WO2006039426A1

WO2006039426A1 - Control method for a static belt reel tension paster

Info

Publication number: WO2006039426A1
Application number: PCT/US2005/035072
Authority: WO
Inventors: Byron Mayyou
Original assignee: Goss International Corporation
Priority date: 2004-09-29
Filing date: 2005-09-29
Publication date: 2006-04-13

Abstract

A feed-forward tension-control system [29] that can be used for controlling the tension in the web [12] in a commercial printing line. The system has a controller [60] that is programmed to calculate a desired pit tension for a pit cylinder [28] using a tension set point and a formula based on the instantaneous wrap angle of the static belts [20] around the roll [10] that the web winds from. A paster-carriage-position indicator [36] may affect the equations used to make the pit tension calculation. The wrap angle is calculated using input from a roll-diameter indicator [41]. The roll-diameter indicator [41] calculates the roll diameter from input from a press-speed sensor [30] and a revolution sensor [32], and may be trimmed based on input from a dancer-position sensor [34].

Description

CONTROL METHOD FOR A STATIC BELT REEL TENSION PASTER

BACKGROUND OF THE INVENTION

[0001] This disclosure relates to static belt reel tension pasters (RTP 's) used in commercial printing, and more particularly to a system for controlling the tension in the web of material to be printed.

[0002] In conventional printing operations, the web of material to be printed is wound from rolls. The paster of an RTP is used to splice the start of one roll to the end of an expiring roll end without the need to shut down the line, and also to add tension to the web of material.

[0003] RTP' s use belts that extend around part of the roll to provide tension in the web. In a static belt RTP system, the belts remain relatively motionless, applying a frictional braking force on the spinning rolls. The amount of friction (and, accordingly, the amount of tension in the web) can be controlled by a cylinder that presses against a length of the belt, tensing the belt.

[0004] Conventionally, control of the amount of force applied to the belt has been tied to observations in the position of a "dancer" that is positioned upstream of the rolls. The dancer is a resiliently-mounted roller that the web winds around in its path from the rolls to the printing press. Increased tension in the web causes the dancer to move in one direction, shortening the path of the web. As tension in the web decreases, the dancer moves in the opposite direction, lengthening the path of the web. Thus, movement of the dancer provides an indication of the tension in the web, providing feedback that can be used to adjust the amount offeree applied to the belts.

BRIEF SUMMARY OF THE INVENTION

[0005] Unlike conventional systems that control belt forces based on feedback from the dancer, this system uses a "feed-forward" system that attempts to predict the appropriate amount offeree to be applied to the belts based on current operating conditions and parameters.

BRIEF DESCRIPTION OF THE DRAWINGS

[0006] The invention may be better understood by referring to the accompanying drawings, in which:

[0007] Figures 1 and 2 are perspective and schematic views of an RTP in which the new control system can be used.

[0008] Figure 3 is a block diagram of one embodiment of a feed-forward control system in accordance with this invention.

[0009] Figure 4 is a schematic of the angles between various structural elements in one example of an RTP.

[0010] Figure 5 is a schematic of the angles showing the influence of movement of the paster carriage on the calculations used to estimate pit pressure in the system of fig. 3.

DETAILED DESCRIPTION

[0011] This discussion will begin with a description of the mechanical structure of a typical tension-control system used for commercial printing, then provide an overview of the new feed-forward system for managing that kind of system, and finally provide a detailed explanation of one management system utilizing such an approach.

[0012] I. The mechanical structure of a tension control system

[0013] As seen in figs. 1 and 2, a typical tension control system used in a printing process includes a roll 10 of the material to be used in the printing. The roll can rotate around its axis, allowing a web 12 of the material to wind off the roll. As seen in fig. 1, the illustrated roll is carried on rotating reel 14 that also carries a second roll 16. As one roll expires, the system pastes the start of a new roll of material to the end of the expiring roll, allowing continuous operation of the printing line.

[0014] The web 12 follows a path that ultimately leads to the presses (not shown).

As seen in fig. 2, that path includes a trip around a dancer 18. As discussed above, the dancer is a resiliently-mounted roller that can move laterally, thus changing the length of the path that the web travels from the roll to the presses.

