1. Introduction
The distribution environment includes the handling, storage, and transport operations that a product is subjected to as it is moved from the manufacturing facility to the final customer [
1]. Within the distribution environment, several hazards have the potential to damage products, such as compression, vibration, and shock events. Mechanical shock occurs when a packaged item’s position, velocity or acceleration suddenly changes. Shock can be characterised by a rapid increase in acceleration followed by a fast decrease over a relatively short time duration. A longitudinal shock of a package in a unit load may typically be 10 ms (0.010 s) long and have a magnitude around or over 4G. [
2] Shock events include sudden stops in transportation, drops during manual handling, and impacts during distribution; basically, any event which causes a drastic change in a short period of time [
3].
Because of the high probability that products will face distribution hazards before reaching the final customer, packaging engineers must ensure their products and containers can withstand these hazards. A thorough understanding of the risks to which the package is exposed to during distribution, and an efficient distribution loss tracking system are useful to aid in the design of cost-efficient packaging solutions that will protect products and prevent damage.
Equipment such as forklifts and pallet jacks are ubiquitous in modern warehouses [
4]. Due to excessive handling, there is a high potential of this equipment causing damage to pallets. Fork tines are listed as the fourth most common cause of damage, after protruding nails, dropped boxes, and hitting the bars in racks [
1].
Pallet testing standards are often used to simulate the damage caused by forklift handling in laboratory settings. ASTM 1185 [
5] and ISO 8611 [
6] both incorporate incline impact tests that stress the pallet deck edges, blocks, and stringers in order to determine their resistance to impacts by forklifts. The effect of forklift handling is also incorporated into the ISTA 3B testing standard, which includes a pallet handling sequence that simulates damage caused by the forklift [
7]. In 1993, Virginia Tech developed a comprehensive pallet durability simulation, called FasTrack, where the overall life of a pallet is assessed using a handling sequence that mimics what pallets experience in the field from industrial forklifts and pallet jacks [
8].
Knowledge of the locations and types of damage experienced by pallets during forklift handling can help to improve the durability of pallet designs. It was observed that pallets used in the field feature a high percentage of top lead deckboard and stringer damage, which is often caused by the use of forklifts and pallet jacks. Wallin and Whitenack used this information to investigate the effect that selectively placing high-quality wood on the edges and ends of the pallet would have on pallet durability [
9]. This information has been used in standards and industry guidelines, such as the ASME MH1-1997 [
10].
An extensive number of research projects have focused on understanding the vibration environment of common transportation modes such as parcel trucks [
11,
12], rail cars [
13,
14], and industrial trucks [
15], as well as vibration transmissibility in a unit load [
16]. The influence of different factors, such as speed, top load and suspension, have been extensively investigated too [
17,
18,
19]. With this information, recommendations regarding how to simulate these conditions have been proposed [
20]. Meanwhile, only a limited number of research projects have focused on the investigation of shock impacts during material handling [
21].
More information is needed on the effects that factors such as the weight of the load and the speed of the forklift have on shock damage. Rodriguez et al. equipped pallets and unit loads with sensors to measure the level and transmissibility of shock impacts during forklift handling for three different unit load configurations. His study created different impact condition levels based on driver experience, operating speed, and load weight using a counterbalanced forklift truck. The obtained results were compared to the conditions recommended by testing standards, and it was determined that the levels recommended by ASTM D 4003 were, at that time, excessively severe. So, authors proposed the use of an equation based on impact velocity, pallet weight, forklift weight, and the coefficient of restitution to determine impact levels during the simulation of pallet marshalling [
22].
While factors affecting vibration, such as load weight, equipment type, and speed, have been extensively researched, there is a lack of information on these factors’ effects on shock events which could limit the applicability of current standards to the wider range of conditions seen in supply chains.
