This is a §371 of International Application No. PCT/JP2008/065221, with an inter-national filing date of Aug. 20, 2008 (WO 2009/025384 A1, published Feb. 26, 2009), which is based on Japanese Patent Application No. 2007-217180, filed Aug. 23, 2007, the subject matter of which is incorporated by reference.
This disclosure relates to a method and an apparatus for detecting concavo-convex shape surface defects existing on a ferromagnetic metal and, in particular, to a method and an apparatus for detecting concavo-convex shape surface defects, wherein the concavo-convex shape surface defects each have a concavo-convex level in the range of 0.5 to 6 μm and an area of 10 mm2 or larger and exist on a rough surface, with Ra being in the range of 0.5 to 2 μm, of a steel sheet having a thickness in the range of 0.4 to 2.3 mm.
In a manufacturing process of a ferromagnetic metal, in particular, thin steel sheets, foreign matter adhering to or caught up in a roll installed in a manufacturing line sometimes produces concavo-convex shape on the roll itself, and this concavo-convex shape may be transferred to resulting steel sheets as roll-generated flaws.
Such roll-generated defects include concavo-convex shape surface defects that each exist on a rough steel sheet surface (Ra=0.5 to 2 μm) and have a smooth outline (radius of curvature R≧10 mm), a concavo-convex level of 5 μm or lower, and an area of 10 mm2 or larger. Such defects are hereinafter referred to as fine concavo-convex shape surface defects.
Most of roll-generated defects have a relatively high concavo-convex level and are visible, and thus they can be easily found in the course of a manufacturing line. However, such fine concavo-convex shape surface defects have concavo-convex levels as small as the surface roughness of steel sheets. The optical difference between these fine defects and surface roughness is small, and thus the defects cannot be easily found in the course of a manufacturing line when observed without any treatment. However, once the surface is coated and the surface roughness is filled with a coating agent to make a smooth surface, the fine defects become clearly visible thereby posing serious problems in terms of appearance. Therefore, detection of fine concavo-convex shape surface defects prior to coating is an important issue in quality control.
The shape of fine concavo-convex shape surface defects includes point flaws like roll-generated flaws described earlier and flaws extending in the longitudinal direction of a steel sheet, such as linear marks or buckling marks.
Such fine concavo-convex shape surface defects are formed by transfer of concavo-convex shape of a roll to a steel sheet, and this situation persists until the roll is replaced or the process is improved. Therefore, early detection and countermeasures are very important also for yield improvement.
To find such fine concavo-convex shape surface defects, threading of steel sheets is stopped during operations in each inspection line in an iron-making process, and thereafter inspectors rasp all the coils with a whetstone and perform visual inspections thereof. After being rasped with a whetstone, convex portions, which are more strongly rasped than concave portions, have a higher reflectance than concave portions. This makes the difference between convex and concave portions clearer and more visible in visual inspections. This method is called whetstone inspection.
Unfortunately, this method necessitates stopping of inspection lines and requires a substantial period of time, thereby reducing the efficiency of operations. To address this situation, a method for automatically inspecting concavo-convex shape defects each having a concavo-convex level on the order of a few micrometers and a smooth outline has been developed. Examples of such an automatic surface analyzer may include the techniques disclosed in Japanese Unexamined Patent Application Publication Nos. S58-86408, H5-256630, H6-58743 and 2000-298102.
However, the technique disclosed in Japanese Unexamined Patent Application Publication No. S58-86408 is a technique for inspecting a mirror surface, and thus cannot be applied to an object with a large surface roughness because of the problem that beams converging into or diffused from concaves and convexes of flaws are masked by beams diffused by surface roughness and thus the flaws cannot be detected.
Also, the technique disclosed in Japanese Unexamined Patent Application Publication No. H5-256630 is intended for steel sheets, but still cannot be applied to objects other than materials having a mirror surface such as stainless steel sheets. Furthermore, this technique is admittedly effective in detecting concavo-convex defects extending perpendicular to the illumination detection, but is insufficient in terms of detection sensitivity for those extending parallel to the illumination detection.
Additionally, the technique disclosed in Japanese Unexamined Patent Application Publication No. H6-58743 is intended for unpolished wafers having rough surfaces, but still has the problem of low detection accuracy because it determines whether a flaw is present on the basis of the total light intensity and thus cannot detect specific signals of individual flaws.