[0015] Primary control of the tension in the. web is achieved through static belts 20 that wrap around a portion of the surface of the roll 10. Although the precise arrangement can vary, the illustrated belts have upper and lower ends 22 that are fixed to the frame 24. As seen in fig. 2, a portion of one of the illustrated belts wraps around a pit pulley 26 that is connected to a tension-control actuator 28. In this illustration, the tension-control actuator takes the form of a pit cylinder, though other types of actuator could be used. The pit cylinder and pit pulley apply force to the belt and, consequently, on the roll. The force on the roll in turn adds tension in the web 12 feeding from the roll. Varying the force applied by the pit cylinder therefore allows the operator to control the amount of tension in the web.

[0016] II. Control system description

[0017] The fundamental concept of this tension-control system for a static-belt

RTP is based on a feed-forward control strategy. A feed-forward control strategy is one that attempts to predict what the control point should be for a given set of operating conditions and parameters. The predicted value is then summed with critical feedback to develop an actuator control point. This differs significantly from a pure feedback system, such as a system that attempts to control the pit cylinders based solely on the dancer position.

[0018] With this design, the major control point is the pressure applied by the pit cylinder 28, the "pit pressure." This design works by estimating what this pressure should be based on variables such, as the angular extent of the surface of the roll that is in contact with the belt (the "belt wrap angle") and the desired tension in the web.

[0019] Previous designs have relied on the dancer 18 as the major command/feedback element. However, there is no direct relationship between the dancer position and the pit pressure. Consequently, the illustrated arrangement only uses feedback from the dancer position to "trim" the estimated pit pressure. In this design, the dancer is viewed as a disturbance, not a demand set point. In fact, this control system's adjustment of the pit pressure to influence the speed of the roll may be used to try to minimize movement of the dancer. This system can control movement of the dancer, rather than being controlled by

it.

[0020] In order to understand this static-belt RTP feed-forward design, it may be useful to understand the system conditions and parameters that influence the pit pressure. Preliminary analysis suggested that the major factors are the tension set point and the belt wrap angle. However, it may sometimes be beneficial to fold additional parameters of influence into the design. In the illustrated embodiment of the invention, press-speed changes, paster-carriage impact, and coefficient of friction are all used in calculating a desired pit pressure.

[0021] III. Example system

[0022] Fig. 3 outlines one embodiment of a feed-forward control system 29. As discussed in more detail below, the illustrated system estimates the appropriate pit pressure by combining input from a press-speed sensor 30, a revolution sensor 32 on the expiring roll, a position sensor 34 on the dancer, and a paster carriage position indicator 36 with a tension set point (obtained from a tension set-point source 38) and information about this number of belts in use (obtained from a belt indicator 40). [0023] Ultimately, tension in the web 12 is controlled by adjusting the force that the static belts 20 transmit to the roll 10. That force is a function of the pressure of the belts, surface area of the roll contacted by the belts, and the relevant coefficient of friction. The surface area of the roll contacted by the belts is a function of the angular extent of the circumference of the roll that the belts wrap around (the "wrap angle"), which in turn is a function of the changing circumference of the roll.

[0024] Briefly, the illustrated system 29 works by first using a roll diameter indicator 41 to estimate the circumference of the roll. This estimated circumference is then adjusted based on input from the dancer-position sensor 34. The corrected roll diameter is used to calculate the wrap angle. Knowing the wrap angle and the number of straps in use allows an appropriate pit pressure to be calculated. Each of these steps will be discussed in turn. _;

[0025] (A) Roll circumference estimate

[0026] As noted above, the pit pressure required for a given tension in the web 12 is a function of the tension in the static belts 20 and the area of contact between the belts and the roll 10. The area of contact is a function of the wrap angle. For convenience, wrap angle is generally calculated indirectly, rather than directly measured, and is a function of the circumference of the roll.

[0027] Although other arrangements could be used, the illustrated roll diameter indicator 41 estimates the roll circumference using a counter 45 to count the number of press pulses that occur during a revolution of the roll 10. In the system illustrated in fig. 3, the counter 45 counts the number of press pulses that are sensed by the press-speed sensor 30 over the period that the revolution sensor 32 indicates is necessary for the roll to complete one revolution. Knowing the number of press pulses that occur during a revolution of the roll and the linear length of the web used during each press pulse, a simple calculation can be made to estimate the linear length of the roll circumference.