The objective of this research was to investigate the effects that forklift type, top load, pallet design, and entry speed have on horizontal shock impacts exerted by forklifts on pallets during their interactions. Additionally, this research investigated the acceleration levels measured on the pallet as compared to on the forklift. Understanding the differences in shock transmissibility between the forklift and the pallet allowed us to also obtain data on the handling equipment, rather than just on the unit load. While pallets are handled an average of 80 times per year (four to six times per trip) [
9], forklifts handle hundreds of unit loads during regular operation. A similar approach could be taken for other testing equipment, which could be equipped with sensors to gather data over multiple samples, further reducing research costs and investigation efforts.
2. Materials and Methods
2.1. Forklifts
The forklift types utilized for this study included a Clark model CQ30L gas forklift, Clark model TMG15 electric forklift, and a Crown model RR 5715-35 reach truck (
Figure 1). The specifications of the forklifts are listed in
Table 1.
2.2. Pallet Designs
Two pallet designs were investigated during this study including a wooden and a plastic pallet design. The wooden pallet design selected for the study was a 1200 mm × 1000 mm, multiple-use, block-class, non-reversible, perimeter-base pallet (
Figure 2). The wooden pallet design had an average weight of 23.5 kg, and it was made of kiln-dried, Southern Yellow Pine. The number and dimensions of the pallet components are presented in
Table 2. The plastic pallet design was a 1200 mm × 1000 mm nestable, block-class, non-reversible pallet (
Figure 2). The plastic pallet design had an average weight of 11.04 kg, and it was manufactured by the Orbis corporation (Menasha Corporation, Oconomowoc, WI, USA). Because the friction between the pallet design and the floor could influence results, the static coefficient of friction between the bottom of the pallet and the concrete floor was measured for both pallet designs using the method outlined by O’Dell et al. [
23]. The coefficient of friction for the wooden pallet was 0.62, and for the plastic pallet, it was 0.34.
2.3. Data Collection
Acceleration levels were measured with Lansmonts’ SAVER 3D15 and SAVER 3X90 dataloggers (Lansmont Corporation, Monterrey, CA, USA). The SAVER 3X90 datalogger was mounted to the back of the fork tine carriage using Scotch permanent outdoor/exterior mounting tape (3M, Maplewood, MN, USA). The measurement of the horizontal shock impacts on the forklift could be a more effective data collection method because it could yield significantly more shock data to researchers because forklifts handle thousands of pallets a day while pallets are only handled 16 times by forklifts [
8]. The measurements on the pallet were recorded with the SAVER 3D15 datalogger, positioned on the top lead deckboard for both pallet designs (50 mm measured in from the 1000 mm side, and 229 mm measured in from the 1200 mm side of each design). The sensor was secured to the pallet using Scotch permanent outdoor/exterior mounting tape (3M, Maplewood, MN, USA). The locations of both sensors are shown in
Figure 3. Data was collected from all three axial directions. The channel parallel to the direction of the impact was used as trigger. The settings used for the dataloggers were:
The SAVER 3X90, located on the back of the fork tine carriage (
Figure 3a), was oriented with the Z axis parallel to the direction of impact and travel of the forklift (
Figure 3b). In this orientation, the X axis recorded the lateral motion. while the Y axis recorded the vertical motion of the forklift. On the pallet, the SAVER 3D15 was oriented with the Y axis parallel to the direction of impact (
Figure 3c). Similar to the sensor on the forklift, the X axis recorded information about the lateral motion of the pallet, and the Z axis recorded the vertical motion of the pallet.
2.4. Forklift Impact Test
The pallets were impacted by the forklift in a single movement. To ensure the entry speed remained constant between repetitions, marks were drawn on the fork tines. The marks reflected the distance at which the entry speed was 0.22 m/s and 0.45 m/s. Rodriguez et al., 1994 reported that the speed during normal forklift operation is 0.33 m/s and the authors wanted to simulate operating speeds that are slightly lower and greater. To ensure that there was no disturbance in data collection, impacts were repeated after the forklift remained idle for one minute. The condition of the pallet was monitored to avoid bias in the measurements due to pallet fatigue. The pallets were loaded with three different top loads: 227 kg, 680 kg, and 1134 kg. The top load was offset from the sensor to avoid the top load contacting the sensor during impact.
