Tire cord calender as process system: operating for consistent product--part 2.

Last month's Process Machinery column (Rubber World, July 2005) began a discussion of the calender and the compound that is delivered to it, presenting a review and looking at process and machine variables. In the second part of this article, overcoming and correcting for process variation will be explained, along with the effect of maintenance on operation and quality.

Overcoming and correcting for process variation

The fundamental principal that should guide all calender operations is consistency. We want to keep the roll separating force, and therefore the roll defection, as constant as possible. Some variation is inevitable, and can be anticipated. For instance, we cannot start the calender at line speed; we ramp up to it. The surface temperature of the calender rolls will not come to equilibrium until the line has been running for at least a few minutes. Other sources of change include a change of product, compound and/or substrate.

Other sources of variation are expected, but less predictable, including the occasional line problem that necessitates a slow-down while corrections are made, an interruption in the feeding and warming system, etc. These problems tend to have a multiplicative effect, a slowdown inevitably results in more warm-up time and reduced viscosity of the feedstock, as stock is held in the warm-up equipment. This is followed by a rather abrupt change in viscosity as the over-warmed feedstock is fed into the calender, to be replaced by feedstock that once again is at or near the normal work history and temperature. Other disturbances often follow an unplanned line speed change, including an increase in bank size and poor movement of material in the feed bank, typically followed by near starvation of the bank as the line is ramped back up to speed.

Perhaps last, but certainly not least, variation in the feedstock process variables will produce changes in roll loading.

Our objectives should be two-fold. We want to minimize the changes in roll loading by minimizing the variation in the feedstock, the way the feedstock is warmed and delivered to the calender, the size and shape of the feed bank, and in the various machine variables that affect roll loading. Secondly, when change occurs, we want it to be low in amplitude and long in period.

Over the last 50 years, great strides have been made in our ability to measure the thickness of the sheet that the calender is producing. All of our measurement technologies have limitations. Noise is always present. In its simplest form, noise is apparent signal collected along with the actual data. We would like to measure accurately differences in thickness that are on the order of 5-10% of the total thickness; in English units, a half a mil of less. It does not require a very high noise level to make it difficult to separate the signal from the background.

Depending on the measuring technology, we may be measuring thickness (actually, variation in distance between a source and the surface of the rubber on a roll); mass, measured as the loss in energy of a reflected beam of gamma or beta radiation; the change in capacitance between a metal surface and the roll, resulting from variations in the dielectric characteristic due to changes in the amount of rubber compound between the two. A moment of reflection tells us that quite a few things might affect the readings produced by these techniques, some being due to real differences in thickness or weight coat, others due to changes in density of composition, or simply surface roughness.

The point of all this is that for any measurement to be useful, the measurement signal must be filtered and averaged. We are looking at trends, not true instantaneous thickness measurements. This is very definitely a good thing; the localized and regional short-term variation in weight coat thickness is not something we want to try to correct, at least not with the calender. We are concerned with adjusting for trends, for steering the measured average thickness steadily between two closely spaced bounds, the upper and lower limits. The more consistently we can control the roll separating force, the tighter our practical upper and lower limits can be.

Most on-line measuring systems measure and control three zones, including water end edge, center and drive end edge. All three are really bands; in the usual mode, each comprising roughly a third of the width of the sheet. (This statement assumes a single scanning sensing head. Some calenders are equipped with three fixed-point measurement heads, which, of course, do not scan, and whose measurement band is a tight, narrow strip, much less than a third of the sheet width.) The sensed data are added to the last several measurements for that zone or point, the oldest data point in the series is dropped, and the result averaged. If the result is within tolerance, nothing else happens. What happens if the measurement is out of tolerance depends on the device, the built-in logic and how any user programmable variables are set. It may be nothing until the end of specification condition has persisted for a certain number of scans, or it may immediately initiate a control action.

Prior to a control action being initiated, the measuring system control logic will have made some calculations. No change in roll positioning of crown compensation is truly independent. Opening or closing the roll adjust at one side of the calender will pivot the roll about the centerline of the main bearing on the opposite end of the calender; in other words, opening the water end of the nip a bit will also open the drive end, a much smaller amount, but it will still open a bit. Increasing or decreasing roll crossing will also change the openings at both the water and drive ends of the roll as well. Some control systems calculate the likely effect of the anticipated control action, and output simultaneous corrections to offset the undesirable of unwanted change.