[0028] (B) Roll circumference correction

[0029] The simple calculation described above may not be perfect. While the press is running at a constant speed, changes in the dancer location may cause the expiring roll to temporarily speed up or slow down compared to the speed of the web through the presses. As a result, a calculation of the roll circumference that uses the length of the web traveling through the presses as a measure of the length of web winding off the roll may not be accurate when the dancer is moving.

[0030] It is possible to compensate for dancer movement. The amount of error introduced by dancer movement can first be calculated using the simple relationship relating angle (radians) to arc length (inches), S=Γ_DΘD.

[0031] In the arrangement illustrated in fig. 3, if the dancer 18 is a 6" diameter roller that can swing through a range of approximately 50 degrees on a 6" long arm 50 (see fig. 1), then the potential error in the calculated circumference of the roll is:

[0032] ΔCirc = | — |*ΔΘn Eqn 1

[0033] where ΔΘ_# is the change in the angle of the dancer 18 on the arm 50.

[0034] This analysis suggests that the worst-case error that would arise from end- to-end travel of the dancer 18 during the period of measurement would be a distance of -10.5" of web travel. In other words, if the measurement occurs while the dancer is moving from the extreme forward position to its extreme rearward position, the roll will have wound off 10.5" less web than actually used by the presses. If the measurement occurs while the dancer is moving from its extreme rearward position to its extreme forward position, the roll will have wound off 10.5" more web than actually used by the presses. Since the calculation of roll circumference described above uses the amount of web used by the presses as an indication of the amount of the web wound off the roll, this error in the assumed length of the web wound off the roll can provide an error in the calculated roll circumference. For the illustrated geometry, the error results in an error of ±3.5" in the calculated roll circumference. (This error not only affects web tension control, but can also result in error in final butt size or in web loss if the circumference is perceived to be greater than the core circumference when the core is actually reached.)

[0035] (C) Correction for reel shaft movement

[0036] A similar error in the calculated circumference of the roll can be introduced whenever the reel 14 on which the roll 10 is mounted is rotated within the frame 24. If the reel is rotated away from the paster carriage pivot shaft, the calculated roll circumference will decrease as a result of the roll speeding up to make up an increase in distance. If the reel is rotated in the other direction, the roll will slow down, introducing an opposite error in the calculated roll circumference.

[0037] . This reel-shaft-induced error may be smaller than the dancer-induced error, hi the illustrated system, it is ignored. In other cases, an appropriate correction may be made using the same kind of analysis discussed above to correct for dancer movement.

[0038] (Dϊ Wrap angle calculation

[0039] Figure 4 represents the geometric relationship between the roll 10, the static belts 20, the shaft 52 of the reel 14 (fig. 1), the shaft 54 of the paster carriage, and the pit pulley 26. In order to use the roll diameter to calculate a belt wrap angle, it is useful to understand the relationship between these points.

[0040] The following lengths were measured in two RTD 's, one with a 42" reel and the other with a 45" reel. For other RTD's with reels of other sizes, different lengths may apply. Similarly, different lengths may apply for specific RTD 's if there are variations in reel shaft height or in other fixed parameters.

[0041] It is desirable to calculate the relationship between wrap angle ®_w shown

in fig. 4 as a function of the roll diameter Φ_a . hi the illustration,

®_w = 2π - (Θ_j + Θ₂ + Θ₃ + Θ₄ ) . Given that the reel geometry can be represented as a series

of triangles, the fundamental law of cosines for a general triangle is given by the equation

C² = A² + B¹ - 2AB cos(Θc). This equation can be used to determine Q₁ , Θ₂, Θ₃, and Θ₄.

[0042] From the basic equation, the length between the reel shaft and paster carriage center can be found:

[0043] PC² = PR² +RC² -2PRRC∞s{®_R -Θ_P) Eqn 2

[0044] Although the center of the paster carriage pivot shaft is used here, in actuality the point of interest lies on its surface. However, given the relative size of the shaft diameter relative to the distance between the reel and the pivot shaft, the error induced with this simplification is minimal.

