US2244606A - Rail flaw detection - Google Patents

Description

June 3, 1941. J, H BlGELw Y 2,244,606 i l RAIL FLAw DBTECTI'ON Filed sept. 10, 1938 F153- F"|G.2.

ma 5,72 L -)TL A k Y E; l Plas.

man. f j\/ N l j. F'iGJZ. C'

INVENTOR JUL/AN H. else/.ow

` ATTORNEY Patented June 3, imi

.trillian H. Bigelow, Endicott, hl. Y., assigner to Sperry Products, Inc., Hoboken, lil. i., a corpo-= ration ci New York Application September iii, w38, Seriali No. 229.2%

(Ci. ITE-193) d iliainns.

This invention relates to the art of rail Haw detection, and particularly to the type ci rail Haw detection employed in the Sperry rail flaw detector car which operates on the principle of energizing a portion of the rail locallywith current as the car travels along the tracks, thus establishing a local Velectromagnetic field surround-1 ing the rail and detecting variations in thisiield causedl by the presence of fissures. Such detec tion hasr heretofore been accomplished by means oi a plurality of pairs oi.' similar induction coils carried bythe car at a fixed distance above the rail head so that all coils normally cut the same number of lines of magnetic flux: but in the presence of a defect. and hence o i' a concurrent distortion in the electromagnetic iield, the individual coils cut a different number of lines, gen erating a resultant M. F. which is suitably amplified by a'thermionic amplifier, thereafter operating a plurality ,oi indicators such as a re cordin'g pen on a chart withinthe car and paint spraying device for marking the spot on the rail where the defect occurs.

A study of the characteristics of the voltage surge generated in the type oi detector unit heretofore employed by the presence of a defect disclosed the fact that by the'construction of said arrangements of detector coils in accordance with my invention.

Figs. 9 to 12 inclusive are diagrams illustrating the theory of flux impulse generation by a plurality of pairs of coils and illustrating the theory of my invention for the case of a two-pair detector unit.

Recent `investigationscarried on to determine the nature of the voltage impulses generated within the induction coils which are used as the detector means in rail iissure detection, have indicated that the fluir distortion in the vicinity of transverse and compound rail defects has a peculiar and characteristic formation. Plotted on Aan exaggerated scale, the flux intensity along a 1 line parallel to the longitudinal axis of the rail unit it sometimes occurred that the several voltage cycles generated in the individual coils of the unit tended to nulliiy each other totally or partially' and the resultant outputs sometimes were of such characteristics as to maire amplification diiiicult or impossible. The present invention has for its principal object, therefore, the provision oi a method and means for securing a maximum resultant output from the detector unit without danger of nullification of the various impulses generated in the individual coils of the unit, and in a manner such that the said resultant output impulses may be readily amplified.

Further objects and advantages ci this invention `'will Ibecome apparent in the following detailed description.

In the accompanying drawing,

Fig. 1 is a side elevation of a portion of rail, indicating diagrammatically the presence of a fissure.

Figs. 2 to 5 inclusive are diagrams illustrating the theory of` generation of voltage impulses in the detector coils.

Fig. 6 is a plan view of a portion of railhead disclosing a form o! iiux distribution occurring 'in ,the presence of a iissure oi double polarity.

and located at a fixed distance above the head oi?- the rail, has been found to vary in either of two ways orin a combination of these ways. Thus.. if 'a flaw F is present'in the `rail R, as shown inFlg. 1, the ux intensity may take the form of the plot in Fig. 2 or the plot in Fig. 3, that is to say, there may be an increase 'or decrease in the ux intensity determined by the nature of the aw. Along a given line over the head of the rail either phenomenon ri or B may occur; and under certain circumstances phenomenon A may occur along one line parallel to the head of the rail while phenomenon B occurs on another such line elsewhere. This is illustrated graphically in Fig. 6 which is a view looking down on the head of the rail from above. It sometimes Occurs that on one side of the head (the gage side, for instance) phenomenon A occurs, the diminution in iiux being indicated by declination curvature, While along another path phenomenon B occurs, the amplification in flux being indicated by inclination curvature. These curvatures reach their maximum or minimum, as the case may be, directly opposite or over the derect.

Let us consider rst a hun distortion of either type ii or type B and assume that but a single induction coil is .being passed through the electhe region ci distortion. and then resumes its normal value, it is evident that there must be two which is negative (downward), there is a point f at which the flux intensity is itself either a maximum or minimum, and at this point c the flux plot is parallel to the rail axis and has zero slope.

