Cushioning performance of flexible polyurethane foams.
Most items have the durability to withstand mechanical damage in the workplace due to minor misuse or mishandling. However, in most cases, potential hazards during shipping exceed the inherent ability of the item to resist damage. Consequently, package protection is often afforded to the item during transit by carefully designed foam cushions of the appropriate material, thickness, and area to protect against specified impact conditions (1).
The foam cushions mitigate impact forces by providing a gradual deceleration mechanism, which dissipates some of the impact energy such that the item does not exceed its critical fragility, which is a dimensionless ratio measured in multiples of gravitational acceleration (G). Design is based on the fragility factor, which is 80% of the experimentally measured fragility, thus providing a degree of safety and allowing for product variability. Package design engineering is concerned with conferring protection that is just adequate so that packaging and shipping costs are kept to a minimum while maintaining protection according to the specified hazards.
Design procedures are based on dynamic performance curves (alternatively known as cushioning, drop or peak G curves) of a variety of foams that are specific to the base polymer, foam density, cell structure, i.e., open or closed cell, cushion thickness, drop height, number of impacts, and test temperature (2). They are generated by standard drop tests (3, 4), which yield curves similar to the schematic representation shown in Fig. 1. The curves highlight three particular aspects: (i) different foams provide optimum cushioning at different static load ranges, (ii) effective cushioning is achieved over a limited static loading range where the flatter the curve, the wider the working range, and (iii) increased thickness always gives greater protection.
Polystyrene and polyethylene closed-cell foams are widely used in the cushion packaging industry for the protection of items of intermediate to high static loading. The closed-cell structure absorbs energy by cell wall stretching, bending, yielding and buckling. Cushioning ability under specified conditions tends to deteriorate with the number of impacts as the cell walls yield and eventually rapture (5) with increasing impacts. This leads to increases in peak G with the number of impacts, and consequently, manufacturers publish dynamic performance curves based on the 1st drop and 3rd drop (or the average of 2nd to 5th drops) to enable design protection against single or multiple impacts. More delicate lightweight items often use low density open-cell polyurethane (PU) foams as optimum cushioning performance is achieved at relatively low static loadings. These highly elastic foams suffer little structural damage after repeated impacts in the optimum cushioning range and thus provide reasonable multi-impact protection (6).
Traditionally, cushion area is minimized by designing at the highest appropriate static loading [though due consideration must be paid to possible dangers from compressive creep, compression set, and buckling (7) when using minimum cushion areas] and designers calculate the cushion area by dividing the package mass by the appropriate static loading. A cushion (or cushions) of equivalent area is then produced. These usually take the form of one or more rectangular cushions taking into account the product dimensions, yield from the plank or board, and overall support of the packaged item. However, peak G curves are determined by varying the drop hammer mass to alter static loading using square impact samples 15 x 15 cm (3) or greater than 10.1 x 10.1 cm (4). The standards do not take into account the established design practice that often transposes the calculated cushion area to one or more rectangular strip cushions of equivalent area.
The objectives of this work were therefore to study cushion shape effects on cushioning and deformation of open-cell flexible polyester PU foam and to relate the findings to practical aspects of package design engineering.
All studies were performed on grade 4274 open-celled polyester PU foam of 27 kg/[m.sup.3] supplied by Caligen Foam Ltd. To avoid significant variation, all samples were cut from the same batch of foam.
Right parallelepiped specimens of 5, 7.5, and 10 cm thickness were cut to the required dimensions on a vertical continuously honed bandknife. All samples were conditioned at 23 [degrees] C and 50% relative humidity for a minimum of 72 h prior to testing. All impact tests were carried out in the direction of foam rise (due to anisotropy of cell structure).
Investigations also included the effect on cushioning of introducing controlled centrally placed voids into 5-cm-thick samples of external dimensions 15 x 15 cm. Voids of 5 x 5 to 13 x 13 cm were introduced, giving wall thicknesses surrounding the void from 5 to 1 cm respectively.
Dynamic Performance Curves
Dynamic cushion curves were determined on an apparatus represented schematically in Fig. 2. While most commercial cushion performance is determined at drop heights typically 30 cm and above, the minimum mass of the hammer was such that the optimum portion of the cushion curve of 15 x 15 cm flexible PU foam specimens was not as clearly defined at drop heights [greater than]20 cm. For this reason, 20 cm was used as the drop height for all determinations. Data points were the mean of five separate determinations.
