Cierre De Envases De Hojalata

Non-round Can Seaming Systems A. Bizkarguenaga, J. Ariño, G. Jaio SOMMETRADE S.L. · P.T. Zamudio · 48170 Bizkaia · Spain

Introduction Unlike the case of round cans, the very nature of their non-round siblings negatively impacts the regularity of the seam formation. The design of the blank body and lid flanges, the body structural stiffness, the curvature changes and the seaming process itself are some of the non-axisymmetric characteristics that make much more difficult to get a good double seam with non-round cans. In this lecture we will survey some of these causes, their effects and how to handle them. Regarding the machine itself, the pros and cons of the different seaming programs adopted by the industry are discussed from the point of view of the productivity and the double seam quality. We will adopt a descriptive approach, based mainly on everyday experience of preparing seam machines for a wide variety of cans. Anyhow, due to the complexity of the involved phenomena, an important amount of qualitative studies and numerical simulations have been required before bringing the concepts to the descriptive level.

Figure 1. Non-round cans involve many additional difficulties with respect to round cans seaming.

Keywords: Can seaming, non-round can, seaming defects, Seaming basics The double seam configuration and requirements are essentially similar for round and non-round can -see figure 2. The involved sheet metal forming process is analogous in both cases. The body’s and lid’s customarily pre-formed edges are first hooked, and after ironed by means of specially shaped groves engraved around appropriate seaming rolls which orbit along the edges juncture. The juncture is back supported by the seaming chuck, which fits inside the lid countersink. Both can elements are held together by the constraint effect of the bottom lifter.

Figure 2. Seaming of round and non-round cans is basically similar in terms of initial and final state of the seam section. The differences arise during the seaming process.

The round cans are typically seamed in two sequential operations, which set up the seam progressively as the corresponding rolls orbit spiraling around the can. Actually, in most cases the can is spinning as the roll approaches following a predetermined law. The so-called 1st operation performs its cycle in 5÷10 turns and the 2nd one in 1.5÷5. The actual values depend on the machine design. Both rolls do at least one final orbit at their nominal settings. In this paper we assume that this process is sufficiently known for round cans and we will concentrate ourselves in those aspects which make a difference in for the non-round cans. Non-round cans seaming operations The first obvious peculiarity of non-round can seaming is the fact that the seaming rolls must orbit the can following exactly or near exactly

Figure 3. Round cans are typically seamed in only two operations, which are plunged gradually in several orbits. Here are shown in their final settings.

the lid contour. There is a wide variety of shapes: elliptic, ovals of different nature, oblongs, filleted rectangular, etc. Except in the elliptic case, most non-round cans contour can be defined in terms of tangent arc/segment polylines. The major difficulty on the side of the seaming machine comes from the necessity to accurately generate these orbits. We will come back later to this important point when we examine the seam defects due to non-conformal orbits. Now we will describe which seaming operations are generally used for non round cans.

Figure 4. Typically, the can supplier curls the non-round lids only partially. The transition between the curled and non-curled segments is a real discontinuity with great impact in seaming formation. In rectangular cans, the effect is magnified by the start of a curved segment of the lid.

Figure 5. Curling during seaming can be deeper than lid pre-curl because the body flange is already in place. In such case this preliminary seaming operation goes beyond a simple smoothing of pre-curl discontinuities and may become a real folding of the lid hook for those cases in which the first operation alone does not succeed.