The technique disclosed in Japanese Unexamined Patent Application Publication No. 2000-298102 was developed to address this situation and, accordingly, an apparatus for this technique has very high detection sensitivity. However, this technique requires an angle of incidence as large as close to 90°, thereby making it difficult to introduce a necessary apparatus into an actual operation line. It has also the problem that adjustment of optics is difficult.
In view of general defect-detecting methods without limiting the target thereof to fine concavo-convex shape surface defects, there are detection methods wherein magnetic flux is applied to a sample, for example, the technique disclosed in Japanese Unexamined Patent Application Publication No. H8-160006 for detecting an inclusion existing inside an object using a magnetic flux leakage inspection method. This patent document describes that surface defects can be detected also by using this magnetic flux leakage inspection method in the form of leakage flux signals generated by changes in the surface shape (signals generated by variation or disturbance of magnetic flux due to changes in the shape).
However, the intensity of signals for this magnetic flux leakage inspection corresponds to the shape change level of surface defects (concavo-convex change level), and thus the lowest shape change level of surface defects (concavo-convex change level) detectable in an automatic inspection is limited. Those skilled in the art consider this lower limit to be approximately 100 μm (Magnetic Flux Leakage Inspection Method for Iron and Steel Products, Feb. 28, 2001, NDI Department (Technology and Technique Transfer Committee), Working Group for Quality Control, the Iron and Steel Institute of Japan). In other words, it is recognized that stable detection of surface defects having a shape change level of approximately 100 μm or lower is difficult and signals generated by changes in the shape of approximately 100 μm or lower serve only as noise sources observed in flaw detection for other purposes (e.g., detection of inclusions).
Although such a simple comparison may be inapplicable, signals for detecting inclusions similarly correspond to the size of defects, and thus the lowest detectable size of a defect in the thickness direction of the test steel sheet is still limited. Those skilled in the art consider this lower limit to be approximately 10 μm (Development of a Minute Inclusion Detector for Thin Steel Strips, CAMP-ISU, Vol. 10 (1997)-289, Development of Gouge Defect Detector, Chiba 2CGL, 131st Control Technology Working Group Meeting, Manufacturing Technology Department, the Iron and Steel Institute of Japan, Development of Non-metallic Inclusion On-line Detection System, CAMP-ISIJ, Vol. 7 (1994)-1270 and State-of-the-art Techniques for Internal Quality Measurements in Steel Sheets Production Processes, Kawasaki Steel Giho, 31 1999 4.211-215).
Furthermore, in such magnetic flux leakage inspection, an increase in the sheet thickness would result in a widened range of flaw detection in the thickness direction, thereby resulting in increased background noise generated by the steel sheet, increased magnetizing force being needed for magnetization of the steel sheet, reduced flatness of the surface of the steel sheet leading to difficulty in scanning the steel sheet with a sensor, and other problems. For these reasons, the thicker the test sheet is, the more difficult flaw detection is. With the inclusion detector described in the publications listed above, the lowest detectable size of an inclusion in a steel sheet discussed, whose maximum thickness is 2.3 mm, would probably be much larger than 10 μm (size in the direction of the steel sheet thickness) if they can be detected at all, considering the fact that this inclusion detector is intended for thinner steel sheets used for cans.
Meanwhile, Report of Explorative Research on Seed Technology for Detection of Whetstone-inspection-level Flaws, July 1995, WG for research on cold rolling in rolling, refining, and temper systems, Iron and Steel Technology Committee, the Japan Iron and Steel Federation discloses the results of explorative research on the detection of whetstone-inspection-level flaws conducted by WG for research on cold rolling in rolling, refining, and temper systems, Iron and Steel Technology Committee, the Japan Iron and Steel Federation from September 1994 to July 1995 (Report of Explorative Research on Seed Technology for Detection of Whetstone-inspection-level Flaws). Previous approaches to automatic inspections of fine concavo-convex shape surface defects were mainly based on a light-detecting method, and, to date, there have been no approaches using a magnetic flux leakage inspection method.
Therefore, it can be said that the use of magnetic flux leakage inspection for detecting fine concavo-convex shape surface defects that can be visible only after the surface of a test steel sheet is rasped with a whetstone would not normally be considered by those skilled in the art of magnetic flux leakage inspection.