2.5. Experimental Design
The experimental design is shown in
Table 3. It was set up as a full factorial design with acceleration as the main response. An analysis of variance (ANOVA) was used to test the effect of forklift type (3 levels), top load (3 levels), pallet design (2 levels), and entry speed (2 levels). Ten replicates were conducted for each variable.
2.6. Statistical Methods
Results were analyzed with the Minitab Statistical Software (Minitab LLC, State College, PA, USA) as well as MS Excel (Microsoft Corporation, Redmond, WA, USA). To compare the effects of the main response, an ANOVA was conducted with a significance level of 0.05. Peak G force, duration, and delta velocity were manually selected for each event using the SaverXWare v4.1 software (Lansmont Corporation, Monterrey, CA, USA).
A statistical model was used to test the effect of each factor on the peak acceleration values:
where
is the response of interest (acceleration),
is the overall mean,
= effect of the
ith forklift,
Tj = effect of the
jth top load,
Pk = effect of the
kth pallet design,
El = effect of the
lth entry speed,
= interaction effect between
ith forklift and
lth entry speed,
= effect of the interaction between
ith forklift and
kth pallet design,
= effect of the interaction between
kth pallet design and
lth entry speed,
= effect of the interaction between the
jth top load and
kth pallet design,
= effect of the interaction between
jth top load and
lth entry speed,
= effect of the interaction between
ith forklift,
jth top load,
kth pallet design, and
= random error with expectations (0,
).
A Tukey pairwise comparison test, at a significance level of 0.05, was used to compare the factors in the study.
3. Results and Discussion
3.1. Forklift Behavior
The acceleration, duration, and delta velocity response registered for each forklift is shown in
Table 4. The average acceleration measured on the forklifts was 2.98 G. The coefficient of variation obtained for these measurements was 37%, reflecting values as low as 1.38 G (for the reach truck) or as high as 5.1 G (for the gas forklift). The average event duration recorded was 13.6 ms. However, durations recorded in the reach truck were 50% higher than for the gas and electric forklifts. Durations in the gas and electric forklift averaged 10.2 ms. The increase in event duration for the reach truck is due to the reach mechanism being fully extended for its impacts. The coefficient of variation obtained for that event duration was 23%. While high, the variations found in this study are considered acceptable as shock is traditionally considered a difficult parameter to characterize [
24].
The acceleration response measured in the forklifts is used as the main response for further analysis. The p-values obtained in the analysis show that all the main factors, including forklift type, pallet design, entry speed, and top load, significantly affect the acceleration experienced by the forklift during its interaction with the pallet (p < 0.05). Similarly, there are significant two-way interactions between most variables, except between the forklift type and the top load (p = 0.508). Only one three-way interaction was found to be significant in the model (p = 0.000), corresponding to the forklift, pallet design, and top load.
The effects of the main variables were further investigated using a main effects plot (
Figure 4) and a Tukey’s pairwise comparison (
Appendix A). The measured acceleration response was significantly higher (15%) when the gas forklift was used, while the reach truck and the electric forklift were not significantly different. However, event durations with the reach truck were 50% longer than the electric forklift.
When wood pallets were used for testing, the acceleration levels were significantly higher (20.7%). This result could be explained by the greater coefficient of friction between the pallet and the laboratory floor, which increases the resistance of the pallet to impact. The higher stiffness and greater weight of wood pallets could also contribute to this trend. Pallet design did not affect the duration response (
Appendix A).
When the impact speed was increased, the measured acceleration increased by as much as 32%, but entry speed did not affect the duration response. Comparing entry speeds during pallet impacts to the height of a package drop, higher drop heights do not affect shock response durations for package drops [
25].