[0045] The angle between a vertical line through the center of the reel shaft and a line from the center of the reel shaft to the paster carriage center is:

[0047] Substituting Eqn 3 into Eqn 2 and taking the square root yields:

[0048] PC = J Pi?² + RC² -2PRRC cos ®_R -cos^"1 S- 1 1 Eqn 4

[0049] From Eqn 4, angles 1 and 2 can be derived:

+RC² -PR⁷

[0051] Θ₂ = cos -1 PC²

Eqn 6

2PCRC

[0052] Repeating this process with the distance between the center of the reel shaft 52 and the center of the pit pulley 26 yields:

[0053] CB² = RC² +RB² -2RCRBcos{π-®_R -®_B) Eqn 7

[0054] Again, use of the center of the pulley 26 rather than the actual point of interest on the surface of the pulley 26 induces a small error. Nonetheless:

[0055] ®_B , so Eqn 8

[0056] Eqn 9

[0057] From this equation, Θ₃ and Θ₄ can be determined:

[005 and Eqn 10

[0059] Θ₄ = cos .-^'1¹I M ^r~= . Eqn 11 [0060] Given these equations and the fixed geometric lengths for a 45" reel, the wrap angle can be plotted against the reel shaft position, giving the following graph at different roll diameters:

Graph 1

[0061] In this example, if the reel-shaft angle is near 110 degrees for the illustrated 42- inch reel during normal operation, there is a generally linear relationship between roll diameter and wrap angle (in radians) for that unit:

[0062] θ_r = — (1.218 Φ_α +23.91) Eqn l2

180

[0063] For a similar 45-inch reel at a 90 degree reel shaft position, the equation can be simplified as:

π

[0064] ® 'Ww = \ Vl. l«6 Φ_β +27.32) Eqn l3 .180

[0065] As described below, the appropriate equations can be used in controller 60 for calculating a desired setting for the actuator 28. Potential errors in these equations for the wrap angle based on the corrected roll circumference calculation are discussed below. Although movement of the dancer 18 can compensate for some of these errors, it may sometimes be desired to keep the influence of the dancer as small as possible. The potential sources of error due to the wrap angle estimates and conversion can be reviewed so that use of the dancer to correct or compensate for errors can be reduced. As larger errors are introduced, increased influence of the dancer may sometimes be required. In the illustrated system, this is not preferred.

[0066] Sources of potential error in the wrap angle calculation include:

[0067] (a) RTP size

[0068] Eqns 12 and 13 are influenced by the size of the reel 14. Different equations may apply when a reel of a different size is used. Eqns 12 and 13 should be fixed based on the actual reel size in question.

[0069] (b) Variation from site to site

[0070] As in the RTP-size-induced error, the actual placements of the paster carriage and pit pulleys 26 relative to the reel shaft 52 for a given structure can induce different results. Some sites may have these critical geometric points in different locations even if the same size reel is used. It may thus be desirable to review each installation to determine a specific roll circumference-to-wrap-angle conversion appropriate to that setting.

[0071] (c) Reel-shaft angle

[0072] Reel-shaft error may be more difficult to compensate for because the reel-shaft position changes during operation. It is assumed that the expiring roll reel-shaft position will be at or around 90 degrees (ATDC). Extending the graph above to 70 and 110 degrees yields the following chart: 5072

12

Graph 2

[0073] In this example, the wrap angle error increases as the roll size decreases. In this instance, the error can be as much as 5 degrees over a potential 40 degree variation in reel- shaft position. Given a nominal position of 90 degrees at the end of a paster sequence, the operator will make a series of shaft movements to load a new roll when the expiring roll diameter is large. At large roll diameters, this error influence is smaller. The conjectured 40- degree swing is believed to be in far excess of what a reel operator will do during the normal operation of the equipment.

[0074] Based on the analysis developed from these equations and the specific RTP parameters, the running reel-shaft position that yields minimal reel-shaft position error is around 100 degrees ATDC.

[0075] (d) Paster carriage position

[0076] As seen in fig. 5, when the paster carriage 60 is lowered in a paster cycle, the belt wrap angle changes. In the illustrated arrangement, lowering the paster carriage essentially shifts the paster carriage pivot point down 32 inches. This affects the wrap angle, -suggesting that it may be desirable to use different equations when the paster carriage is down.