It could be helpful to provide a practical method for stable detection of concavo-convex shape surface defects that exist on a test object having a surface roughness Ra approximately in the range of 0.5 to 2 μm, wherein the concavo-convex shape surface defects are usually almost invisible and detectable only by whetstone inspection, are difficult to detect by automatic detection, and each have a concavo-convex level on the order of a few micrometers, and also provides an apparatus for the same objective.
We thus provide:
We provide for stable detection of fine concavo-convex shape surface defects that exist on a test object having considerable surface roughness, wherein the fine concavo-convex shape surface defects are usually almost invisible to the naked eye and detectable only by whetstone inspection, are difficult to detect in automatic detection, and each have a concavo-convex level on the order of a few micrometers and a smooth outline.
To measure fine concavo-convex shape surface defects generated during rolling (roll-generated fine concavo-convex shape surface defects), we first conducted X-ray diffraction analysis of such defects in some sheets to determine the physical characteristics of the defects. As a result, it was found that these fine concavo-convex shape surface defects had strain that probably emerged during transfer of flaws from the roll in the process of defect formation.
Therefore, we focused on the possibility that sensing signals attributed to the shape, which are insufficient for defect detection alone, together with such strain enables the detection of roll-generated fine concavo-convex shape surface defects and experimentally tested whether this strain generated during the formation of defects could be measured using a method based on magnetism. In this experiment, we first prepared a simple apparatus for magnetic flux leakage inspection, and then tried to detect roll-generated fine concavo-convex shape surface defects in some sheets to confirm that this apparatus could sense signals. After that, the samples were annealed at 850° C. for 10 minutes so that the strain thereof was fully removed, and then subjected to magnetic flux leakage inspection once again. As a result, it was found that the intensity of signals sensed after annealing was substantially lower than that sensed before annealing.
The defect shown in
Consequently, we found that roll-generated fine concavo-convex shape surface defects cannot be detected only with signals attributed to the concavo-convex shape of such defects, whose concavo-convex levels are of the same order as the roughness. However, sensing signals of internal strain, which emerges during the formation of the roll-generated fine concavo-convex shape surface defects, together with the signals attributed to the concaves and convexes enables defect detection using a detection method based on magnetic flux (here, examples of such a detection method based on magnetic flux may include magnetic flux leakage inspection using direct current, magnetic flux leakage inspection using alternating current, eddy current inspection, residual magnetic flux inspection, and magnetic particle inspection). On the basis of these findings, we optimized various conditions to be suitable for the characteristics of such fine concavo-convex shape surface defects.
The principle of detection of strain in magnetic flux leakage inspection is described below. Strain generates leakage flux signals, and this is probably because strain changes the crystal lattice interval in the sample and accordingly changes the interaction between spins, thereby changing the magnetic characteristics of the sample. However, a usual magnetic flux leakage inspection method generally uses a saturated magnetic flux for measurement to improve the signal intensity and remove the influence of noise generated by unevenness of magnetic characteristics (magnetic permeability) of the test object.
If a sample is very strongly magnetized to magnetic saturation as with those used in such usual magnetic flux leakage inspection, spins are all aligned in the same direction and thus the signals of strain are probably suppressed. Therefore, a magnetization level that is lower than magnetic saturation (rotational magnetization region) is considered preferable because it results in stronger signals of strain. This probably corresponds to the fact that the comparison between the measurement data of the B-H curve obtained in a steel sheet strained over the entirety thereof by applying an external force thereto and that obtained in the steel sheet without any strain gives a larger difference in the region of magnetization levels lower than those in the magnetic saturation region (rotational magnetization region).
Therefore, the measurement of roll-generated fine concavo-convex shape surface defects involves measurement of signals attributed to strain, and sensitive measurement of such signals can be more easily achieved by using magnetization levels lower than saturated magnetization. It should be noted that the intensity of a magnetic field applied to a sample described herein means not the total magnetic field generated by a Magnetizer for magnetization of the sample, but also the magnetic field directly applied to the test site of the sample, i.e., the quantity corresponding to H in a B-H curve obtained in the test site of the sample.