After the control action is initiated, there will be a time delay before it is accomplished. There will be an additional time delay before the result of the change appears at the measurement device. These delays must be accounted for in the measuring system logic, in order to avoid initiating a second correction before the effect of the first can be measured. If the measured variable is fluctuating with a relatively short period, or in a random fashion, it is entirely possible that by the time the correction has taken effect, the disturbance that led to it will have passed. When the time constants are such that this occurs frequently, the control system becomes part of the problem, rather than part of the solution.

The majority of the calenders now in existence make use of some form of nut and screw to position the rolls. The nut and screw are driven through a gear train, usually comprised of a two-stage worm gear reduction system. These systems are capable of making very small and precise movements, but by the nature of their design and construction, there are lags in the response of the roll to position corrections. Fundamentally, the necessary clearances in the system mean that when a change is made in the opposite direction of the previous correction, the backlash in the gear train must be reversed. The friction between screw and nut, and in older designs, between the end of the screw and the bearing box, produces a certain amount of wind-up in the system; energy and torque are necessary to initiate movement that unwinds as movement is accomplished. Over the years, many improvements were made in these systems to minimize these effects, including tighter clearances in the gear train, antifriction bearings between the turned device and its bearing surface, the use of brakes to reduce coasting when the drive motor is stopped, antifriction bearings between bearing boxes and box windows, among other examples.

All of these issues apply equally to the roll crossing (cross axis) system. For both the roll crossing mechanism and the roll nip or gap adjustment system, frequent changes accelerate the wear on mechanical elements, increasing backlash and slowly degrading response time.

In recent times, hybrid roll adjust systems have appeared, making use of a screw and nut for gross positioning and short-stroke hydraulic cylinders interposed between the screw and the bearing box for fine adjustment. Such systems have been explored for over 50 years, and employed for a variety of reasons, including the ability to very quickly open a roll gap the full length of the cylinder stroke; the ability to pass a fabric splice without adjusting the roll positioning system; and the ability to laminate in a constant pressure mode. Positioning the roll precisely using the hydraulic cylinder in the roll adjust stack-up, or even singly, with no mechanical adjustment ability, has always been attractive. In theory, it offers the possibility of an almost immediate and stepless response to a corrective action, and offers the possibility of combined functions in a single package. Friction in the adjusting system is reduced to the seals in the cylinder. In practice, there have been a number of issues to overcome.

Such systems include sensors for roll position and hydraulic pressure, logic to filter these signals and determine appropriate output, and output devices to adjust hydraulic pressure and modulate roll position. Sensors have effective ranges, sensitivities, resolutions, response times, environmental (temperature, electromagnetic noise, etc.) and signal to noise ratios to consider. Output devices have similar issues, including hysteresis, response times, accuracy and repeatability, among others. The logic itself is based on certain assumptions and models that predict the expected outcome from a given control action. Designing and tuning these systems so that they reduce product variation rather than induce it requires considerable experience, a recognition of the limitations of the measurement and control devices employed, and the willingness to spend considerable time investing the theoretical with the empirical.

Despite these difficulties, the latest tire cord calenders employ these systems, and the issues have been overcome to the extent necessary to keep the lines running successfully.

The effect of maintenance on operation and quality

It should be clear from the previous discussion that the subsystems that respond to gauge correction actions are complex bits of mechanical and hydraulic components. Unless they are functioning correctly, response will be erratic, unpredictable and not repeatable, rendering the most sophisticated gauge measurement system worthless. Similarly, the most mechanically perfect calender will produce useless material if the gauge measurement system is not working correctly.

The primary cause of malfunctioning roll adjustment systems is lubrication failure, usually between the screw and the nut, and with lesser frequency in the lifter assembly that couples the adjusting screw to the bearing box. Unlike lubrication loss to a main bearing, which manifests itself as a hot bearing or a bearing rapidly shedding metal particles, lubrication failure in the roll adjust system is more invidious, and hard to detect. The reduction ratio in the roll adjust system is very high, high enough to break things if something stalls the drive train. Consequently, quite a bit of damage can be done before a loss of lubrication is apparent.