The results show that there was no significant difference in acceleration when the top load was increased from 227 kg to 680 kg. However, when the top load was further increased to 1134 kg, the measured acceleration increased 22%. The explanation for the lack of payload effect at a lower weight could be that friction between the floor and the lower weight pallet load was not great enough to prevent the sliding of the pallet which reduced the intensity of the impact. However, at heavier weights the resulting frictional force between the floor and the pallet was enough to prevent the sliding of the pallet which increased the resistance and results in greater impact intensities. This result indicates that, at least for heavy top loads, the weight of the top load needs to be tracked during data collection.
Due to the change in mean acceleration, pallet design and entry speed were found to be the most influential factors on the acceleration measured on the forklift at the moment of impact.
Figure 5 shows the interactions between the main effects. A change in pallet design affects the mean peak acceleration for the different types of forklifts. The change from a plastic to a wooden pallet design is associated with higher peak accelerations for the reach truck and the electric forklift.
Entry speed also affects the acceleration response in each forklift. The effect of the change in speed was greatest for the reach truck (46% increase). The changes in the behaviors of the gas and the electric forklifts were similar. Pallet design was also affected in a different manner by a change in speed. The wooden design shows a greater peak acceleration response when the speed increased 0.22 m/s.
Pallet design affected the peak acceleration response measured for the different top loads. When wood was impacted, there was a significant increase in the peak acceleration for the 1134 kg top load. Although the top load was significant, this behavior is pallet-dependent. The behavior is explained by the higher weight of the loaded pallet and the coefficient of friction, both of which create a greater resistance to impact.
The acceleration measured during impacts with differently weighted top loads is also dependent on the entry speed. At higher speeds, there are significant differences between the 227 kg, 680 kg, and 1134 kg top loads (12–14% difference). At lower speeds, the acceleration measured was not significantly different for the 227 kg and 680 kg top loads.
3.2. Pallet Behavior
Table 5 shows the acceleration, duration, and delta velocity responses registered in the pallets for each shock event in the study. The peak acceleration measured on the pallet was, on average, 13.15 G. The data has a coefficient of variation of 40%, with the lowest peak acceleration value of 3.7 G in the reach truck and the highest acceleration value of 19.51 G measured in the gas forklift. The impact event duration was 12 ms with a coefficient of variation of 37%. Impacts to the pallet show an average change in speed of 0.73 m/s, with values as low as 0.23 m/s measured in the reach truck and as high as 1.35 m/s measured in the gas forklift. The coefficient of variation for delta velocity was 42%.
The p-values obtained in the analysis show that the main factors, including forklift type, entry speed, and top load, significantly affect the peak acceleration response measured on the pallet (p < 0.05). Significant two-way interactions were found in the combinations of forklift type and entry speed, forklift type and top load, and pallet design and top load. The three-way interaction between pallet design, entry speed, and top load is also significant in the model (p = 0.001).
The effect of the main variables on the peak pallet acceleration response was investigated with a main effects plot (
Figure 6). The forklift type plot shows an increase of 16% in peak pallet acceleration response due to changes in forklift type. The acceleration response change seems to be affected by the weight of the forklifts. No significant differences were found in pallet impact event duration for the different forklifts (
Appendix A).
The entry speed plot shows a 14% peak pallet acceleration increase when the speed was 0.45 m/s. Similar to the forklift type plot, there was no change in the duration response for the increasing levels of speed. The effect of the top load shows a significant increase in peak pallet acceleration for the 227 kg top load (13.8%). The Tukey pairwise comparison shows no significant differences between the responses for the change of top load from 680 kg to 1134 kg.
Significant two-way interactions are shown in
Figure 7. Significant differences were evaluated with the Tukey pairwise comparison method.
The measured accelerations obtained by forklift type is dependent on the entry speed. At lower speeds, there are differences between the forklifts. The reach truck presented the lowest acceleration response. This behavior could be explained by the reach mechanism of this forklift, which could act as a cushion during impacts. At higher speeds, the greatest acceleration was measured on the gas forklift.
