Blog

Regulator Tuning – Metrics

At this time, we do not have good statistics for the speed and tension regulator responses used in web handling.

Recently I performed a drive audit on a line. I found that the slitting tension zone had a load cell tension regulator tuned for a Bandwidth (BW) of 0.5 radians/second. That is a time constant of 2 seconds. A tension disturbance will be settled out in 8 seconds.

Right next to the slitting zone (on the same line) was a calendering zone with a load cell tension regulator. This was tuned with a BW of 0.001 radians/second. This regulator would calm down a tension disturbance in 4000 SECONDS (66 minutes). In that time, this line produces 50 rolls.

Something is wrong and we don’t know what. Is the calender tuned correctly and the slitting section wildly over-tuned? Is the slitter tuned correctly and the calender far too slow? Are both sections far too slow?

Years ago, a major paper slitter/winder vendor scolded me, indicating the drive supplier I worked for was always “far too aggressive” in tuning their precious equipment.

Many times I have been chided by operators because the slow tension regulator was giving “bad starts”.

Note that most drive suppliers and many equipment suppliers have internal specifications for tuning drives for specific applications. They may tune all sections the same as the slowest section, all as fast as they can be, duplicate a previous contract or use some other guidance.

In order to bring some sanity to this situation, we need metrics. It is difficult to get these metrics because the tuning parameters are proprietary. It may be no one has measured the values. It may just be that no one knows.

I have been on my knees begging you readers that tuning information on some real-world lines be made available. I promised to collate the information and present it through AIMCAL and other web handling forums. No response.

Tuning theory is in every textbook, in drive setup manuals and all over the web. Just because you don’t know how your lines are tuned, doesn’t make that information proprietary.

E-mail me when people leave their comments –

You need to be a member of Association of International Metallizers, Coaters, and Laminator to add comments!

Join Association of International Metallizers, Coaters, and Laminator

Comments

  • “OF COURSE IN THE STEADY STATE, THE SPEED REGULATOR RESPONSE IS LESS IMPORTANT.” Reference to steady state was only setting up initial conditions (t<0) and final state (t>infinity) for the example provided. This is the same process used to setup transient analysis problems in Electrical Engineering 101 courses.

    “AS YOU MENTIONED, THE RESPONSE TO DISTURBANCES IS PRESENT AND IN MY EXPERIENCE CRITICAL.” You are asking how to tune speed loops and referencing speed loop tuning differences between different vendors.

    DISTURBANCES ARE SEEN IN WEB HANDLING LINES WHEN:
    • "STARTING FROM A STOP" (Followed by a speed change)
    • "COMING TO A STOP" (Proceeded by a speed change)
    • "CHANGING SPEED" (Most frequently observed disturbance in any web handling system)
    • "PERFORMING AN UNWIND SPLICE OR WINDER CUTOVER" (Reason dancers are used to handle these disturbances. Tuning controls amplitude of disturbance seen by the dancer.)
    • "CLOSING A NIP OR LAMINATOR NIP" (Reason why nips should be closed prior to starting or these axes must be well designed to have a high bandwidth but the lowest available bandwidth in the web handling system. Thus, all other axes will have their speed regulators detuned to this bandwidth instead of detuning this bandwidth to a lower value.)
    • "INITIATING COATING OR PRINTING" (coating being applied by a coater acts like a lubricant which significantly reduces the impact of starting/stopping the coating action. Printer plate/anilox rolls are driven via motors/gears. Thus, on/off print results in very minor disturbances if any.)
    • "ETC."
    • "IN ALL OF THESE TRANSIENT CASES, THE RESPONSE OF THE SPEED REGULATOR AFFECT THE TIME TO GET TO STEADY STATE. THE RESULT IS OFTEN SEVERE ENOUGH TO CREATE SCRAP OR EVEN A WEB BREAK DURING THE TRANSIENT." Yes but the designer must not fix one problem while ignoring side effects caused by their decisions.

    My example was attempting to demonstrate that an important consideration in web handling systems is the relative bandwidth between nips. Different speed loop responses create their own tension disturbance during a coordinated speed change. If speed loops are tuned the same, no disturbance is observed during a speed change. As you stated in your blog, one drive manufacturer suggests all speed loops should be tuned the same as the weakest speed loop bandwidth. My example supports this statement.

    QUESTION:
    Can you explain the following result? I have a theory.

    While working for an OEM, we were setting up an 8-color flexographic press. It consisted of turret style center-driven unwind. Dancer feedback modulated unwind speed (correction of 10% of maximum speed) and diameter calculation. Dancer loading was air pressure regulated via electronic adjustment. Web was fed into an 8-color press. Next, web passed through a drier and cooling roll. Load cell at output of dryer modulates cooling roll speed. Next, web passes into a load cell modulated draw roll zone. Next, web passes into a gravure printing station. Mechanical draw of about 1% was built into gravure roll. Gravure roll drive was positioned locked to a press plate roll to maintain print registration. Because we were experiencing and unable to correct zero tension prior to the gravure roll, we called in the customer. After reviewing the problem, customer’s operator increased dancer loading (unwind web tension) which almost immediately increased web tension at the gravure roll. No other changes were made to the system. Can you explain this?

  • In response to the comment by Dr. Birmbaum.

    Of course in the steady state, the speed regulator response is less important. 

    As you mentioned, the response to disturbances is present and in my experience critical.

    Disturbances are seen in web handling lines when:

    • starting from a stop
    • coming to a stop
    • changing speed
    • performing an unwind splice or winder cutover
    • closing a nip or laminator nip
    • initiating coating or printing
    • etc.
    • In all of these transient cases, the response of the speed regulator affect the time to get to steady state. The result is often severe enough to create scrap or even a web break during the transient.

    Clarence Klassen

  • There are very few, if any, web handling systems that consist of one-drive. In a single drive web handling system, one could argue any speed regulator response is OK.

     

    Most web handling systems consist of multiple drives. The real question is "how does different speed regulator responses impact the web?"

     

    Simplest web handling system consists of two nips. To analyze this system, let’s assume:

    • Input tension remains constant.
    • For t<0, system has run long enough that all tensions have reached steady state.
    • Draw between input nip and output nip is zero (therefore, input web tension is transferred into tension zone between these two nips).
    • At t=0, speed of input and output nip are stepped/ramped/s-curved by 10%.

    Results:

    • If both drive systems respond with the same response (bandwidth and characteristics (first order response)), the speed difference between input and output nips is zero for all time. Therefore, draw is always zero. Thus, no tension change.
    • If response of one drive is twice the other, there is a speed difference between input and output nip during this speed change. This causes tension changes. After sufficient time, speed difference will return to zero. Thus, tension will return to its previous value.
    • As rate of change of speed command is lowered (step/ramp/s-curved), speed difference is reduced. Thus, tension changes are reduced.
This reply was deleted.