Meanwhile, such a magnetic flux leakage inspection apparatus is usually used for product inspections in a final line, so that there are various processes between occurrences of defects and detection of them. These processes include, for example, a step wherein the steel sheet is released from strain, which emerges during the formation of defects, through the removal of the strain by heat, the release of strain by means of another stress such as tension applied during threading, or other circumstances. In such a process, detection of signals attributed to strain may be impossible. Therefore, the location where the detection is performed should be discussed.
Manufacturing processes of cold rolled steel sheets usually include a thermal process, such as an annealing process, to remove the influence of hardening that occurs during rolling. This thermal process sometimes involves a temperature increase to recrystallization temperature. Heating a steel sheet in this thermal process may result in the suppression of defect signals as with the experiment described earlier. Therefore, measurement should be carried out at a point just after rolling that causes defects, like “Position A” in
Even after strain is removed once through such a thermal process, another strain is provided by temper rolling. Because of concaves and convexes, the size of strain given to defect portions is different from that given to normal portions. Therefore, detection of defects is possible even after the thermal process as long as it is carried out at a point after temper rolling, like “Position C” in
For detection of defects generated by every roll that causes defects described above, a preferred point to carry out the measurement is a point after temper rolling. Furthermore, a point just after temper rolling, “Position C” in
It should be noted that a manufacturing process of surface-treated steel sheets, such as galvanized steel sheets and tinned steel sheets manufactured using CGL, EGL, or other lines, basically includes the same processes until temper rolling although the manufacturing process further includes a plating process or other surface treatment processes. Therefore, the preferred point of such measurement is the same as that described above.
Even if an annealing process (CAL) or a plating process (CGL, EGL) does not include temper rolling, the same result can be obtained by conducting such measurement at a point after tempering performed with a refining/temper line, such as a recoiling line.
In particular, in a plating process and later processes, defects that transfer concaves and convexes to the plated layer on the surface without affecting the substrate metal may occur. Such defects can be better detected after strain is given to the substrate metal by temper rolling, and thus a method wherein the measurement is carried out after temper rolling is especially preferable in this case.
Meanwhile, rolling at a high rolling reduction by rerolling or the like totally eliminates defects. Therefore, for successful detection of roller-generated defects, it is also obviously important to carry out the measurement between occurrences of the defects and rolling at a high rolling reduction. In addition, such rolling at a high rolling reduction admittedly eliminates defects totally; however, for the prevention of recurrences of defects, it is important to detect such defects also in the case that such rolling at a high rolling reduction is used.
The steel sheet 1 has a roller-generated fine concavo-convex shape surface defect 2 having a level in the thickness direction of a few micrometers, which is equivalent to the level of roughness. The roller-generated fine concavo-convex shape surface defect detector 9 is placed by the steel sheet 1, and the detector is configured as follows: the magnetizer 4 and the magnetic sensor 5 are arranged on the same side facing the steel sheet 1; the magnetizer 4 magnetizes the steel sheet 1 using direct current supplied by the magnetization power source 3.
Magnetic flux generated between both magnetic poles by the magnetizer 4 passes through the steel sheet 1. If a defect 2 exists on the steel sheet 1, strain that emerges during the formation of the defect 2 and surrounds the defect 2 interrupts the magnetic flux, and then the change in the magnetic flux can be detected by the magnetic sensor 5. The signal output from the magnetic sensor 5 is amplified by the amplifier 6, and then noise contained therein is removed by the filter circuit 7. After that, the defect analyzer 8 determines the point that corresponds to a signal having intensity equal to or higher than a threshold value to be a defect.
Although this example uses a direct current signal for magnetic flux leakage inspection, an alternating current signal may be used instead. In this case, a synchronous detection circuit is required. Furthermore, in this example, the magnetizer and the magnetic sensor are arranged on the same side facing the steel sheet. However, if a direct current signal is used, they may be arranged on opposite sides across the steel sheet, and even if an alternating current is used, they may be arranged in the same way, on opposite sides across the steel sheet, as long as the excitation frequency is sufficiently low for the sheet thickness.
In addition, although the defect analyzer 9 uses the intensity of a signal attributed to a defect for determination of the defect, the length, width, or area of a point that gives a signal having an intensity equal to or higher than a threshold value may be used for determination together with the signal intensity, and two or more of them may also be used in combination for determination. Also, the phase of alternating current may be used for determination in the case where alternating current is used.