Curling. Round lids are generally pre-curled during the manufacturing process. The curl is an important feature for a good performance of the automatic de-stacker device, but it also helps the seam formation. Pre-curl depth is limited by the necessary clearance to fit the lid on the body flange. Due to manufacturing reasons it is uneconomical to curl non-round cans uniformly along the whole contour, so they are generally curled just along the straightest sides. This fact induces seaming problems, not only because more forming work will be necessary during first operation, but even more because the change between the curled and uncurled zones is a discontinuity which most of the times acts as a seam defect nucleation point (droops and vees). We must recall that the seaming process is a kind of sheet metal forming in which the material is not completely guided by the seaming rolls during the formation. We must rely in the tendency of the surrounding material to drag the non-constrain ed parts of sheet to the desired place. If a sharp discontinuity is present, like the transition between curled and uncurled sections, this desired dragging is ruined and the edge is ejected from the hook. The detailed mechanism by which this happens is complex, we will only mention that it consist of an edge buckling initiated by abrupt curvature change of the sheet along the rolling direction. In addition, for rectangular cans the discontinuity coincides with the change between the straight and curve segments. This fact plays an important role in the origin of the defects. See figure 4. For the above-mentioned reasons some non-round seaming machines include a preliminary operation for smoothing the pre-curl along the lid contour eliminating the discontinuities. This operation is performed in a single parallel orbit, not spiraling. The curling during seaming can provide further advantages. As the lid is already placed on the body mouth, this operation can be much more aggressive than a simple curl smoothing, folding the lid curl around the body flange up to a level that the lid hook is nearly set to its final state. As in this folding process the material can be reasonably well driven, the lid hook can be formed nearly without wrinkles. This feature has proven a good solution for some difficult cans. See figure 5. 1st Operation. The goal of the first operation with non-round cans is entirely similar to the case of round cans: to form the lid and body hooks ready to be ironed by a second operation. It can be said to a great degree of exactitude that for analogous blank sections, analogous 1st operation shapes have to be reached. However, the work is more difficult to do with non-round cans.

Figure 6. “Classic” 1st Operation is very much like the typical 1st operation of round cans. Though most of the times the real process is accomplished by two opposite rolls, instead of one, orbiting in two interlaced spirals.

In the past the round can technique was mimicked as close as possible, so the first operation roll was introduced in 8÷15 spiraling orbits with the intention to do the forming work as progressively as possible. Progressive meant to divide the hook forming process in several steps of lower deformation rates, and this was believed to be good for better drawing the lid material. Different mechanisms and tactics have been used to this end, but it is not the aim of this paper to review them. We will only mention that in most cases the 1st operation task was in fact performed by two or four opposite rolls orbiting in interlaced spirals instead that by only one roll in a single spiral. This technique is called “the classic method” and it is still widely used by most existing non-round seaming machines. See figure 6. As a matter of fact, the classic method reached its ceiling when modern cans came into play. While the market was continuously increasing the quality standards, factors like two piece cans, aluminum, thinner and less plastic sheet metals made more and more difficult to obtain good seams. The 1st operation is the most critical point with respect to these factors by the reason already discussed that the material is not positively guided during the deformation. During the eighties it was discovered that a different strategy could be more efficient in term of seam quality and process speed. This technique is known as “plunge method”. We will explain how it works. It is simpler that the classic method, indeed. The first operation roll is abruptly plunged to its nominal setting and a single parallel orbit is executed around the can. No spiraling orbits. See figure 7. Contrary to the old believes, the material is better controlled with this method and much better results can be obtained. The orbit start point is chosen to plunge the roll within the less curved segment of the contour because this is the less sensitive area. In some machines two opposite rolls shares the work, each of them performing only half an orbit.

Figure 7. “Plunge” 1st Operation has proven more efficient in drawing the lid material into the desired hook. In addition it allows arranging the whole seam cycle very efficiently increasing the mechanism performance 2 or 3 folds.

2nd Operation. During this operation the material is better driven than in the 1st operation. Here the key point is to use a suitable grove profile. This profile must fit the shape of the seam plied-sheets, while constraining any tendency to the seam to drop. So 2nd operation is less delicate than 1st one and this is true for round and non-round cans. As for the case of the first operation, there exist classic and plunge methods in non-round can seaming but, notoriously enough, some kind of plunge strategy has been already used for round cans second operation. Plunge 1st operation is always used in combination of plunge 2nd. It is worth commenting that, with plunge method, the operations can be arranged very efficiently, because they are brief and can be overlapped to a high degree. The full effective cycle can be reduced to only one turn plus a small arriving-delay per each subsequent operation –the arriving-delay depends on the architecture of the seaming mechanism, typically ¼ or 1/3 of a turn. If two rolls per operation are used, the full effective cycle could be reduced in an additional ½ a turn. These leads to a dramatic grow in the performances of the machine, increasing the output 2 or 3 times with respect to the classic method for equivalent mechanism rates.