A piezoelectric accelerometer, magnetically fastened to the drop hammer, was used to measure the deceleration (in G's) as a function of time. The accelerometer signal was suitably charge amplified for display on the storage oscilloscope, which was calibrated to read directly in G's. Peak G was thus obtained from the maximum in the C-time trace.
Initial studies investigated dynamic performance curves of 5-cm-thick foam by varying the static loading in different ways:
(a) by adjusting the drop hammer mass using 15 x 15 cm specimens (i.e. in accordance with international standards) and
(b) by keeping the drop hammer mass constant and varying the area of square cushions from 18 x 18 cm to 8 x 8 cm.
Also, cushion curves of samples containing voids were obtained where static loading was based on the impact bearing area of foam (i.e. the area of the block less the area of the void) using a constant drop hammer mass. This was compared against the cushion curve generated from non-voided samples by varying the sample area.
Impact deformation was assessed by vertically and horizontally scribing one face of the sample, dividing the width into five equal sections and the thickness into five equal layers. Impact deformation was monitored as a function of bulk sample compression using a high speed photographic technique by modifying the drop testing apparatus as illustrated in Fig. 2. A camera fitted with a 5 cm lens was positioned normal to the scribed faced of the sample. A flash gun was mounted to project light on to the scribed face where the flash trigger was installed such that one contact was attached to the drop hammer and the other on an adjustable arm positioned vertically below the hammer contact. Adjustment of this contact in the vertical direction allowed the height to be set to trigger the flash at the required bulk compression of the sample. Figure 2 shows the system set up to record the deformation pattern at 50% sample compression.
After the system was set for the required bulk compression and drop height, the lights were turned off in the test room, the camera shutter opened, drop hammer released (triggering the flash at the required compression), and the camera shutter closed before reilluminating the room. A series of photographs was obtained for bulk compressions between 0% and 80% to investigate sequential macroscopic deformation. Analysis of resultant photographs showed that time lag was minimal, thus validating the method.
RESULTS AND DISCUSSION
Initial drop test results on 5-cm PU foam at a static loading of 0.023 kg/[cm.sup.2] gave peak G levels of [approximately]9 on the first drop, 10 on the second, and a mean of 11.4 ([+ or -]0.4) on the third to the twentieth drops, where 60 sec was allowed between each drop. This variation was considered acceptable, but, contrary to what was previously found (6), there appeared to be a deterioration from first to third drops, presumably due to destruction of the small percentage of closed cells during the first two drops. For the above reasons all data was obtained from third or subsequent drops.
The effect of altering static loading by either changing drop hammer mass (using 15 x 15 cm samples) or varying sample areas (at constant drop hammer mass) was investigated on 5-cm-thick foam. Figure 3 indicates there was a substantial difference in the G curves as static loading increased and also there was evidence of minimum swift. Peak G values continued to diverge as the static loading increased beyond the range shown in Fig. 3. The peak G value was [approximately]50% greater at 0.055 kg/[cm.sup.2] based on area reduction compared with increasing the drop hammer mass using a 15 x 15 cm specimen.
This has particular implications on package design, as standard curves (produced by changing hammer mass using 15 x 15 cm samples) are used for design calculations, but have been shown to be valid only for the standard specimen size. Where an item mass is such that a smaller cushion area than standard is used, the peak G obtained from the cushion curve is likely to be exceeded (possibly incurring damage in a practical drop). Conversely, if the item mass is such that the cushion area is greater than standard, Fig. 3 suggests that in a practical drop situation, a slightly lower peak G than originally expected would occur. Consequently, although package designers use different foams so that for any static loading the foam used is around the minimum of the peak G curve, results indicated that the common practice of using variable numbers and areas of cushions (which comply with the buckling criterion and calculated cushion areas) may present more danger than is realized. This will be detailed later.
Flexible foam deformation has been studied under static compression (8). This work extended the study to dynamic compression in an attempt to explain the discrepancy illustrated in Fig. 3. A number of impact compression investigations were made showing that parallelepiped specimens deformed asymmetrically with respect to local deformation.
Impact deformation patterns as a function of increasing bulk compression showed similar features for 5-, 7.5-, and 10-cm-thickness foams. Deformation photographs of square 5-cm-thick specimens suffered from relatively poor reproduction clarity; consequently, typical impact deformation is best represented schematically as shown in Fig. 4. The following salient points were observed:
(i) while local deformation was asymmetric, bulk deformation about the central x and y axes was reasonably balanced;
(ii) local layer deformation bands appeared at the surfaces in contact with the hammer and the anvil. Crushing was observed to propagate from these bands toward the center of the specimen, tending toward a more uniform distribution of strain as the bulk compression increased;
(iii) sectional deformation bands were also observed that were more pronounced towards the edges of the specimen and contributed in the extreme to expansion in the edge center layer;
(iv) local asymmetry increased the thicker the sample.