We measured B-H curves for some steels with different compositions. The result is shown in
As can be seen in
Furthermore, as shown in
Also in detection of roll-generated fine concavo-convex shape surface defects, concaves and convexes are present although they are as fine as a few micrometers at minimum, and thus such signals attributed to concaves and convexes can be obtained as well. Therefore, the following two conditions for flaw detection can be combined to improve the sensitivity for detecting defects: flaw detection at a magnetic field intensity less than 25000 A/m for sensing signals attributed to strain; flaw detection at a magnetic field intensity of 25000 A/m or higher for sensing signals attributed to concaves and convexes. Furthermore, comparing the signal component attributed to concaves and convexes with that attributed to strain enables comparing of the concavo-convex level and the strain level, thereby making it possible to identify which roll causes defects by, for example, assuming that the combination of a low concavo-convex level and large strain represents a defect caused by a roll used at a high rolling reduction.
Meanwhile, in this example, the value of magnetic flux density was calculated from a magnetic field intensity on the basis of a predetermined B-H curve. The value of a magnetic field intensity for this calculation may be measured in the vicinity of the test site of a sample.
Although this example used leakage flux based on direct current for measurement, a leakage flux method based on alternating current, an eddy current inspection method, or a magnetic particle inspection method may be used instead as long as the method allows for sensing of signals attributed to strain.
Furthermore, although this example used a hall element as the magnetic sensor, a coil, a magnetic resistance element, or a SQUID may be used instead as long as the element can sense magnetism. Also, a single magnetic sensor or a plurality of magnetic sensors may be used. If plurality of magnetic sensors are used, a wider area can be simultaneously measured by arranging the magnetic sensors in the direction perpendicular to the traveling direction of the sample and parallel to the sample itself. However, a too large pitch of magnetic sensors would cause overlooking of defects passing between the magnetic sensors, whereas a too small pitch would reduce the efficiency. Although a pitch of magnetic sensors in the range of 0.5 to 3 mm is enough for detection, one in the range of 0.8 to 2 mm is the most preferable for sufficient detection sensitivity and efficiency.
Moreover, the lift-off used in this example was 1 mm, and this value was chosen on the basis of the following findings. Especially small ones of roll-generated fine concavo-convex shape surface defects sometimes give very slight leakage flux signals. To detect such defects, optimization of the lift-off, which is a distance between a sensor and a sample, is required in addition to the measures described earlier.
In general, defects necessitating whetstone inspection for detection thereof in an iron and steel manufacturing line are, as described earlier, those each having concaves and convexes equivalent to roughness (approximately a few micrometers) and a radius of curvature R of 10 mm or larger and thus masked by roughness of a few micrometers. Such defects often each have a diameter approximately in the range of 4 to 30 mm and an area approximately in the range of 10 to 1000 mm2 on the surface of a steel sheet. In a usual magnetic flux leakage inspection method, a smaller lift-off results in a higher sensitivity and thus is more advantageous. However, in detection of such a defect, which has a concavo-convex level as small as a few micrometers and a large area, a too small lift-off would lead to the situation that only signals from a small portion of the defect are sensed, thereby posing a problem of poor efficiency in detection of defects, for example, the need for arrangement of a plurality of sensors.
In detection of small defects, the results of which are shown in
This situation is discussed below with reference to
It is generally known that, in magnetic flux leakage inspection, the leakage flux density coming from a defect increases as the detecting apparatus comes closer to a sample, and thus a smaller lift-off results in a stronger defect signal being detected, whereas a larger lift-off results in a weaker defect signal being detected. Another consideration is that the detection area detected by the sensor is widened as the lift-off is increased.