From the above discussion it can be seen that the voltage plot is of the form shown on Fig. 4. It starts fromA a normal of zero, ascends to a positive maximum, in the case of phenomenon A, or descends to a negative minimum, in the case of phenomenon B, then returns through zero, ascends to a maximum of opposite sign, and then falls to zero again. This has also been verified experimentally. For a single detector coil, therefore, it is apparent that there are two critical inflection points in the flux plot; that closed in Fig` 2 or type B disclosed in Fig. 3; and

that the distances between the critical inflection points a and b have been found to have a deflnite mean value for all transverse defects and that in no instance did this critical spacing distance vary greatly from the mean value.

l In all detector units that have ever been employed on the Sperry rail fissure detector cars, it has been necessary to use not a single detector coil but at least one pair of such coils, the coils of the pair being connected in series opposition. This condition is, of course, imposed by the necessity of eliminating the effects of total flux variations such as are caused by excitation current variations and by unavoidable transverse and vertical motions of the search unit relative to the railhead. Under such conditions the flux variations affect the coils of the pair equally and oppositely, so that there is no differential effect upon the pair of coils. It was discovered at an early date that when these pairs of coils were separated by the predetermined critical distance, an external phenomenon of wave addition took place. Thus, for instance, referring to Fig. 5 which illustrates a pair of coils I and 2 connected in series opposition and passing over a region of ux distortion which may be of the type A or the type B, there results two voltage curves which are indicated in Fig, 5 as a, c, b and a', c', b. As coil l passes over the region of fiux distortion it generates a voltage peak in one direction at point a, and then a voltage peak b in the other direction. As the second coil 2 passes over the same region it generates first a peak a" C," which consists rst in a normal voltage peak corresponding to peak a, then an opposite polarity voltage peak of twice normal value as the first coil passes over the second inflection point and the second coil passes 'over the first inflection point, and last, a normal voltage peak of the same polarity as the first as the second coil passes over the second inflection point. There thus results the curve a", c", b. Thus it will be seen that the method of utilizing a critical separation between the two induction coils of a pair results in one impulse of double normal amplitude and two impulses of normal amplitude for all such defects; and this has been invaluable in detection and in discrimination against flux irregularities other than those of rail defects.

If the flux path along which the coil pair travels is of type A, shown in Fig. 2, then the double amplitude surge will haveV one electrical polarity; but if it is of the type B shown in Fig. 3, the double amplitude peak will be of opposite electrical polarity. It should be remembered that the voltage surge must be amplified so that the output may operate suitable recording means, and if it is borne in 'mind that the major or double amplitude voltage peak may be of either polarity, then it means that the amplification and recording equipment which is to be actuated by the double amplitude peak of a single coil pair should` be capable of operating in identically the same manner by such wave impulses of either polarity. This condition is nearly impossible to obtain in known impulse amplifiers, with the result that if the amplifier is set to respond to one impulse polarity correctly it may be insensitive or ultra-sensitive to impulses of the opposite polarity, thereby failing to record true defects or recording frequent spurious defects.

Besides the matters of wave addition and polarity sequencedescribed above, there is the matter of coverage to be considered. In order to examine a rail properly for internal defects, `it is necessary to have flux distortion detection means covering the entire head of the rail. This might be done by constructing a pair of detector coils Wide enough to extend from the gage side to the outside of the head. Such an arrangement is ineffective for two reasons; first, it is likely to be extremely sensitive to wide surface defects, such as wheel burns and flats, and second, it is extremely insensitive to all defects in which the phenomenon shown in Fig. 6 occurs, that is, where aflux decrease at one sideof the railhead completely nullifles a flux increase on the other side of the railhead, since both effects act upon each coil at the same time. For this reason the Sperry detector cars have employed two pairs of coils as disclosed in the patent to Harcourt C. Drake, No. 1,960,968, dated May 29, 1934. for Flaw detector unit, each pair covering approximately half the-railhead. The individual coils of each pair have been properly spaced relative to each other at very nearly the correct critical distance, that is to say, each pair yields a t curve such as the curve C/in Fig. 5, and the net result of the curves of the two pairs of coils is as shown in Fig. 12, there being two curves C and C' which overlap. From this diagram it will .be seen that there are portions of one of these curves C' which are opposite in polarity to portions of the curve C. Thus the portion of curve C' is in a direction opposite to that of portion Il of the other curve and will at least partially nullify the' eil'ect of what would otherwise be a double amplitude peak. Similarly, the portion I2 of curve C is opposite and will tend to nullify at least partially what would otherwise be the double amplitude peak I3 of curve C. Thus the action ofy the outputs of the two pairs of coils as they have vheretofore been employed may be detrimental, at least partially eliminating what should be the double amplitude peaks which are relied upon for the input to the amplifier.