Figure 8.a.. Seaming force Fs develops as consequence of the normal and frictional pressures in the grove while the lid hook is drawn. The stress distribution of this force Fs(ϕ) spans the contour zone being seamed with a tear shape distribution.

Seaming forces Before examining seam irregularity aspects we need to understand some of the forces which are involved in the seaming process. We will concentrate on the 1st operation because it is this step where the seaming forces more severely deform the can.

Figure 8.b. The seaming force Fs is counterbalanced by the constraint reactions of the chuck and body flange. Many times these constraints cannot hold locally the force Fs and the countersink slips down and other mechanisms come into action..

We have already mentioned how the juncture under seaming process is held in place by means of the chuck and the lifter. The chuck lip acts as an anvil providing a stiff back support to the juncture area. As the 1st operation roll pass by a given section, the lid flange slides following the grove profile, setting up the lid hook in this way. This process develops substantial contact pressures and friction forces, which accumulates along the profile, giving the resultant FS of the figure 8.a. The resulting force depends mainly on the friction coefficient between the grove surface and the lid sheet external face; and on a complex interrelation of the grove profile curvature and lid sheet bending resistance (This is a very simplified viewpoint. In fact, roll diameter and lid contour curvatures have also a very important influence). It is important to notice that this force is developed over all the zone of the lid being drawn in each instant. The stress distribution in this zone is shown in the top view of the figure 8.a. The seaming force FS is considerably stronger with plunge method than with classic method. As the roll travels along the contour, the stress shape translates and produces a drift effect, which sums up an undesired tendency to move ahead the lid material. We will discuss some consequences of this drift when we discuss the lid countersink sinking in the next paragraphs. The seaming force FS is counterbalanced by the constraining reactions of the chuck and the body flange –figure 8.b. The condition for keeping in place the juncture while seaming is approximately given by, FB ≥

FS µC Sinθ + Cosθ µB − µC Cosθ − Sinθ

where FB is the minimum body lip force to avoid countersink tendency to slip. θ is the chuck lip angle, µC is the friction coefficient between chuck and lid, and µB is the friction coefficient between body flange and lid. When this condition is not locally fulfilled in the seaming zone, other mechanisms must come into action, as we will see in the next sections. In the remaining part of the paper we will survey some of the principal causes of seam irregularities in non-round cans. Countersink irregularities Figure 9. When body flange pressure cannot withstand countersink slippage, the countersink starts working structurally. This response is very different in the straight and curved areas. While curved zones support the seaming force almost without lid sinking, the straight parts fail to support it and a considerable sinking takes place. The longer the straight parts the deeper the sinking. The larger the countersink angles the larger the depth differences between straight and curved zones.

This is one of the most troublesome points when dealing with nonround cans. The countersink depth changes along the contour. Long straight segments tend to have deeper countersinks than sort or tight curved segments. Why is this a problem? First there are aesthetic reasons (if very intense can also affect the functionality of the automatic can openers), but the main problem is that irregular countersink depth means irregulars lid hook length and hook overlap, which in turn are critical parameters for seam safety.

The reason of the countersink depth irregularity is the uneven response of the can structure to the seaming forces along the different points of the contour. We have already seen that many times, especially with large countersink angles (θ), the bodies flange fails to support the seaming force and the countersink tends to sink. In such cases the countersink reacts structurally against the sinking. At the contour straight segments the countersink panel is a plane strip and works like a slender beam bending under the seaming force. The stiffness in these points is not very high. At the curved segments the panel has a conical shape offering much higher stiffness against the sinking -see figure 9. Consequently, countersink depth increases during seaming much more at long straight sides than at curved segments.

Figure 10. Compensation of countersinks sinking. The depth of the countersink sink is subtracted to the lid flange leading to shorter lid hook and overlap. Adding the expected amount of sinking to the lid curl roughly compensates this problem.