The acceleration obtained with the different forklift types is also dependent on the top load of the pallet. High differences in behavior are seen for each type of forklift depending on the top load. The greatest acceleration was measured in the gas forklift carrying a 227 kg top load.
Pallet design affects the acceleration values obtained for each top load combination. The greatest acceleration response was recorded for the plastic design using a 227 kg top load. This increase in acceleration could be caused by the plastic pallet sliding around more easily when it is impacted by the forklift.
3.3. Pallet vs. Forklift
The mean peak acceleration for impacts measured on both the pallet and the forklift were compared via the Tukey’s pairwise comparison. The differences were evaluated according to the main effects of the study (
Table 6). Significant differences were found in the acceleration levels between the pallet and forklift. There is a pallet acceleration increase of 17–21% when heavier forklift trucks are used to impact the pallet.
Comparisons made by pallet design show significant differences between the acceleration responses measured in the pallet and the forklift. However, there is no difference between the pallet designs, which show an increase of 10 G in the pallet when compared to the acceleration response measured in the forklift. The acceleration responses measured on the forklift and the pallet significantly change with different entry speeds. This acceleration difference ranges from 8 G–12 G.
When varying the top loads, there are significant differences between the peak acceleration responses measured in the pallet and the forklift. However, the 680 kg and 1134 kg top loads show a 9 G difference when comparing pallet to forklift. There is a 12 G difference in the acceleration measurements when a 227 kg top load is used.
The acceleration of the pallet is approximately 4.4 times greater than the forklift. However, the acceleration peak increase was not the same at different types of forklifts, presumably due to the varion in forklift mass and operator manoeuvre styles. Variations in the mass of the forklift types and operator error may explain these scenarios. The impact durations for pallet and forklift are consistent, ranging from 10 ms–12 ms. However, extended durations were obtained with the reach truck, which averaged 20 ms.
Figure 8 contains the shock curve for the same impact event, as measured on the pallet (3D15) and the forklift (3X90). Acceleration measured on the forklift for this event was 4.6 G. The acceleration measured on the pallet was 8.6 G (
Figure 8).
The results obtained in this study were compared to the results of Rodriguez et al. [
21] who measured the shock response on a hardwood pallet when impacted by a forklift truck. The authors used similar levels of pallet top loads (227 kg and 680 kg) measuring the shock response in the direction of impact. The duration of the events found with this research is consistent with the levels obtained by Rodriguez et al. [
22], which ranged from 8.1 ms to 12.2 ms. However, impacts on the 680 kg, the severe impact condition for Rodriguez et al. [
22], showed a significant reduction in event duration (4.9 ms). The duration response for the events measured by Rodriguez et al. [
21] seems to be affected by the different conditions in their experimental design, which were not observed for this research. The shock acceleration response from this research (10 G–15 G) significantly differs from the acceleration response obtained by Rodriguez et al. [
22], which ranged from 31 G–46 G for the hardwood pallet. Rodriguez et al. [
22], however, featured different unit loads (bulk bins, drums, and corrugated boxes), which were not used in this research. This study focused on a kiln-dried, southern yellow pine, block-class pallet design and a plastic pallet design. Rodriguez et al. used a standard GMA wooden pallet design during their experiment. Stiffness differences between pallet designs could be an additional source of these variations in acceleration results.
Rodriguez et al. [
22] recommended the following equation to simulate the acceleration response in a pallet as produced by the impact of a forklift:
Data not included within these limits are not representative of impacts conducted at 0.30 m/s–1.22 m/s forklift speed or by half-sine shock events [
22]. The results obtained in this research are consistent with this interval.
3.4. Limitations of this Study
It must be noted here that this study did not investigate the direct effect of impacts on unit load integrity; thus, the severity of the recorded impact events is not suitable to define the possible damage of unit loads in general.
Determining a complete damage boundary curve for a unique unit load design requires extensive damage analysis based on appropriate test series.