A further objection to the Fig. V12 arrangement lies in the fact that the output leads from these pairs of coils do not actuate the amplification and recording equipment independent of polarity, but, rather, each coil pair actuates a circuit having definite polarity discrimination. `Thus separate circuits are provided with separate leads from each coil pair to said circuits, each circuit being designed to discriminate against one polarity, the discrimination of the two circuits being in opposite directions. This means that for at least one polarity sequence of a divided flux of the type disclosed in Fig. 6 the output from both coil pairs may be discriminated against and the defect remain unrecorded. i

The action of the detector equipment which has beenemployed on the Sperry detector cars and consists of the plurality of pairs of coils hereinbefore described, is further complicated by the fact that Ithe two amplifying circuits are not independent in action but enter separate amplifying channels and eventually combine within the amplifier, resulting in a random addition and cancellation in whole or part of the input signals.

The method and means which I shall now describe obviate all of the difficulties which have been set forth above in connection with the use of a plurality of coils as shown in Fig. 12. It obviates the possibility of impulses wiping each other out and also obviates the possibility that because of polarity discrimination these impulses will not be amplified. The method consists in arranging the detection coils in a series or chain, the number of units of this chain being even, each coil being connected in series opposition to the next and each spaced from the other by substantially the critical distance which has been described in connection with Fig. as being that distance in which any two adjacent coils connected in series opposition generate a double amplitude peak similar to c. This may best be described in connection with Figs. 7 to 11 inclusive. Thus, in Fig. 7 it will be seen that four coils are employed. Each coil is connected in series to the adjacent coil and is in opposition to the adjacent coil. Each coil is spaced from the other by a distance substantially equal to the critical distance. This means that not only willcoils I and 2 of the first pair be spaced a critical distance, but the adjacent coils 2 and 3 of the pairs of coils I, 2 and 3, 4 are spaced bythe said critical distance. As a result, there is obtained the voltage surges shown in Figs. 9 to 11. Thus, if pair I and 2 generates a cycle of pulses similar to the cycle C of Fig. 5, then, as shown in Fig. 9, the spacing is such that in addition to the two ordinary double amplitude peaks designated as d and d', the single amplitude peaks overlap and produce a third double amplitude peak in the opposite direction, as shown at d". If both coils generate impulses in the opposite direction, as shown in Fig. 10, the same thing occurs except that in addition to the two double amplitude peaks e and e' there is generated by the overlapping of the single amplitude peaks a third double amplitude peak e" in the opposite direction. If the pair of coils 3 and 4 generates a double amplitude peak of opposite polarity from that of coils I and 2, then the middle double amplitude peak is partially wiped out, but by the very nature of this condition two double amplitude peaks f and f are nevertheless obtained, one of each polarity. Y

The advantages of the above method and construction are many. First ofl al1, if such a. series of coils be arranged so ,thateach two adjacent coils track in the same path. then these pairs such as I and 2, 3 and 4. 5 and. 6, may be located anywhere on the railhead so that coverage is easily obtained. Secondly, since there are only two output terminals and hence only one means of amplification and recording is required, there may be as many pairs of coils located to track in as many paths as desired, so that the coverage may be made as complete as necessary without any complication of the amplifying-recording equipment. In previous forms, each coil pair would require a set of terminals and a separate input channel into the amplifier and also random mixing and cancellation would occur. Thirdly, since the nearest two coils, whether in the same pair or in adjacent pairs, are always separated by the critical distance and these are always of opposite lpolarity connection, it follows that detrimental cancellation of impulses cannot occur. Fourth, since the critical spacing is utilized not once but a plurality of times and appears between pairs as well as between the units of each pair, the discrimination between transn verse defects of which the flux is of the characteristic form (as shown in the graphs of Figs. 9, 10 and 11) and surface defects (where the flux distortion is of another form) is improved.