The surfaces in contact with the anvil and drop hammer appeared to have been restricted in movement possibly as a result of friction. In addition, edge and/or corner effects were suspected. The observed deformation mechanism was consistent with air flow, which would tend to expand the sample, particularly where the surfaces were unrestrained (i.e. the center layer) as the bulk compression increased. Layer crushing and sectional bulging tended to suggest that air flow through the open cell structure influenced the deformation mechanism and therefore had some bearing on cushioning performance. The air flow was envisaged, as illustrated in Fig. 5, from progressive surface crushing forcing air from collapsed cells into the bulk of the specimen and escape from the bulk of the sample through the most open cell structure (i.e. the middle layer) to the edges of the specimen, which contributed to the bulging.
The mechanism was more clearly elucidated in dynamic cushion curves and deformation of samples containing central voids. Dynamic performance curves for 5-cm-thick non-voided foam samples (by changing area) and voided specimens indicated that the introduction of square voids significantly improved cushioning ability at higher static loads [ILLUSTRATION FOR FIGURE 6 OMITTED]. As the void size increased, it appeared that the gaseous phase in the central void became a more and more dominant factor in flattening out the curve. Observations for 7.5-cm and 10-cm foam thicknesses were similar. It was concluded that the central void acted as a huge "cell." This cell, under impact conditions, was bounded by the drop hammer and anvil on the top and bottom surfaces and by four open-cell PU foam walls on the other four sides.
It was therefore inferred that controlled voiding in cushion packaging applications could produce potential material savings, though further more detailed work is necessary to validate the concept. A more positive conclusion may be drawn, however, on the basis that the introduction of a 5 x 5 cm void included in a 15 x 15 cm specimen produced little change in peak G level. Similar cushioning performance was achieved even though slightly more than 11% of the foam cushion volume had been removed. A practical implication is that the results suggested that foam with a wide cell size distribution (including random voids formed by cell coalescence), which would normally be rejected on the grounds of appearance, would perform perfectly satisfactorily in cushion packaging applications.
Typical results of deformation studies are shown for a 10-cm-thick, 15 x 15 cm cushion containing a 10 x 10 cm void, at 0%, 20%, 40%, and 60% compression [ILLUSTRATION FOR FIGURE 7 OMITTED]. The deformation was more exaggerated than for non-voided samples, showing significant bulging of the sides, and at 60% bulk compression (and 80%, not shown), definite corner effects were noticeable.
The 2.5-cm, 2-cm and 1-cm wall thicknesses would drastically fail the buckling criterion in a free standing cushion, yet they held up well in the voided cushion. This led to the concept that the walls deformed by "supported buckling" due to the integrity of the hollow sample and internal cell pressure as shown in Fig. 8a, compared with unsupported buckling seen with thin free standing strip cushions [ILLUSTRATION FOR FIGURE 8B OMITTED]. Figure 8c schematically represents slight bulging at the sample surfaces and more exaggerated deformation in the middle layer, owing to the preferential air flow through the more open cell structure. As the impact surface area reduced (i.e. with increasing void size), bulging became more prominent because of a combination of factors:
(i) reduced frictional forces between the foam impact surface and the hammer and anvil and less integrity of the hollow block, allowing some movement of the top and bottom surfaces;
(ii) effective solidification of the surface layers with progressive crushing;
(iii) void pressure acted on all layers, including the compressed layers, but the decay mechanism occurred through the more open structure in central layers.
Deformation mechanisms strongly suggested that initial impact caused gas compression in the central cell, increasing the pressure and concurrently initiating a second mechanism of air flow through the open cell structure of the wall. This resulted in progressive collapse of the system. Air flow was presumed to become more difficult because of gradual restriction of escape pathways by surface layer compression, counterbalanced by pressure increase during further compression, but eventually decaying to ambient.