[0077] Although the following calculations for the wrap angle when the paster carriage is down were derived from a 45" reel, they may apply to both a 42" reel and a 45" reel if the physical geometry of the paster carriage is the same for both reels, and the location of the pivot shaft relative to the reel shaft is the only important difference in the arrangements.

[0078] This wrap angle development follows the same approach as used when the

carriage is up. In fact, angles Θ₃ and Θ₄ are the same. Equations ®_\ and Θ₂ are as follows,

and are developed in a similar way to when the carriage is up.

[0079] Θ, = cos^"1 Ϊ.L-2- Eqn 14

P C

[0080] where

[0081] P_BC = ^RP² +RC² -2RP_ARC cos(Θ_R -®_P -Θ_X ) Eqn 15

[0082] where

[0084] and

[0085] W_B =2>2² -^-^ηy^ Eqn 17

[0086] and

[0087] W_B = (PPA - ^iπY^ + (^ - 32J² Eqn 18

[0088] Finally,

[0090] From these equations for Θi and ©₂, Θ_w can be determined when the paster carriage is down. The wrap angle can be plotted versus the roll diameter at a number of different reel-shaft positions. Graph 3 shows what the wrap angle will be in this example when the carriage is up. Between 80 and 110 degrees of reel-shaft angles, the difference in

actual wrap angle is roughly 2-3 degrees.

WRAP ANGLE WITH PASTER CARRIAGE UP @ DIFFERENT REEL SHAFT ANGLES

» 70 degrees — •— 80 degrees

A 90 degrees — *— 100 degrees x 110 degrees

<£> ^b ><b _Nfc $, ^ <b ROLL DIAMETER (IN)

Graph 3

[0091] Graph 4 represents the results when the carriage is down. Comparing this graph to that of the previous one shows that at a 90-degree reel-shaft position, there is nearly a

doubling of the wrap angle when the carriage is down instead of up.

WRAP ANGLE WITH PASTER CARRIAGE DOWN @ DIFFERENT REEL SHAFT ANGLES

-70degrees -80degrees -90degress -100degrees -110degrees

r$> Φ φ _\* Φ ι$ <b <b ^ ROLL DIAMCTER (IN)

Graph4 [0092] Based on the calculated data from the change in geometry, when the paster carriage is down, the following Eqns 20 and 21 may be preferred instead of Eqns 12 and 13. Eqns 20 and 21 give the wrap angle (for a 42" and a 45" reel, respectively) when the paster carriage is lowered. These equations are based on a reel-shaft position of 100 degrees in this example, a position that was chosen since it is the most likely position at paste for a full- diameter roll. This choice of reel-shaft angle was based on the assumption that a new roll will be at full diameter when it is moved into the paste position. In situations where a new roll diameter is less than full diameter, other equations may be preferred.

[0093] Φ_a + 27.734) Eqn 20

[0094] ®_w +[ — 1(2.043 Φ_fl +48.265) Eqn 21

[0095] Compared to Eqns 12 and 13, these equations suggest an increased wrap angle when the paster carriage is lowered. An increase in wrap angle suggests that the pit pressure should be adjusted down to prevent too much force from being applied to the roll. The lowering of the pressure may apply equally to the tension and press deceleration components.

[0096] The illustrated example system 29 (fig. 3) takes the position of paster carriage 60 into account. When the carriage is up (as indicated by the paster carriage position indicator 36, fig. 2), the diameter-to-wrap angle conversion is made using Eqn 12 (for a 42" reel) or Eqn 13 (for a 45" reel). When the carriage is down, the wrap angle is calculated using Eqn 20 (for a 42" reel) or Eqn 21 (for a 45" reel). In order to prevent an abrupt change in the pit pressure, modifications may be made to ease the change in pit pressure.

[0097] Other shift positions may be used when formulating the appropriate circumference-to-wrap-angle calculation.

[0098] E. Estimated running-tension pit pressure calculation [0099] As seen in fig. 3, once the wrap angle is determined, the pit pressure needed to provide the desired web tension can also be calculated in the controller 60.

[00100] The following shows the development of the pit pressure relationship to the wrap angle, the tension set-point, and the number of belts. Figure 4 depicts the geometric relationships between an expiring roll 10, the belts 20, and the pit pully 26.

