What do you mean by web machine quality?

We have lots of ways to define web quality. But how do we judge the machine's quality? Here, we can seldom use any conventional web or uptime measurements. For example, let's say that you have side-by-side machines with drastic differences in runnability. It could be due to a different product mix or raw material, neither of which is the machine's fault. Even if raw material and product were identical, one could not make conclusions. The machines have different operators. Even if operators rotated between the machines, you couldn't conclude anything. The machines undoubtedly have some differences in their design—even if they were made from the same set of blueprints. So, we can't really pin down machine quality by comparisons. At best, we can say there is a difference in machine performance.

I will offer four different machine metrics. All are important to waste and delay, are measurable and could be written into purchasing specs. We could hold the builder or maintenance to task for allowing any of these to slip. These metrics are for web handling alone. Obviously there are other considerations unique to each process.

The first is mechanical precision. As given in my Mechanics of Rollers book, roller precision can be measured as diametral variation, alignment, deflection and so on. In general, mechanical precisions are important; precisions are held to the thickness of a human hair +/- an order of magnitude or so depending on the parameter, application and web.

The second is control precision. The TNTs of controls (Tension, Nip and Temperature) can all be measured. We could insist, for example, that tensions be held to 5 percent of setpoint during steady state and perhaps 10 percent during speed changes as read by fast-acting load cells. Nip loads can generally be held within 5 percent except on winders where 10 or even 20 percent (nonwovens, tissue) might be tolerated. Temperature could be held to 2-5 deg F at important locations. No less important: The builder must provide measurements in engineering rather than arbitrary units and calibration procedures for all control parameters.

The third measure of quality would be MTBF (Mean Time Between Failures). Here, we mean mechanical breakage or electrical faults. A web break is a messy metric and could not be used meaningfully. For example, we could insist on 200 consecutive, successful flying splices or turret transfers before the builder gets the last check. (You did hold back payment pending successful startup, didn't you?)

The last metric is not so measurable and is only recently acknowledged by web handlers; it's tolerance, meaning the fewest number of rollers (and components) as possible. It also means avoiding excessive traction. Tolerance is a different concept than flexibility for running different web types.

We must learn to distrust the common sense of comparing one machine with another. Instead, we must replace that sense with a science of controlled, comparative experiments with a dispassionate statistical evaluation. Lacking that, we can only turn to the easily measured parameters suggested here and which are completely determined by the machine rather than messily mixed with material and operation.

920/312-8466 drroisum@aol.com, www.roisum.com