The cell pressure argument would result in radial pressures being generated, which would suggest that deformation behavior is dependent on sample configuration. For instance, longer flow paths to the corners of square samples would complicate air flow, and the majority of gas escape would be through the path of least resistance, i.e., toward the middle edge of the specimen. Air flow toward the comer of the sample would be deflected in the comer areas to accommodate preferential air flow. This provides a rational explanation for edge/corner deformation effects previously noted and an additional reason why central edge bulging in square (or rectangular) samples was observed. The interpretation also suggested that drop testing of open-cell foams might be more accurately represented by circular specimens where air flow paths are equal in all directions. Deformation studies were therefore extended to circular specimens of 15-cm diameter; Fig. 9 shows deformation at bulk compressions of 0%, 20%, 40%, and 60%. The complex sectional band deformation seen with square samples was virtually absent as shown by parallel layer lines, i.e., no corner effects were observed, whereas strain localization in the layers was evident but less pronounced. The comparative absence of bulging of the wall indicated that the major contribution to bulging in square or rectangular samples was a result of the complicated air flow paths introduced. Pressure relief therefore occurred primarily through the path of least resistance, i.e., from the center of the sample to the midpoint of the side. This conclusion also validates the approach of various studies of Ramon, Miltz, and Mizrahi (9-11), who used circular samples presumably for symmetry reasons.
Results suggested that the relative magnitudes of pneumatic damping effects were a complex function of void size, degree of compression, effective wall thickness, and sample geometry. Previous workers (8, 12, 13) indicated that energy dissipation of open-cell foams is strain-rate sensitive and related to the ability of the gaseous phase to escape. Air flow resistance was considered to be due to friction between the polymer matrix and the gas due to turbulence created by the tortuous irregular flow path. It was also assumed that at high deformation rates, gas was initially more readily compressed than forced to flow.
It was assumed that the drop hammer and the anvil prevented air flow from the top and bottom surfaces of the foam, which seems valid considering the deformation mechanisms shown above. Resistance to air flow out of a foam block or voided specimen was therefore likely to be a complex function of open surface area of the four sides and the maximum path length through which air is expelled, which in a non-voided specimen may be simplistically related to the open surface area to volume ratio of the test sample.
Returning to Fig. 3, the curves generated by changing drop hammer mass were based on a sample size of 15 x 15 x 5 cm, giving an open surface area of 3 x [10.sup.2] [cm.sup.2] and a volume of 1.125 x [10.sup.3] [cm.sup.3], equating to an open surface area-to-volume ratio of 0.27 [cm.sup.-1]. Curves generated by varying areas were performed on sample sizes of 18x18x5 cm to 8x8x5 cm. Correspondingly, open surface area to volume ratios increased gradually from 0.22 to 0.5 [cm.sup.-1]. Now the gas escape rate would be expected to be more readily achieved as the open surface area to volume ratio increased lessening the ability of the gas compression element to assist in energy absorption due to reduced air-flow resistance. This postulation was tested by plotting the difference in surface area to volume ratios and the difference in peak G values of the specific data shown in Fig. 3. Figure 10 therefore explained not only the divergence of peak G values obtained by the two methods at higher static loadings, but also the shift in the minimum values. This work has therefore conclusively proved the dependence of peak G on open surface area to volume ratio of open cell foams. This finding can have potentially serious implications on cushion design, as there may be considerably more danger than currently realized in moving from square to strip cushions. The effect of maintaining equivalent cushion areas and foam volumes on surface area to volume ratios but moving from a single square cushion (say 15 x 15 cm) to two cushions 15 x 7.5 cm or 4 cushions 7.5 x 7.5 cm or to single strip cushions of 22.5 x 10, 30 x 7.5, 37.5 x 6, 45 x 5, or 60 x 3.75 cm is shown in Fig. 11. In all cases the foam volume is identical, but changing specimen geometry significantly altered the open surface area to volume ratio. Strip cushions of high length to width ratio or four corner blocks of equivalent area to a single square cushion increased the open area to volume ratios by a factor of approximately two. If this difference is equated to the data presented in Fig. 10, an increase in peak G of more than 15 would be predicted. It is, however, difficult to verify this practically, as standard drop testing apparatus is designed with a hammer dimension only slightly larger than the 15 x 15 cm standard test specimen. Future work will incorporate modification of the hammer to test the prediction.
Specimen geometry considerations therefore indicated that the established practice of changing numbers and shapes of cushions (provided they equate to the calculated cushion area and do not cause buckling) may potentially cause problems when designing at the margin, i.e., toward the maximum appropriate static load to reduce foam usage. In such instances it is recommended that a minimum number of cushions be used and that the dimensions be approximated to a square to maximize protection.
The contribution of air compression and flow in flexible open-cell cushions possibly impinges on cushion design in two additional areas that appear to be currently overlooked. Firstly, in practical use, a solid packaged item or inner crate and the outer casing generally provide an effective seal at the top and bottom of the cushion. However, in the generation of cushion curves, the base is sealed by the anvil but the top surface is open until the hammer strikes the cushion. As air is such a significant parameter, some inaccuracies may occur.
