce a considerable effect in the indicator system. It follows from this that in many cases the threshold of sensitivity should be limited in advance, or only deflections above some definite magnitude should be taken into account. This limiting magnitude is different in each individual case, depending on the object of investigation and on the requirements imposed on the investigation. In accordance with the sensitivity of the method, the localization of the inhomogeneity is also determined quite accurately.

Further, one of the essential difficulties that must be encountered in applying the magnetic method is the analysis of the obtained results of the investigation. Already at the very beginning of § 2 we encountered the fact that inhomogeneities different in their nature have exactly the same effect on the characteristic of the specimen. On the basis of this characteristic alone, it is very often (if not always) simply impossible to draw any conclusions about the nature of one or another inhomogeneity. Meanwhile, such inhomogeneities, depending on their nature, may have an entirely different influence on the quali-

…properties of one or another specimen. One inhomogeneity may render a specimen completely unsuitable, while another may have no substantial influence on the quality of the product. In such cases, when it is not enough merely to determine the presence of an inhomogeneity, but it is also necessary to know its nature more or less precisely, additional investigations by other methods must be carried out, which may considerably complicate the whole work and nullify all the advantages of magnetic analysis. Here it will not be superfluous to draw attention once again to the following circumstance: as was mentioned in § 1 and as could be seen in Fig. 7, inhomogeneities that differ in their nature behave differently, depending on the strength of the magnetizing field. It is possible that one of the methods which will make it possible to obtain an answer to the question of the nature of an inhomogeneity is precisely the method of repeated investigation of the specimen at different values of the magnetizing field. But apart from the fact that the very question of the connection between the nature of inclusions and the magnitude of the magnetizing field has not been developed at all, it is quite obvious that such repeated investigation may considerably complicate the methodology itself.

But if we do not speak for the moment of such a complication, it must be acknowledged that the magnetic method of inspecting products, in comparison with many other methods, is simple; the investigation can be carried out much more quickly, and, moreover, it is relatively inexpensive.

It is necessary, however, to bear in mind that magnetic analysis is still very poorly developed. The question of the connection between magnetic properties and those local changes which are possible in specimens and which may play an essential role in the selection of products is still far from being clarified. Each type of product requires special study.

For this reason, it seems to us premature to speak of the applicability of this method to industry. Despite the extremely numerous articles about it in the specialized and popular literature, the application of magnetic analysis in the West apparently remains, up to now, extremely rare.

1. Kuhnke. See Auwers: loc. cit. ↩↩↩↩↩↩

2. Capp. Proc. Amer. Soc. for Test. Mat., V. 27, p. 268, 1927. ↩↩