A web may not be flat for two reasons. First, it may have internal stresses built-in by manufacturing. Second, it may have external stresses imposed by web handling. It is often easy to tell the difference between the two. If the web does not lay dead straight and dead flat on a table, then it is distorted by residual manufacturing stresses. If the web is dead flat on the table, but not in the machine, then web handling has distorted it.

Whether inspecting the web in situ or in the lab, the reference is always the same—dead flat like a sheet of glass. By this measure, all webs are flawed to some degree. However, this is the approach we must take as troubleshooters. If you take a more liberal reference, such as comparing it to a mere "good" web, you are throwing away information. It is analogous to the near-miss in accident reporting. Just because the person didn't get hurt (web didn't get rejected) this time, doesn't mean that there is no potential for an accident (rejection) next time. Skilled web handlers can look at features so subtle they are best described as shades or shadows. They know that these subtle features may hint at the seeds of layflat or web-handling problems that are just waiting to sprout.

Thus, troubleshooters and quality inspectors must use different standards. In troubleshooting, you try to prevent a problem from occurring. In quality inspection, you try to prevent a problem from reaching the customer. In troubleshooting, for example, you can see a wrinkle long before it has occurred because the trough (shadow) hints at a foldover that may happen at some other time when conditions are just a bit more pronounced. In quality, the subtle trough grown to a bulge or foldover crossing the roller will leave a visible mark to which the customer may object.

Only after determining whether flatness problems are due to processing or handling do we go to the next step of breaking it into subcases. Let us assume the problem is internal residual stresses from manufacturing. If the stresses vary through the thickness, you have curl. If the stresses vary across the width in a linear fashion, you have pure camber. If the stresses vary irregularly across the width or length, you have the more general case of baggy lanes or patches. To improve the quality, you try to level the product in forming/processing. It is usually difficult to stress-relieve the material after the fact by mechanical or thermal yielding.

If we assume the problem is with external residual stresses imposed by web handling, we will have an entirely different set of subcases. The first thing we look at is the angle of the wrinkle, or as we now know, angle of the shadow. If it is aligned with the MD (machine direction), the web wants to be wider than it currently is and has buckled due to CD (cross direction) compression. This case may be further broken down into subcases of excessive tension, tension drop, hygro-thermal expansion, improper spreading and so on. In most MD cases, we either eliminate the root cause or treat with spreading.

If the shadow is slightly angled with the MD, we know that either the web is crooked and/or the machine is crooked. The angled orientation is the fingerprint of shear buckling. In most cases, we begin our work with the web-handling aspects of the machine. First, we have more control over our machine than raw material coming from upstream. Second, the machine is easier to make precise than the web. Finally, the list of web-handling things to look for on the machine is shorter: rollers do not deflect excessively, roller diameter does not vary across the width, rollers are aligned, and nips are level.