This was simplistically tested by impacting square cushions (15 x 15 cm non-voided and voided specimens) with and without a sheet of paper placed on the top surface of the test specimen, as shown in Table 1. While experimental error was [+ or -]0.3G for the foam block and [+ or -]0.5G for the voided samples, the results suggested that under standard test conditions, air was pressurized ahead of the falling drop hammer and transmitted into the foam sample, giving slightly better cushioning than would occur in a practical package drop situation. Consequently, cushion curve generation of open-cell foams may more accurately represent practical use by deflecting air flow from the top surface.
Second, full face cushioning may also cause problems in practice in the opposite manner by restricting air flow out of the cushion. If the item to be protected is surrounded by full face cushions within an outer container, then air flow resistance would be increased, which would give cushioning performance somewhat more analogous to the introduction of some closed cells as in semi-rigid foam-in-place PU foams and effectively increase the static loading range.
Cushion curves of open-cell PU foam generated by varying sample area differ from those generated by varying the drop hammer mass on fixed dimension cushions. Supporting studies of samples containing voids indicated that beneficial cushioning results could be obtained. The phenomena were directly attributed to the balance of gas compression and air flow effects in voided samples and in non-voided samples as a result of changing open surface area to volume ratios. Separate studies showed that peak G values were affected by air pressurized ahead of the drop hammer being transmitted into the foam sample.
Table 1. Effect of Covering Foam Samples on Peak G. Peak G Peak G Void Size (covered (uncovered (cm) samples) samples) None 9.9 9.4 5 x 5 7.4 6.0 6 x 6 8.8 6.7 10 x 10 9.0 8.4
Impact deformation studies showed significant local deformation strain asymmetry in square cushions, increasing in severity in voided specimens, which was dramatically reduced in circular specimens and therefore attributed to corner effects complicating air flow paths. Deformation proceeded from initial compression of surface layers, which propagated toward the center of the specimen, tending toward more uniform strain distribution at high bulk compression strains.
Practical implications on cushion design procedures tended to indicate that
(i) best protection could be achieved by minimizing the open surface area-to-volume ratio of the cushion(s);
(ii) void introduction, within limits, may improve protection ability;
(iii) samples containing irregular cell structures, or random void concentrations as a result of cell coalescence, perform satisfactorily in cushion packaging applications;
(iv) cushion curves may be more accurately represented when based on circular specimens;
(v) in cushion curve generation, dissipation of the air flow ahead of the hammer away from the sample prior to impact would more accurately represent practical package drop situations;
(vi) full face cushions may impair calculated cushioning ability.
1. M. Bakker, ed., Wiley Encyclopedia of Packaging Technology, Wiley Interscience, New York (1986).
2. N. C. Hilyard and A. Cunningham, eds., Low Density Cellular Plastics: Physical Basis of Behaviour, Chapman and Hall, London (1994).
3. BS 4443, Part 3, Method 8 (1988).
4. ASTM 1596a, ASTM, Philadelphia (1991).
5. T. L. Totten, G. J. Burgess, and S. P. Singh, Pack. Technol. Sci., 3, 117 (1990).
6. V. O. Shestopal and B. Chilcott, J. Test. Eval., 16, 312 (1988).
7. R. K. Brandenburg and J. J. Lee, Fundamentals of Packaging Dynamics, 3rd Ed., MTS Systems Corporation (1988).
8. M. Fatima Vaz and M. A. Fortes, J. Mater. Sci. Let., 12, 1408 (1993).
9. J. Miltz, O. Ramon, and S. Mizrahi, J. Appl. Polym. Sci., 38, 281 (1989).
10. O. Ramon and J. Miltz, J. Appl Polym. Sci., 40, 1683 (1990).
11. O. Ramon, S. Mizrahi, and J. Miltz, Polym. Eng. Sci., 34, 1406 (1994).
12. A. N. Gent and K. C. Rusch, J. Cell. Plast., 2, 46 (1966).
13. A. Nagy, W. L. Ko, and U.S. Lindholm, J. Cell. Plast., 10, 127 (1974).
|Printer friendly Cite/link Email Feedback|
|Author:||Sims, G.L.A.; Bennett, J.A.|
|Publication:||Polymer Engineering and Science|
|Date:||Jan 1, 1998|
|Previous Article:||Fusion bonding of maleated polyethylene blends to polyamide 6.|
|Next Article:||Development of heat-activated joining technology for composite-to-composite pipe using prepreg fabric.|