In general, a defect difficult to detect is one having small concaves and convexes and a small area. Such a defect is inherently smaller than the detection area of a sensor, and thus the area of the defect encompassed in the detection area of the sensor decreases as the lift-off is increased (see
On the other hand, defects to be measured have a very small concavo-convex level and a large area (corresponding to 5 mm or larger in diameter). Also in detection of such a defect, the defect signal and the noise signal coming from a sample itself decrease together as the lift-off is increased. However, a larger lift-off does not lead to a weaker defect signal because the size of the defect is large and thus the area of the defect encompassed in the detection area of a sensor remains constant until the detection area of the sensor exceeds that of the defect even if the lift-off is increased (see
As for noise signals, it can be considered that the random noise component decreases as a result of arithmetic averaging until the lift-off is increased to some extent. Therefore, because of the relationship between the size of the defect and the arithmetic average of random noise, S/N virtually increases until the lift-off reaches a certain value. Further increasing the lift-off would lead to a more significant decrease in the defect signal as with the case of a small defect, thereby probably causing S/N to decrease.
In this way, S/N increases even if the lift-off is increased to some extent as long as the defect is smaller than the area of the sample that can affect the sensor because the increased lift-off reduces noise attributed to the sample itself while keeping the defect signal almost constant. This probably contributes to establishment of the optimum range.
In addition, whetstone inspection usually performed in an iron and steel manufacturing line involves rasping such defects as described above, which each have a concavo-convex level of a few micrometers and a size approximately in the range of 5 to 30 mm in diameter on a steel sheet, with a whetstone to make them visible, but this inspection often further includes inspection of fine defects that can be detected without being rasped with a whetstone. Such defects can be detected without being rasped with a whetstone and thus our method does not handle them.
The results based on actual data obtained for defects that are not easily visible (usually almost invisible and detectable only by whetstone inspection) are shown below.
As shown in
Then,
Furthermore,
As can be seen in
As shown in
Number | Date | Country | Kind |
---|---|---|---|
2007-217180 | Aug 2007 | JP | national |
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/JP2008/065221 | 8/20/2008 | WO | 00 | 12/31/2009 |
Publishing Document | Publishing Date | Country | Kind |
---|---|---|---|
WO2009/025384 | 2/26/2009 | WO | A |
Number | Name | Date | Kind |
---|---|---|---|
2814019 | Bender | Nov 1957 | A |
3535625 | Pratt | Oct 1970 | A |
7417425 | Machi et al. | Aug 2008 | B2 |
20020121896 | Kato et al. | Sep 2002 | A1 |
20070108974 | Machi et al. | May 2007 | A1 |
Number | Date | Country |
---|---|---|
58-086408 | May 1983 | JP |
5-256630 | Oct 1993 | JP |
6-058743 | Mar 1994 | JP |
8-160006 | Jun 1996 | JP |
2000-298102 | Oct 2000 | JP |
3271246 | Apr 2002 | JP |
2004-191220 | Jul 2004 | JP |
2005-061940 | Mar 2005 | JP |
2005-227133 | Aug 2005 | JP |
2007-071659 | Mar 2007 | JP |
Entry |
---|
Suechika et al, Partial Translation of JP 2004-191220 A, Aug. 7, 2004. |
“Magnetic Flux Leakage Inspection Method for Iron and Steel Products,” NDI Department (Technology and Technique Transfer Committee), Working Group for Quality Control, the Iron and Steel Institute of Japan, Feb. 28, 2001, pp. 144-147 and English translation (1 sheet). |
Development of a Minute Inclusion Detector for Thin Steel Strips, CAMP-ISIJ, 1997, vol. No. 10, p. 289 and English translation (1 sheet). |
“Development of Gouge Defect Detector,” Chiba 2CGL, 131st Control Technology Working Group Meeting, Manufacturing Technology Department, the Iron and Steel Institute of Japan, Date, two-page coversheet and pp. 1-5 and partial English translation (1 sheet). |
Yokota, H., “Development of Non-metallic Inclusion On-line Detection System,” CAMP-ISIJ, 1994, vol. 7, p. 1270 and English translation (1 sheet). |
Takada, H. et al., “State-of-the-art Techniques for Internal Quality Measurements in Steel Sheets Production Processes,” Kawasaki Steel Giho, 1999, vol. 31, No. 4, pp. 211-215 (English Synopsis on first page). |
“Report of Explorative Research on Seed Technology for Detection of Whetstone-inspection-level Flaws,” Iron and Steel Technology Committee, the Japan Iron and Steel Federation, Jul. 1995, three-page coversheet and pp. 1-78 and partial English translation (1 sheet). |
Number | Date | Country | |
---|---|---|---|
20110000338 A1 | Jan 2011 | US |