WWW.TEMPERATURECONSULTANT.COM

LINK: SYSTEMS

An Overview on Strip Guidance

Modern control technology has been around for many decades. That doesn’t mean it is applied intelligently. Let’s take two interrelated examples of control problems I was called to consult on: strip steering roll systems and strip coiling systems. These are so common throughout steel mills that you’d think they have been fully understood.

Light gage steel strip usually has a certain amount of camber that changes throughout the length of each coil. Take a section of strip, roll it out onto a flat surface, and you will see that it is not perfectly straight or flat. Blame tandem mill setup if you wish (which is an entirely separate problem). On most strip processing lines, coils are welded together, and the strip camber changes abruptly at these welds, which results in a sharp change in strip tension profile. As a result, the strip can “walk” off the rolls carrying it if steering roll systems are not used to compensate for the camber and/or tension profile changes.

This, of course, has been known for years, and various types of steering roll systems have been installed on almost all processing lines. Some pivot a single steering roll about a center, some lift or shift one end, and some (probably the best) use a pair of steering rolls pivoted at the entry end on a common base. The steering rolls usually are followed by a strip position sensor of some type which provides feedback to tell the steering roll system what to do. If the strip starts to shift to one side, the sensor detects this and tells the steering roll to correct the shift.

However. And this is a big however! What most steering roll system manufacturers provide to the customer (the mill) is a simple PID controller. And the fact (which few seem to know) is that standard PID controllers can be correctly “tuned” only for one set of conditions. There is one line speed and one tension setting for which the “tuning” will be correct. Change either of these and the control system will either overcorrect (hunting or oscillating) or undercorrect. The solution? You do not use a standard PID controller.

You use an imbedded PC with necessary input and output cards to create a controller capable of adapting to line conditions, modifying gains etc. as a function of line speed and tension. The cost for a simple system of this type is not all that much more expensive than the PID, and it can often be implemented on any old computer lying around which might otherwise be scrapped. You don't need a Pentium 4.

If you consider the cost of just one coil that has been ruined by “walking off” a roll and the resultant mill down-time to correct the situation, it becomes amazing that no-one has really addressed this problem. Literally millions of dollars per year could be save through implementing “intelligent” steering roll systems.

Coilers are a bit more complex (and are usually preceded by steering roll systems). The thinner the strip gage is, the better a coiler control system must be as if the coil is flipped onto its side, any thin wraps that stick out will be crushed, causing quality problems. Typical customer specs allow 1/16 inch variation in the wrap.

Coilers, again, usually just use standard PID controllers which can be tuned only for one set of conditions, and conditions for coilers change even more than for steering roll systems. 

Most coilers use a hydraulic system to change the mandrel position. The mandrel structure  is fairly massive, so the rate at which the mandrel can move is limited. It is often forgotten that the mass being moved includes the coil mass which continually increases as the coil is wound. Thus, the system time constant increases with coil size and this affects system stability.

On most coilers, the strip passes over a deflector roll before being coiled onto the mandrel. The amount of “wrap” on the deflector roll starts out fairly high and decreases with coil size. Usually (not always) the strip position sensor is located just prior to the deflector roll, and is (unfortunately) a simple brightness unit, such as an Askania (which, if used, should always be connected to a stabilized voltage source). I’ll have  a few more words on this later. Anyhow, the mandrel shifts position as commanded by the strip position sensor to try to keep the same strip position. If the amount of deflector roll wrap is high, it is more difficult to shift the strip position than if the wrap is low. What this implies is that the “gain” of the control system should be varied with the amount of deflector roll wrap or coil size.  

Also obvious, the gain of the control system should be varied with both strip speed and tension to obtain improved performance.

An obvious improvement to the strip position sensor is to move it to the coil side of the deflector roll. One commercially available sensor to do this uses an extended laser parallel light source which projects a wide, narrow beam to a sensor located on the other side of the strip. A less complicated device of my own invention uses a laser source and TV system located on the same side of the strip. The laser source projects a diverging beam of light onto the strip that exactly compensates for the change of TV field of view with distance of the strip from the TV. The advantage of this system is that it is cheaper and less subject to damage as it is mounted only on one side.

Again, a simple old computer with A/D and D/A cards can be used to cheaply implement this type of system. 

I could go into whether the use of “dither” good in overcoming stiction or cover related topics, but that would make this topic too long for a brief discussion.