This effect is difficult to control or compensate because it depends on factors as unreliable as the can filling, which sometimes helps supporting against the sinking. Nevertheless, can-makers use to foresee a rough compensation for rectangular cans in order to get acceptable regularity in the lid hook. They provide the lid flanges with additional width in the straight sides, while trim the width at the corners. The blank flange width can differ in as much as 0.75mm in a typical club can; see figure 10. A present trend is to decrease the countersink angle (θ) to reduce the influence of the seaming force. Some modern cans have as low as 1.5 degrees instead of the common 4 or 5 degrees. The seaming force increases with the friction coefficient between the grove and the lid surfaces, as well as with the grove profile curvature. So, the control of the countersink sinking depends too on a good choice of roll profile and on the finishing quality of the roll grove. Within the straight segments there is another source of irregularity. The countersink sinking is not symmetrical. It is deeper in the second half of the straight part than in the first one -see figure 11. This is because the countersink panel is partially supported by the already seamed sections. The whole effect is a drifting of the sinking which produces the maximum depth well after the middle point of the straight segments.

Figure 11. The countersink sinking in the straight segments has and inverted bell shape, but it is not symmetrical. The maximum depth point is backward the straight part mid point. The sinking effect tacks place just ahead the roll, while just behind the lid is already seamed and constrained, leading to a drifting that explains the asymmetry.

Distortion induced by asymmetric body stiffness Round cans are axi-symmetrical. This means that any radial sections behave in identical way under compression forces between chuck and bottom-plate. With non-round cans the structural response of each section can be very different. The can body shows different stiffness in different areas. Curved parts are stiffer in both axial and radial directions, while straight parts are weaker -see figure 12. The bottom plate force tends to be transmitted through the stiffer paths, consequently the curved parts are more loaded and more pressure is transmitted to the lid at these areas. We have already seen that this body flange pushing pressure is important for control the seaming operations. In addition the flat parts tend to buckle under the pressure, deforming the can and lowering even more the pushing pressure. In general, the body structure tends to deform irregularly and put the different zones of the contour under different seaming conditions.

Figure 12. The structural response of non-round cans under bottom plate pressure is different at the different sections. Curved zones are stiffer in both, radial and vertical directions. The bottom plate pressure is mostly transmitted through stiffer sections. This uneven behavior has a final impact in the seam regularity.

Two piece cans are typically embossed with special bottom shapes for certain purposes. One of these purposes is to properly redistribute the bottom pressure transmission pattern – see figure 13. Altogether, the lack of structural symmetry of the body is one of the intrinsic peculiarities of non-non round cans that indirectly produce many of the seam irregularities. The only possibility to minimize the effects is to use a correct value for the bottom plate pressure and, sometimes, to design a customized lower chuck as to correctly fasten the body avoiding undesired deformations and rearranging the bottom pressure transmission pattern. Figure 13. Embossed shape of the body bottom can be designed to properly redistribute the lower plate pressure to the body flange.

Effects radial drawing The effect that we will analyze in this paragraph is related to problems like wrinkles, vees and droops in the curved segments of non-round cans. It is concerned to the essential difference of drawing condition of the lid and body flanges in sections of different curvature. The seam formation in a straight zone –hooking and ironing- involves basically metal sheet folding -see figure 14. Material strain along contour direction -(l) in the figure- can be practically ignored in the whole process. Initial lengths δli keep unchanged till the end of the process. Technically speaking it is a plane strain conformation process. This condition completely changes in curved zones -see figure 15. In this case the material is drawn toward the center of the curve and forcefully must shrink in (l) direction, δli lengths reduce to δli´. The amount of shrinking depends on the curvature and the seam panel width. The strain in (l) direction is given by ε l = Ln

Figure 14. Seam formation –here 1st operation only shownis a plane strain sheet conformation process in contour straight zones. Initial lengths keep constant all through the process. The metal sheet is basically bent or folded.

