Defect mapping is a most powerful troubleshooting technique. One, unfortunately, of which few take full advantage. Instead, if you ask "Where do the defects show up?", the response will invariably be a verbal one such as, "It shows up on both sides, but the back is worse." It may be a million-dollar problem, but don't be surprised if there's no map, no graph and no statistical analysis. Same thing with customer returns. It's easy to find the number of rolls that were returned last month. Don't, how-ever, expect to be blessed with even rudimentary mapping such as roll positions, roll diametral locations and shift or crew records.

If you are a detective and want to catch a crook, you must develop a profile or MO (modus operandi). Crooks will betray their identities based on behavior patterns. Some prefer to work in stealth at night, others snatch-and-grab during broad daylight. The same thing is true for industrial problem-solving. Patterns will lead to the cause. These patterns have to do with CD position, MD position and time.

Perhaps the most important pattern is a histogram of CD position. A histogram is nothing more than a graph showing the number of occurrences at each location. Many defects are wide and may prefer the ends or center. Examples here are baggy ends or baggy center, respectively. Their CD positions may only be able to be pinpointed as close as the center of the roll that failed.

Others, such as gauge bands and corrugations, are relatively narrow. Not only can we locate the bands to a fraction of an inch; we also have another pattern to use, namely, width. The CD location and the width of the defect correspond to the location and width of the specific offending element upstream. For example, it is not possible for a die bolt with a 2-in. wide spacing to cause or cure a 1-in. wide streak.

For narrow defects, we visually mark the center of the defect area and use a tape measure to reference its location to some known datum. Differences in manufactured width, necking, edge trim, web tracking and other factors must be accounted for so that a defect location can be traced precisely to the upstream forming element. Defect maps make hidden patterns visible. For example, while gauge band and corrugation locations may seem to come and go, you may find that they actually prefer to come and go in very distinct positions corresponding to an element upstream with a uniformity problem. Spacing between defects rather than positions may also be a clue for some defects such as tin-canning. Sometimes the CD position moves sinusoidally, indicating that an oscillator or randomizer is involved.

Another diagnostic pattern is MD position. Are the defects randomly located, or do they come in bunches? One example of an MD bunch pattern is defects that favor the inside or outside of a wound roll. This demonstrates that winding affects that problem, even though there will always be material factors as well. If there is no wound-roll diametral preference, winding is almost certainly not involved.

Another MD pattern is when defects repeat at a wavelength corresponding to the circumference of an upstream roller, or at a time corresponding to some repetitive cycle in manufacturing upstream. This pattern is often masked by missing defects and the superimposition of defects that are caused by other factors. To find the hidden patterns, you need an FFT (Fast Fourier Transform) analysis of the MD map.

Some people are blessed with map makers. In well-equipped paper mills, for example, color-coded maps made by optical detectors show positions of flaws smaller than a fingernail on a 300-in.-wide sheet traveling 3,000 fpm. If you don't have million-dollar sensors, you can use a $10 tape measure and graph paper to do the same thing. You just need to invest the time. It will be time well spent.