Hydraulic or composite hydraulic and mechanical roll adjust systems need clean, well filtered and conditioned hydraulic fluid. Pressure, directional and positioning valves have close clearances and do not cope well with dirt or particulate matter. Leaks in the system must be attended to, particularly leakage at the cylinders.

In either type of system, sensor operation must be checked and calibrated on a regular basis. The devices themselves must be protected from mechanical of thermal damage.

The gauge measurement system must be calibrated on a periodic basis, and appropriate preventive maintenance performed to keep it operating correctly.

Our focus has been on the measurement and control of the gauge or thickness of the rubber coating, and on the systems on the calender that directly adjust the roll nip. Every system is a contributor to the performance of the machine. Roll bearing, bearing preload and/or pull back systems, and the bearing lubrication system all contribute to keeping the roll rotating at or near a constant center of rotation, a key issue in maintaining control of the gauge. The drive train must maintain constant roll speed, of fluctuations in the gauge will result. Put another way, the equipment must be maintained properly to insure optimum performance. It is not enough that the rolls rotate.

Summary and conclusions

The sources of variability in product output can be divided into two broad categories, including variation in the feedstock and variation in the operation of the calender. Variation in feedstock requires the calender to make adjustments to compensate for the variation. If we minimize the need to make compensatory adjustments, we reduce dependence on measurement, control and mechanical systems.

Variation in the way the calender line is operated has several roots: One is production scheduling; short runs and frequent product changes are an anathema on stable operation and uniform gauge. Another is poor reliability in any system that supports the operation of the line. Anything that requires slowing or stopping the line to correct will result in the production of out-of-specification product, at least while the gauging system (or the operator) corrects for the problem. Another source of variability is poor operating discipline, usually the result of inadequate training. It is not enough that the operators know what to do, they should also know the why and the how. Understanding how speed variation, bank size, warm-up time and other variables at least partly under their control affect the quality of the product is essential. They must be part of the quality team, not simply tools of it. Without this knowledge and understanding, they will make decisions that keep them out of big trouble, and blame the little trouble on other things. For example, they will tend to keep too large a feed bank, accepting gauge control problems in exchange for reducing the risk of bare spots on the substrate.

The successful operation of a tire cord calender line rests on the recognition that the line is a composite of several systems, and that each of those systems, including the operators and foremen, must be functioning correctly for optimum performance.

The important points are as follows:

* Delivery of consistent feedstock to the calender is paramount. All else is avoiding further variation and attempting to correct for variation at the input end of the process.

* The measurement and control functions of the on-line gauging system must be matched to the capabilities of the calender roll adjust and crown correction systems. It only creates excessive wear and induces product variation if the gauge control system attempts more correction than the system can handle. A new gauge cannot compensate for a worn calender, of the absence of certain important features in an older design.

* The calender, and all its support systems, must operate correctly. Erratic or unpredictable operation of any system, subsystem or component may either induce a variation in the product or make it difficult of impossible to make an appropriate adjustment.

* The calender line is in dynamic equilibrium with the process and its environment. It achieves that equilibrium after it has been in operation for some period of time. The more sophisticated and responsive the elements of the calender and train, the more rapidly that equilibrium can be attained. Changes in operating conditions shift the equilibrium point.

* The pre- and post-calender train and the feeding and warming system must operate reliably. Slowdowns and unplanned stoppages will result in scrap.

* Operating personnel need to know how and why each element of the line contributes to the production of on specification material. Equipped with this knowledge, they will make better decisions with respect to the manner in which the line is operated.

In the majority of the existing tire plants around the world, the production of cord and wire ply material for building the tire is carried out in the plant. Quite often there is only a single calender line for that purpose. The entire productive output of the plant rests on the reliability, quality and consistency of the product produced on that line. Attention to its reliable and proper operation is essential, not only to the productive output of the plant, but also to the job security of each and every person who works in that plant.

Lawrence R. Gooch,

Gooch Engineering Associates