Ri′ Ri

being Ri and Ri´ the initial and final radiuses respectively of each strip δli. This in-plane straining of the sheet metal must be compensated by opposing strains εw and εt, in the sheet in-plane perpendicular direction (w) and out-of-plane direction (t) respectively. The material volume must keep always constant, this leads to the condition εt+εw+εl=0 The strain ratio among εw and εt, is determined by the Lankford index of the sheet material (r) –also known as plastic an-isotropy indexr=

Figure 15. In curved zones the material is enforced to flow in a centripetal fashion and it must shrink in (l) direction. To compensate this shrinkage the metal expands in thickness (t) and width (w). The Lankford index of the sheet metal determines how this expansion is shared in by these two components.

εw εt

This index depends on the sheet metal grain structure and gives the readiness of the sheet material to flow in-plane when drawn, instead of changing its thickness. High (r) values are desired for drawing process because it reduces the differential thinning or thickening of the sheet in zones of different drawing ratios. Typical values of (r) for tinplate are between 1.3 and 1.9. Taking, for example, a average value of r=1.5 and the geometry of a ordinary European ¼ Club can, the resulting strain ratios of the lid edge extremity are: εl=-0.21, εw=0.13 and εt=0.08. This imply that the sheet thickens at this point increase in a 9.5% by the end of the seaming process (t1=t0·eεt). A bit more complicated calculation

shows that the total lid flange width stretches about a 8% (strain is different for each point of the flange section and an integration along the profile must be done to compute this value). Note that these lid flange elongation renders higher lid hooks that expected in the can curved zones. Therefore, this effect of curved parts, though small, is adding to the countersink sinking of the straight parts in producing seam irregularities. All the preceding reasoning is basically correct for the whole seaming process if the seam is properly ironed by the second operation. The increase in sheet metal thickness is observed in experiments and it agrees pretty well with the calculated values. The lid hook extra length has been also observed, even though it is usually shadowed by other phenomena. Nevertheless, at the end of the first operation the lid hook rarely behaves in curved zones as described. We have already commented that the first operation is a drawing process with low control of the material flow. The most common response of the lid flange extreme to the in-plane shrinking demand in (l) direction is that of a micro-buckling. The extra length is absorbed by edge waviness, commonly known as lid hook wrinkles, see figures 16 and 17. In those figures it is shown how the lid hook edge has the same length that before the hooking δl, but it is sinuous and spans the reduced contour length δl´. At this stage there has not been major shrinking of length or growth of thickness and width in the lid edge.

Figure 16. In tight curved zones sheet metal buckling assimilates the shrinking requirement in (l) direction during the first operation. Unavoidably, some kind of wrinkles arises. When these wrinkles are uniform and smooth it is not difficult for the second operation to flatten them.

The wrinkles at the end of the first operation are unavoidable in tight curved zones. If these wrinkles persist after the second operation beyond a given percentage of the lid hook length, the seam is considered unsafe (30% with respect to the hook length is considered the maximum acceptable wrinkle depth value for round cans, while up to 50% can be allowed for non-round cans). Leaving aside the sheet metal mechanical characteristics –temper and thickness are the authentic governing parameters-, it is compulsory to start with smooth and uniform first operation wrinkles, like those shown in the figure 16. If the wrinkles are few and sharp like the ones shown in figure 17, the second operation would not iron out them. The key question is how smooth and uniform wrinkles can be provoked instead of sharp ones. Again, lid material yield stress and plastic elongation and sheet thickness are the commanding parameters. It is a commonplace the continuous cost reduction trend to –can weight reduction- by using thinner and thinner sheet metals with harder and harder materials -both characteristics must vary opposing because can strength needs to be maintained-. These properties progress in the bad direction from the seaming point of view and, in fact, it is the seaming ability one of the most important obstacles to use thinner sheet metal. We will not discuss here possible solutions in this respect, like mini and micro seams, etc. The first operation roll grove profile has also an important role, but in our opinion this is a point that has been over assessed. It is actually a variable with obvious room to play with, but once a few design key factors are satisfied, little can be improved by shape refinements. For the time being no very much advantage is expected from profile optimization, yet a reliable theory of seaming profiles would be desirable to clarify this controversial subject.

