Web Handling & Converting Blog

Interested/Disinterested Science

Science and troubleshooting have much in common. There is a tremendous interest in learning the truth. There must also be, however, no interest in the specific outcome. If you let your pet ideas run totally free, they will likely tear up the flowers. Pets must be reined in with the stout leash of facts, double-blind tests and statistics.

Did Blown Blow It?

A starring pattern that differs on the two sides of the same wound roll is often due to a profile problem in manufacturing; i.e. gage variation. (Other causes are usually easy to distinguish such as an extremely crooked winding machine that would yield consistent results and dropping one side of a roll). So let us suppose you have two materials bonded together: film A and film B. A working theory might be that one of the two web materials or one adhesive layer is uneven, causing the roll to collapse (star) in a local area where the relative gage is ever-so-slightly smaller. One other observation is that the starring is not the classic pattern involving the bottom of the roll. Rather, it is vividly banded and these bands vary in width and position for every roll and are different on all sides of all rolls within a set. Further let us constrain the problem even more. Let us say that this is a medical grade so that substituting raw material supplier is not permitted. (Swapping materials is a strong problem solving technique because if the problem moves with the swap, it is that material). Who is at fault?

How Much Gauge Variation is Too Much?

Often, less than you can measure on the web. Gauge variation that can cause the web to be stretched into bagginess might range from less than 1% to occasionally more than 10%.

Alternatively, we may use the wound roll itself to measure the accumulated variation. The wound roll is often the most sensitive measure of gauge variations. More importantly, the wound roll is often the fussiest customer for gauge variation. More about this topic can be found in my 2001 paper ‘Secrets of a Level Process and Product’ that can be downloaded from my website, Roisum.com, as well as through the organizations that hosted the talk such as AIMCAL (members only) and TAPPI.

You may not, however, need to go measuring stuff. If the finished rolls are inconsistent across the width, then it is very likely that the web is as well. If the winder operator complains that the web runs like crap, it may be that the web is crap. If the end-use customer is complaining that the roll looks and runs like crap, it is possible that the root cause is in fact that the web is crap. More specifically, the web has a profile problem.

Cantankerous Coater-Laminator

Let’s do some problem solving. Let us say you have two different substrates that you bond together to make a structure. The problem is bagginess on the front side first seen after unwinding the now laminated wound roll. The problem is perennial. This means that while it may vary in magnitude, sometimes even disappearing from complaint, it stubbornly likes to misbehave just inside the front edge of the composite web. On close inspection you note that the diameter of the roll is ever so slightly larger there. The easiest way to see this is with a straight bar, it rocks over the tiny bump. You may even notice a relatively higher roll hardness reading via Rhometer, Schmidt Hammer or ParoTester. So whose fault is it?

Streakin

Profile means the variation of ___ across the width of the web. Most web manufacturing and converting processes have a profile that tends to be stable with time as measured in minutes, hours, days and even decades. Yes, decades. Some machines have a smile or frown shape from the day they start up to the day they are dismantled. Textiles with their selvedge edges are just one example of edge effects that are a permanent feature of that product no matter who makes it, how it is made and no matter what machine it is made on. That is not to say that the web cannot have down stream or with-time variations. It is just that time variations are usually the smaller of the two when viewed from a sample size of a roll or lot. We could use the word profile shape or fingerprint to describe this dominant cross machine variation. However, it is just simpler to say that webs tend to be ‘streaky.’

The problem is that streaky webs tend to misbehave, particularly when wound.

Secret WH Chapter in Coating/Drying Defects Book

I'm working on developing a seminar focused on problem solving the most common web handling problems. During a web search for some relavent insights on coating chatter and speed variations, I stumbled across Gerald Kheboian's chapter in the book Coating and Drying Defects: Troubleshooting Operating Problems (by Edgar Gutoff and Edward Cohen).

Gerry does a nice job in Chapter 11, "Problem Associated with Web Handling," of covering the importance of web handling to good coating and drying. This chapter is almost 70 pages long with over 20 figures and tables. If you are looking to add to you web handling library and practical knowledge, this chapter alone is worth the price of admission (<$150).

The book is easy to find at Amazon or Barnes&Noble online and also available for eBook download (http://ebooks.ebookmall.com/ebook/229510-ebook.html).