Referring again to Figs. 9 and 10, it` will be observed that with the additional double amplitude peak there results a wave train voltage output instead of random impulses. In other words, a double amplitude peak in one direction is followed by a double amplitude peak in the other direction which is followed `by a double amplitude peak in the iirst direction. This is true in the. Figs. 9 and 10 cases and even in the Fig. 11 cas'e where because of the partial cancellation of two minor peaks there nevertheless resuits double amplitude peaks successively of opposite polarity. If three pairs of coils had been used, or any greater number of pairs, this wave train would appear in the same form except that it would endure for more cycles. The advantage which follows from setting up this wave train is apparent from the statement of the diiiiculty heretofore experienced in amplifying these random impulses. It was necessary to have two separate circuits each of which discriminated against one polarity; and this meant, in those cases such as shown in Fig. 11 where the impulse was of both polaritles, that both outputs might be discriminated against with consequent failure to indicate. By the present invention, only two leads are used, with only a single input channel and amplifier, and since an aquidirectional wave train is necessarily established by every condition of flux distortion that may be encountered, it becomes impossible forthis ampliiier to discriminate against any defect, because both polarities are always present equally. Furthermore, the possibility of converting the signal given by a defect into a wave train is extremely important because it opens up entirely new possibilities of methods of ampliiication and critical filtering to eliminate all but the desired frequency set up by rail fissures. It is almost impossible to ilter random or non-equi-directional impulse signals and they are the most difficult form of signal to am- Dlify.

In conclusion, the advantages accruing from the method and means hereinbefore described may be stated as follows:

1. Reducing the detectorunit output terminals to two in number, regardless of the number of detector coils or pairs of coils employed, which allows the use of 4a single non-adding, non-cancelling ampliiier and recording device.

2. Improving the coverage factor by use of more pairs of coils.

3. Improving the factor of discrimination against types of defects other than internal flssures by using more and probably smaller coils.

4. Improving the discrimination factor by accenting the critical spacing periodicity.

5. Improving the probability of detection by simple repetition of signal peaks.

6. Absolute elimination of complete cancellation in either direction.

7. Further addition of voltage surges to make higher amplitude peaks.

8. Possibility of obtaining an equi-directional periodic wave train from a single defect.

In accordance with the provisions; of the patent statutes, I have herein described the principle and operation of my invention, together with the apparatus which I now consider to represent the best embodiment thereof, but I desire to have it understood that' the apparatus shown is only illustrative and that the invention can be carried out by other equivalent means. Also. while it is designed to use the various features and elements in the .combination and relations described, some of these may be altered and others omitted without interfering with the more general results outlined, and the invention extends to such use.

Having described my invention, what' I claim and desire to secure by Letters Patent is: l. In a rail aw detector mechanism having means for establishing an electromagnetic eld surrounding the rail and means for exploring said field for irregularities due to defects, said exploring means comprising an even number, but more than two, flux responsive means arranged in succession longitudinally of the rail, each ux responsive means being connected with its preceding fiux responsive means and with its succeeding flux rponsive means in series, said means being spaced relative to one another longitudinally oi' the rail a critical distance which is such distance that the outputs of adjacent flux responsive means reinforce each other to multiply the voltage output.

2. In a rail aw detector mechanism having means for establishing an electromagnetic field surrounding the rail and means for exploring said eld for irregularities due to defects, said exploring means comprising an even number, but more than two, flux responsive means arranged in succession longitudinally of the rail, each flux responsive means being connectedwith its preceding fiux responsive means and with its succeeding flux responsive means in series opposition, said means being spaced relative to one another longltudinallyof the rail a critical distance which is such distance that the outputs of adjacent flux responsive means reinforce each other to multiply the voltage output.

3. In a rail flaw detector mechanism having means for establishing an electromagnetic iield surrounding the rail and means for exploring said field for irregularities due to defects, said exploring means comprising a plurality of flux responsive means arrangedin succession longitudinally of the rail, each flux responsive means being connected with its preceding iiuxresponsive means and with its succeeding iiux responsive means inA series, adjacent flux responsive means being arranged in pairs, the' adjacent ux responsive means of adjacent pairs being spaced relative to one another longitudinally of the rail a critical distance which is such distance that the outputs of adjacent ux responsive means reinforce each other to multiply the voltage output.

4. In a rail flaw detector mechanism having means forestablishing an electromagnetic eld surrounding the rail and means for exploring said field for irregularities due to defects, said exploring means comprising a plurality of flux responsive means arranged in succession longitudinally of the rail, each flux responsive means being connected with its preceding iiux responsive means and with its succeeding ux responsive means in series, adjacent ux responsive means being arranged in pairs, the adjacent ux responsive means of adjacent pairs and the flux responsive means of each pair being spaced relative to one another longitudinally of the rail a critical distance which is such distance that the outputs of adjacent flux responsive means reinforce each other to multiply the voltage output.

l JULIAN H. BIGfELOW.ia

Images