Figure 17. The plunge 1st operation –shown in figure 16- is notably more efficient than the classic one -shown in this figure-. In this case few wrinkles arise in the first orbits and these are progressively sharpened in the subsequent passes. If the wrinkles are very deep they become pleats after second operation. Sometimes these protrusions scratch the can body and form vees and drops.

In this section we only will see the importance of using the plunge or classic methods with respect to the wrinkles. One of the characteristics of the plunge method is that, as the roll progress hooking the lid flange, the edge does not buckle immediately. The sheet border is properly put in his final hook shape without buckling, while the exceeding sheet material is accumulated in a bulge before the rolling point. When the accumulated bulge grows too large the roll step on it and the sheet edge slides behind the rolling point gently buckling the already hooked flange into a smooth wave. The key requirement for the roll design is to be able to properly swell back the fore bulges before they grow too large and the hook is ejected down. To this end the grove profile design is obviously important, but the roll diameter is also important. Large roll diameters facilitate very much the task.

Figure 18. The can twisting consists of a rotation of the bottom with respect to the top. It is typically observable in heaps of many stackable cans. The effect is very intense when there is significant gap between the lid countersink and the body mouth. The exceeding body wall is dragged ahead producing the twisting.

So, within certain limits the plunge method handles fairly well the first operation requirements. On the contrary, the classic method is considerably less efficient, in the sense that the wrinkle generation can abort the hook formation or can end in sharp and uneven wrinkles like the ones shown in the figure 17. Indeed, the wrinkles arise after the earliest orbits pretty much like described for the plunge method but obviously in less number and less deep. For the subsequent passes, the flange edge already has buckling initiating ripples sparsely distributed. Instead of increasing the number of wrinkles, the exceeding edge material is absorbed by sharpening the initial ripples. The hook, when formed, ends with few and sharp wrinkles, which will be difficult to iron out. Many times, these wrinkles are so deep that touch the can body wall during the last passes of the roll, scratching it and jamming the hook formation itself –see figure 17. In these cases the result is, not only wrinkles and pleats, but also vees, drops and droops. Can twisting by seam drag

Figure 19. The body wall dragging introduces shear stresses and other effects that can produce irregularities in the seam. This takes place particularly in rectangular cans at the end of the last corner.

Figure 20. The exceeding body wall dragged along the contour accumulates at the end of the round and forms a bulge in the wall. The swelling can be large, and then the roll crash into it producing a dent. Most of the times a slim bottom restrain like (A) can solve the problem if the gap is not very large.

From the seam quality viewpoint this effect is not as important as the previous ones, but it is certainly generated by the seam process and can produce an unappealing distortion of the can. In some cases it can be so intense that bulges and notches appear in the body walls. Finally, in less frequent cases, it can lead to important variations of seam parameters along the can contour. The effect consists of a torsional deformation of the body along the vertical axis. The top and bottom faces remain flat and parallel, because they are constrained by the chuck and the bottom plate, but they are rotated the one with respect the other. The phenomenon is produced by a dragging ahead of the body flange along the seam contour -see figures 18 and 19. The body flange is basically folded during the first operation. This is not a very aggressive forming, but in the curved zones and especially when the body does not fit tightly around the lid countersink, it tends to push ahead the body flange and wall. This effect is accumulative and it is more intense where the 1st operation ends its work. The net consequence is a rotation of the bottom of the can, as it is shown in the figure 18. Most times this is only observable after careful inspection or when many stackable cans are put in a heap. When the fitness between lid countersink and body mouth is poor, the excess of body contour is accumulated at the end of the operation and a bulge arises in the body wall. The roll bottom rim pushes inside the bulge, producing a notch just under the seam. This is