[00101] Starting with the basic equation that the sum of all torques generated by the web

12 and the belts 20 equal zero (])T torques = 0 ), one can begin with:

[00102] (Θ_β / 2)F_0UT + (Θ_β / 2)F_WEB - (®_a I 2JF_1N = 0 Eqn 22

[00103] Another way to look at Eqn 22 is that the torques difference between the belts 20 is the torque transmitted to the roll 10 for tension control.

[00104] Factoring out Θ_fl /2 and doing some equation manipulation on Eqn 22 yields:

[00105] %*&- = 1 - ^L _Eqn 23

^FIN ^Fm

[00106] For a flat bronze belt:

[00107] F_ouτ IF_1N = I I exp(C_F0^ ) Eqn 24

[00108] where C_F is the coefficient of dynamic friction between the roll and bronze belts.

Thus:

[00110] Given the arrangement of the illustrated pit cylinder 28 and the illustrated pit

pulley 26, 2F_1N = PA , where P is the pit cylinder pressure and A is the cylinder area.

[00111] The area of a cylinder is icy _cyl ². The illustrated system can be given to have four

pit cylinders, each with a 2" bore. As a result:

[00112] F_fr^ = 0.5πP(l-l/exp(C_FΘ_fr )). Eqn 26 [00113] Since there are between one and four belts 20 and cylinders 28 depending on the

roll width, F_WEB =0.5N_sπP(l-l/ exp(C _FΘ_W)), where N_s is the number of belts.

[00114] The relationship between a desired pit pressure P and a running tension set point, a given wrap angle, and a number of belts employed can be expressed as:

[00115] Pit_PVT = Tensions, _3?

^PVT 0.5N_sπ{l-\/exp{C_FΘ_ιv))

[00116] F. Estimated roll deceleration pit pressure compensation

[00117] The illustrated system can also be used to slow the roll 10 during a roll deceleration, such as a normal slow-down or an e-stop. To do so, an additional force can be applied to the belts 20.

[00118] Referring again to figure 4 and following a similar approach to develop the running pit pressure, the sum of the torques needed minus the contribution to support the running tension is developed. If the fundamental rotation force relationship for a spinning roll 10 is applied to a sum of the belt-generated torques, one gets:

[00119] T_1N -T_OUT -T_WEB -T_DECL = 0 Eqn 28

[00120] The first two terms represent the torque moment for each of the belts 20. The next term is the torque required to support the web tension. The last term is associated with the roll inertia and is critical only during speed changes. Again, the differential torque supplied by the belts equals the tension and deceleration torques. (Press acceleration is not considered here since there is no way to supplement the torques required to drive the roll during the acceleration. As long as the press acceleration rate is kept within the defined limits, the tension control should be able to manage the small increase in tension during acceleration.)

[00121] Substituting in the appropriate terms for each torque elements produces:

[00122] (Φ_β H)F_1N ~(φ_a /2)F_0UT -(Φ_a I2)F_WEB -Ia = O Eqn 29 [00123] With some simple manipulation:

[00124] i- 0 Eqn 30

[00125] Recalling Eqn 24 for a belt brake results in:

[00126] 1 -X τ = — \ F_WEB + -^— 1 = 0 Eqn 31

[00127] With some further manipulation:

[00128] F_m = ₇ ^ ^ + ₇ w — — — 7 cr Eqn 32

^{1 J m} (l-l/exp(CA)) (Φ_α/2Xl-l/exp(C_FΘj)

[00129] Again, relating the pit pressure to the force applied to the belt 20 yields:

[00130] 2F_λ = τtPit_pvN_s Eqn 33

[00131] Combining both equations,

[00133] The first term here is actually Eqn 27. Consequently, the change in pit pressure to accommodate a press slowdown can be expressed as:

2\nr

[00134] APit_sww = j- -^- Eqn 35 π{Φ_a /2)N_S{1 -l/exp(C_FΘ^ ))

[00135] Relating roll radial deceleration to web deceleration yields:

[00136] Δrø_sw ^E(l^{n 36}

[00137] Given a nominal paper density of 45 lb/ft³ and a full web width of 54 inches, the calculated mass-pound for the four common roll sizes as a function of roll diameter can be represented as follows:

[00138] MaSS₁₁, = 0.0086N_sΦ_α Eqn 37 [00139] Using the definition of inertia for a cylinder l/2mr² in Eqn 36 and then adding in Eqn 37, the change in pit pressure to slow a roll 10 becomes:

0.0086Φ_βQWs

[00140] APιt_sww = ,:^• :, £^■? xx Eqn 38

[00141] Graph 5 represents Eqn 38 for a roll decelerating at 40 in/sec².

