In order to measure the grain size of continuous moving sheet at various stages of processing, the possibility of detecting the areas of grain boundary leakage using electromagnetic sensors was investigated. A test rig was constructed using a number of bands of grain oriented electrical steel (with different grain sizes) slung between two revolving drums. The strips pass over an array of permanent magnets and electromagnetic sensors are arranged above the strip. For a range of grain sizes the pulse repetition rate was found to correspond with the rate predicted by considering the grain size and the speed or travel of the sheet.
I. Introduction
The on-line assessment of grain oriented mate- rial is currently handled by a conventional loss tester which erwraps the complete width of the moving sheet and therefore produces an average loss figure for the full strip width. Studies have however shown that the material produced is not always homogeneous, with serious variations in loss occurrirg from one part ot the sheet to another. Al,ho~gh the loss tester should show gross variations in loss along the iengtil oI a co~l, small variations across the strip width are effec- tively invisible. Because of the nature of the rolling process, defects in the finished material are sometimes found to be linear along the length of the coil. The grain size of the materiai has been found to vary across the strip width in addition to the occurrence of sharply defined bands of fine grained material. These features can be harmfal (in terms of loss) if material is sold slit to a narrow width.
The need for an on-line device capable of detecting variations across the strip width is
Correspondence to: Mr. R. Taylor, British Steel Electricals, Orb Works, P.O. Box 30, Newport, Gwent NP9 0XT, UK.
therefore apparent. This would give an immedi- ate indication of the quality of the steel being produced and would obviate the need to cut samples and etch off the surface coating, a time consuming process which destructively test~ o~:ly a tiny area of the coil.
2. Operating principle
During the development of the magneti .- do- main viewing device [1] it was noticed that grain boundaries in large grained material were clearly shown when the sheet steel was magnetised along its length, the coating being left intact. Fig. 1 shows a typical image obtained. The grain bound- aries are shown by dense black lines in 80% of cases. This is due to the agglomeration of the magnetic p,-rtic:es over the leakage fields at the grain boundaries. The remaining 20% or so of hn , ,ne l .~ r ; c~c o..',~ n n t ~hn,Lvn hoe -~ l~o t h o roJ~,v'~nf
adjacent grains are well aligned with little leakage of magnetic flux from the bounda~' region.
Textures are also shown w;thin the grains. These are due to the surface roughness of the sheet and might possibly be also linked with grain dip angles. Fig. 2 shows a full width grain bound-
96 R. Taylor / On-line grain size measuring aystem
Fig. 1.
a13, viewer which was built to eliminate the need to remove the surface coating for grain size evalu- ation.
The density of the grain boundary lines com- pared to the other features shows that the field in these regions is relatively strong and well defined.
IIL
Fig. 2.
R. Taylor / On-line grain size mt'asutqng ,O'stem 97
Because the steel sheet moves through a produc- tion line at a fixed speed, it was considered that inductive pick-ups positioned above a magnetiscd portion of the strip would produce a series of pulses of a frequency which corresponds directly with the passage of grain boundaries under the pick-ups.
3. Experimental apparatus
Barkhausen noise and possibly domain fields. Be- cause the grain boundary leakage fields have an orthogonal component, a vcrtically oriented sen- sor arrangement was tried. This consisted of a "whisker" of grain oriented material with a wind- ing of several thousand turns of fine enamelled copper wire. The signals from this arrangement were found to be much "cleaner", the difference between grain sizes being much more apparent.
Bands of slit coil of different grain size (3-7) were mounted between two rollers, one of which was driven at 200 f t . /min by a motor and pulley system. The pick-up sensors were mounted on pivoting trolleys which maintain a fixed distance between t!_3 sensors and the steel sheet (1 mm). The sensor outputs were fed to an oscilloscope via an amplifier and filter system. Initially, cas- sette tape recorder heads were used as sensors but these proved to be too sensitive, The practical result was that signals from material samples with widely differing grain sizes were not as "different" as was hoped since grain boundary signals were swamped with signals due to surface roughness,
4. Results
Fig. 3 shows a dual beam storage oscilloscope trace for samples of ATM grain sizes 3 and 7. It is apparent that the trace for the fine grained material (ASTM7) shows more magnetic "events" or discontinuities than the trace for the coarser grained material. This is to be expected because the fine grained material will present more grain boundaries to the fixed sensor per unit time. Since the grain boundary signals from the sensor fall within the audible spectrum, a power ampli- fier and loudspeaker were connected to each of the scnsor coils. The difference between the two
Fig. 3.
93 R. Taylor / On-line grain size measuring ~,stem
signals was plainly apparent, the larger grained material giving rise to a lower pitched tone.
The traces shown in fig. 3 were produced with a time base of 5 ms per graticule division. A complete sweep of the screen therefore takes 50 ms. During this period approximately 50 mm of steel passes under the sensor. ~'or an ASTM grain size of 7 the average grain "diameter" is 3 mm and so we might expect roughly 17 "events" (50/3) to occur. An examination of the upper trace shows 15 or 16 abrupt changes in gradient.
In the same period of time the lower tra~e (derived from the ASTM3 grain size material) shows 5 or so changes in direction for the 50 mm of steel traversed. With an average grain "diame- ter" of 13 mm (for ASTM3)we would expect (on average) to find 4 grains within 50 mm of the material.
material and the frequency of pulses received by an inductive sensor placed over magnetised mov- ing sheet. It should be possible to construct an electronic data handling system (for material with grain size similar to those that were studied) which gives continuous read-outs of grain size for a bank of sensors arranged across the full width of manufactured coils and this approach is cur- rently being pursued.
Acknowledgement
The author would like to thank Orb Electrical Steels Limited for permission to publish this work.
5. Conclusions Reference
It has been shown that a definite correlation exists between the grain size of grain oriented
[1] R. Taylor, Proc. SMM9 Conf. El Escorial (September 1989).