illustrated in the figure 20. When the gap between the lid and the body is not very large, the torsion can be practically eliminated using an appropriate restrain at the lifter plate, (A) in the figure 20. The same solution can eliminate the bulge and notch in these cases of low gap. When the gap between body and lid is too large, the excess of body edge affects the seam formation, particularly at the end of the last corner of rectangular cans -some times at the opposite diagonal too-. In such circumstances the body flange becomes too large and irregular at this point and makes difficult a correct end hook formation. This circumstance is sometimes evidenced by a long and smooth seam droop in the mentioned areas. The plunge method is more likely to produce this kind of effects than the classic one. Effects of machine borne non-conformal orbits The space left between the roll grove and the chuck wall is the seaming pass section –see figure 22. All seaming operation rolls must follow a trajectory around the can contour such that the seaming pass section must be kept constant. This is an obvious requisite to get a regular seam all around the can. It is easy to fulfill for round can seaming mechanisms but it is a rather involved problem for non-round can seamers. In order to be able to evaluate the precision with which the pass section is kept constant we need to define the roll working diameter and the reference point. The roll reference point is the point of the grove profile that is just on top of the chuck lip apex when the roll is in its nominal setting. The roll working diameter is that defined by the reference point. Note that the reference point does not correspond to any fixed material point on the roll, it is a geometric point, which defines the correct setting of the operation roll with respect to tightness or looseness of the seam formation. These ideas are more clearly shown in figures 21 and 22. If, for instance, the reference point is separated 0.2mm left from the chuck contour, the seam is loosen in the same quantity. We will evaluate the regularity of a seam operation orbit by the fitness degree of the trajectory described by the reference point onto the chuck contour. A conformal orbit is the one which coincides exactly with the chuck contour or is separated a uniform distance from it. A non-conformal orbit produces a non-uniform seam.

Figure 21. The relative position of the roll with respect to the chuck is essential for the seam formation. To get a regular seam, this positioning must be exactly copied along the entire chuck contour. This task is quite complex with non-round cans. The concepts we use to control this performance are the roll working diameter ∅W and reference point (see next figure)

Figure 22. For a given roll setting the seaming pass section is defined by its roll reference point and working diameter ∅W. We will evaluate the regularity of a seam operation orbit by the fitness degree of the trajectory described by the reference point onto the chuck contour.

It is worth noting that the roll working diameter depends on the roll nominal setting. We will see that this issue is very important, because normally a machine tooling is calibrated for a particular roll working diameter and the calibration is lost if the operations are tightened or loosen out the original setting. The sources of non-conformity are to be found in the seamer mechanism and may be very diverse. We will restrict our remarks to some of the sources that are common to all the cam-lever type seaming devices. Practically every non-round can seamer is based on this kind of mechanism. Figure 23 shows a schematic view of the cam-lever working principle. The roll, mounted on one arm of a pivoting lever, generates the target conformal orbit as the pivoting axle revolves on a spindle around the chuck. In turn, the lever is driven by a cam through

Figure 23. Cam-lever type seaming mechanism. The lever pivot rotates around the center with constant angular speed. The chuck and cam contour are conjugated silhouettes. In absence of errors, when the follower rolls over the cam the roll orbits the chuck in a conformal trajectory.

a follower mounted on its opposite arm. To get a conformal orbit the chuck and cam contours must be conjugated silhouettes for the actual mechanism parameters. The examples that follow try to illustrate some of the machine borne seaming irregularities They have been obtained by computer simulation of a SOMME 444 seaming machine running on an ordinary ¼ Club can.

Figure 24. Non-conformal orbit produced by a 0.01% orthotropic deformation of the seamer cam. The orbit outrun is magnified 50 times for clarity –the spacing of dashed lines is 0.1mm-. The resulting orbit outrun is 0.2mm. The seaming process is very sensitive to cam errors.

Figure 25. A horizontal eccentricity of 0.1mm in the machine spindle produces this non-conformal orbit. The orbit outrun is magnified 50 times. Each mechanism parameter error produces a characteristic non conformal orbit.

Figure 26. The follower recall springs generate variable forces, which slightly deforms the mechanism, leading to seam irregularities. In this real case the maximum outrun error remains at 0.08mm after readjusting all other mechanism parameters. The tight seam at two opposite corners is very characteristic of this defect. The orbit outrun is magnified 50 times for clarity.