_j Graph 5

[00142] This analysis omits the change in inertia from the core as well as the spindle and the brake assembly. The change in pit pressure impact may be small relative to the pressure change required to e-stop a full-diameter roll.

[00143] G. Dancer trim PID

[00144] The dancer position feedback is used to trim the estimated pit pressure to center and to stabilize the tension control system. The proportional term provides sufficient feedback gain component without system instability. The integral term provides loop damping and reduces dancer feedback following error. The differential term adjusts for rapidly-changing feedback error. It is believed that the differential term will be minimal, since the change in dancer error and its subsequent correction does not undergo any rapid change. [00145] H. Dynamic coefficient of friction for newsprint/bronze

[00146] From previously published material, the dynamic coefficient of friction between a bronze tension belt 20 and newsprint is roughly 0.296. This value may change with paper manufacturers as well as with roll characteristics such as moisture content, so modifications may be desired in appropriate circumstances.

[00147] This description of various embodiments of the invention has been provided for illustrative purposes. Revisions or modifications may be apparent to those of ordinary skill in the art without departing from the invention. The full scope of the invention is set forth in the following claims.

Claims

1. A tension control system for a commercial printing line that has a roll [10] from which a web [12] of material is unwound, the system characterized by: a feed-forward control system [29] for controlling the tension in the web as the material is unwound.

2. A tension control system as recited in claim 1, in which the feed-forward control system has a controller [60] that is programmed to calculate a desired setting for a tension- control actuator [28] using input from a source of a tension set point [38] and a roll-diameter indicator [41].

3. A tension control system for a commercial printing line that has a roll [10] from which a web [12] of material is unwound, a source of a tension set point [38], at least one static belt [20] that passes around a portion of the roll, a pit cylinder [28] that applies force to the belt, and a roll-diameter indicator [41], the system characterized by: a feed-forward control system [29] for controlling the tension in the web as the material is unwound, the control system having a controller [60] that is programmed to actuate the pit cylinder to a desired pit tension that is calculated using the tension set point and input from the roll-diameter indicator.

4. A tension control system as recited in claim 3, in which the roll-diameter indicator [41] calculates the roll diameter from input from a press-speed sensor [30] and a revolution sensor [32].

5. A tension control system as recited in claim 3 in which the roll-diameter indicator [41] calculates the roll diameter from input from a press-speed sensor [30], a dancer-position sensor [34], and a revolution sensor [32].

6. A tension control system for a commercial printing line that has a roll [IUJ from which a web [12] of material is unwound, a source of a tension set point [38], at least one static belt [20] that passes around a portion of the roll, a pit cylinder [28] that applies force to the belt, a counter [45] that counts the number of press pulses that occur over a period of time while the printing line is running, a press-speed sensor [30], and a revolution sensor [32] that senses the completion of one revolution of the roll, the system characterized by: a feed-forward control system [29] for controlling the tension in the web as the material is unwound, the control system having a controller [60] that is programmed to actuate the pit cylinder to a desired pit tension that is calculated using a tension set point, the counter, the press-speed sensor, and the revolution sensor.

7. A tension control system as recited in claim 6, in which the desired pit tension is calculated using input from a dancer-position sensor [34].

8. A tension control system as recited in claim 6, in which the desired pit position is calculated using input from a paster-carriage-position indicator [36].

9. A method of controlling the tension in a commercial printing line that has a roll [10] from which a web [12] of material is unwound, a tension set point, at least one static belt [20] that passes around a portion of the roll, a tension-control actuator [28] that applies force to the belt, a press-speed sensor [30], and a revolution sensor [32], the method characterized by: using input from the press-speed sensor, the revolution sensor, and the tension set point to calculate a desired pit tension; and actuating the tension-control actuator to the desired pit tension.

10. A method as recited in claim 9, in which the desired pit tension is calculated using a formula based on the instantaneous wrap angle of the belt around the web.