I took the seminar offered Gutoff, Cohen, and Kheboian in the early 90s while I was at 3M (it's still available today).

- Timothy J. Walker

Spreader Friction

Does changing the friction of a spreader affect its operation?

Most spreaders can be neatly divided into two categories: those requiring traction and those that operate in sliding. Of the more than one dozen spreaders, only two span those modes effectively.

So if, for example, you have a bowed roller operating in the bowed roller mode (bow pointed along the sheet) and your friction drops, you could very well lose it for spreading. It would be lost on the ends first because that is where the greatest amount of sideways movement is required of the web. Traction is absolutely required for this spreader and it will not operate whenever and wherever you break loose. This is why attention to details (such as glazed covers, binding bearings, improper drives and many other vital considerations) delineates spreading from contracting; success from abject failure. The devil truly is in the details.

Sliding in web handling, on the other hand, tends to be more tolerant. This is also true of spreading. If you have a d-bar, for example, it is already sliding. We would not expect that changing friction coefficients (by changes to either web or bar) to noticeably affect spreading. (However, it would affect other things such as the magnitude of tension increase across the bar, possibly wear etc.) The same is also true of the bowed roller operating in the bent-pipe mode (bow pointed up into the sheet) because the device is operating in an entirely different mode even though the machinery may be identical.

Find out how spreaders work and what you must do to make them work and how to troubleshoot them when they do not work. In fact, find out this and many other must-know areas of web handling. Join me March 10-13 in Cleveland.

Mean Time Between Failures

MTBF stands for Mean Time Between Failures. It is an objective measure. Time is, well, time; measured in days or weeks. Failure is, well, failure; it doesn’t run. Since the machine builder has nearly complete control of many of the failure modes in the initial phase; why not specify it in a purchasing contract. (See previous post that claims that initial failures are primarily due to bad machine design, construction or startup.) The last check will not be paid until the machine does not fault out electrically or break mechanically for ___ hours. If it breaks, the builder will fix it at his or her own expense and the clock starts over. Obviously, a machine shutdown due to operator goof up or other factor in the plant’s control will not restart the clock.

MTBF Definition

The Bathtub Curve

The Weibull Distribution, more colloquially known as the bathtub curve, gives you the failure rate of a machine as a function of the life of the machine. The failure rate on the Y-axis might be measured in units such as breakdowns per week. The X-axis is time and spans from the very first production to when the machine is finally retired perhaps years or decades from now.

The Weibull distribution is known as a bathtub curve because of its shape. It has three phases: infant mortality, normal wear and wear out. The initial period has a high failure rate due to infant mortality. Here, poor design or manufacturing shows up when you fire up the machine. Indeed, electronics are often ‘dead right-out-of-the-box.’ Soon or eventually these defective parts will be replaced and the machine enters something of a steady operation with the occasional failure. Finally parts of the machine will start failing at an ever-increasing frequency. On a car it may be time to think about replacement before maintenance bills exceed the worth of the vehicle or the cost of a new car lease.

Does this apply to web machines? Of course it does. Machine builders have complete control over the first phase and some control over the last two. The plant has significant influence over the middle phase and a lesser amount over the last phase.

http://www.weibull.com/hotwire/issue21/hottopics21.htm
http://www.weibull.com/hotwire/issue22/hottopics22.htm

Necking

Like any elastic material that is stretched, a web will neck down (contract in width slightly) under tension. For typical webs, at typical tensions, this change is too small to see. But, anyone who's studied mechanical engineering knows it's there.

 

The change can be calculated by something called Poisson's ratio. This is the ratio of the change in width to the change in length and it's a pure number (no units) ranging from 0 to 0.5. [Nonuniform materials like nonwovens and textiles are exceptions. They can have much larger values.] For plastics it's typically in the neighborhood of 0.3. So if a web is stretched lengthwise by 0.1% (a typical value for process lines) it shrinks in width by 0.3 x 0.1% = 0.03% in width. In a 60 inch wide web that's only 0.018 inch - way to small to see without some good instrumentation.

 

Textbooks often illustrate the Poisson contraction with a sample of highly elastic material stretched between two clamps in an Instron tester. The sample will have a slight hourglass shape - narrow in the center and wider at the ends. So, it might be natural to assume that a web span also has an hourglass shape. Most of the time it doesn't.