Cam silhouette errors. We have seen that the cam silhouette must be conjugate of that of the chuck. If the cam is not manufactured exactly to the required shape we will have seam irregularities. The figure 24 shows the non-conformal orbit that results from a cam with an orthotropic error of 0.01% -this roughly means a maximum cam shape outrun error of 0.075mm-. The resulting irregularity in the seam is ± 0.2mm. We can see that the seam quality is very sensitive to cam silhouette errors. The only solution to this kind of problem is to correctly reshape the cam. Mechanism dimensions and geometry. Although the basic working principle of the cam-lever mechanism is very simple, the actual devices are rather complex, with many pieces and joints that potentialy accumulate manufacturing and assembly errors. Any dimensional error or misalignment that has an influence in one of the mechanism basic parameters is prone to produce a non-conformal orbit and seam irregularities. There are nine really basic parameters –two of them are easily compensable-. Some parameters are more sensitive than others with respect the seam regularity. Just to give an example, the figure 25 shows the effect of 0.1mm eccentricity of the spindle with respect to the mechanism nominal axis. This eccentricity may seem too large for a precision device but, when three-dimensional misalignments are taken into account, it is not difficult to reach effective values like this in the reality. The only way to compensate this kind of errors is to reshape the cam directly on the machine with a special grinding apparatus. Deflections by static loads. The mechanism elements and joints are not absolutely rigid. A small flexibility is enough to produce significant non-conformances under variable forces of dynamic and static character that take place in the mechanisms. The static character does not mean that the force does not change, it means that it is driven by position instead of velocity or acceleration. In the static category we can find the seaming force Fs already discussed, or the one produced by internal springs. For instance, some mechanisms make use of recall springs to ensure positive contact between follower and cam. In our example the springs deflect the mechanism producing a large orbit outrun between 0.17mm minimum and 0.33mm maximum. This error can be roughly compensated readjusting the pivot radius. However residual non-similarity remains with a maximum outrun of 0.08mm – see figure 26. This effect, which produces characteristic tighter seams at two opposite corners, is very common and some times causes difficult to handle seam defects. Deflections by dynamic loads. The dynamic forces deflect the mechanism much like the static ones. The important difference is that the dynamic forces depend on the mechanism rate. Moreover, this dependence is of quadratic character. If the machine rate is doubled, the dynamic forces increase four times. These forces depend too on the can shape. For a

given rate the forces are lower for near round cans –as oval cans- and larger for slender and tight cornered cans –like long rectangular-. The dynamic forces put real limits to the maximum rate at which a nonround can seaming machine can run. From the seam regularity viewpoint, the effects of a large speed can be seen in the real case shown in the figure 27. In this example the machine is running at 120 cans per minute. The spring forces are also considered in the example. We see that the overall effect is loosening of the operation, but the loosening is double at the corners than at the straight parts. This is a typical signature, frequently observed in closed can. Roll working diameter. The figure 28 shows a problem that we have advanced at the beginning of this section regarding the roll working diameter. It is frequently found during the tooling setting. In the project phase the cam silhouette is designed for a roll nominal diameter –58.00mm in this case. However, due to the real setting needs of the different operations, the effective working diameters of the rolls differ from the nominal one –in this case 57.66 for the first operation and 57.22 for the second one. The actual setting needs are often determined in a trial and error process-. Consequently, some basic parameters of the mechanism have to be readjusted and the new reference point yields a non-conformal orbit. The resulting seam irregularity is 0.06mm for the first operation (still acceptable), but for the second operation it rises to 0.13mm, showing a kind of twisting commonly mistaken for chuck misalignment. The correct approach to fix the defect must be determined in a case by case fashion, but typically it needs to modify the physical diameter of the rolls.

Figure 27. This machine, running at 120 c.p.m., shows an average outrun of 0.1mm at the straight parts while at the corners the outrun is double. This is a typical result of the mechanism deflections caused by the dynamic loads. The effect of the springs is also included in this case. The orbit outrun is magnified 50 times for clarity.

Figure 28. The problem of the effective roll working diameter can only be fully solved modifying the physical diameters of the rolls to restore the reference point to the nominal diameter. Cam reshaping solution will not work if, like in this case, the working diameters of the different operations are not the same.