What is the best winding curve?
David Roisum, Ph.D., Consulting Technical Editor -- Converting Magazine, 9/1/2008 12:00:00 AM
The best winding curve is one that straddles two types of defects at two different ends of the wound roll. Tight defects include some blocking, one type of core crush and many others. Loose defects include one type of loose cores, out-of-round rolls and a few others. Some defects don't respond to changes in tightness, such as cores not aligned with the roll edge, or product/process design issues such as blocking. In these cases, winder adjustments would be a waste of time.
The best winding curve begins by looking at the core area and asking, “Do I see more tight defects or more loose defects?” If more tight defects, then I loosen the core area and vice versa. For this example, no changes are made at the outside because there's nothing wrong with the it. Then I look at the outside and ask the same question, “Do I see more tight or more loose defects?” Based on what you see, you'll either tighten (more loose defects), loosen (more tight defects) or make no move (no defects that favor the outside.)
This simple strategy works for most problems except the global defects of one type of starring and a couple of the telescoping types. For these, the best curve is maximum taper. However, in the lingo we've just described, it would be best to start as tight as possible at the bottom and finish as loose as possible at the top. This strategy is simple because it just requires the operator to know whether defects are tight, loose or global and whether defects favor the bottom or top of the roll.
Now comes the real challenge. Winder builders have not always made programs that accommodate a simple strategy. In past times, controls were too limited. One didn't have much adjustment and no compensation for geometry and gravity. Now, overwhelming complexity is common. On some rewinders, I recently documented up to 14 adjustments available for winding tightness. In addition, there were dozens of other parameters that were part of the hidden math inside the PLC. How many knobs do you want to give the night shift to avoid what might be a half-dozen defects on hundreds of different grades? Often, the winder computer can make a near-infinite combination of tension, nip, torque and speed curves!
To make matters worse, the builder does not often include calibration methods so that one can ensure that once a good curve is found, it can be reproduced next year or on the next machine. Finally, almost no advice can be found on how to adjust the curves for any defect. This state of affairs has left operators and engineers alike totally baffled.
My proposal: First, get rid of knobs on the benchboard and replace them with transparent and calibrated math in the computer. The most the operator will normally need is a starting and ending tension and nip. All complications should be inside the computer and well documented in the user's manual. The manual should include suggestions to guide the process engineer in setting up grade recipes. The manual should include test points and procedures, such as cylinder pressure, to make sure, for example, that a zero-nip setting is really zero and a nip of two lb/in. (or kN/m) is really two.
920/312-8466, drroisum@aol.com, www.roisum.com
