For over 30 years the leading pallet and sawmill magazine in America.
Virginia Tech Study Compares Methods for Splicing Pallet Parts
Based on the results of a pilot study, Virginia Tech embarked on a larger study to fully examine the strength and stiffness of hardwood pallet parts assembled using both finger jointing and metal connector plate splicing technologies.
By Hamner, White, Rupert, Han
Date Posted: 2/1/2007
(Editor’s Note: This Virginia Tech column is related to the one carried in the February 2006 issue of the Pallet Enterprise. To keep this material as simple as possible and still get the results through to our readers, we have elected to place the entire column, along with accompanying data tables and pictures, on our Internet site www.palletenterprise.com. For brevity and simplicity, the version provided here is condensed from the column submitted by our friends at Virginia Tech.)
In February 2006, the Pallet Enterprise reported the results of a Virginia Tech pilot study that discussed the potential for finger jointing short pallet part segments into full size and longer pallet parts. The important finding from this research was that finger jointed hardwood pallet components at moisture levels between 12 and 37 percent can perform at flexural strength and stiffness levels approaching those of non-spliced solid wood. Considering the volume of wood material left over from pallet part operations (approximately 540 million board feet annually), splicing short sound segments to create full size and longer pallet parts has the potential to increase yields, add value, control rising raw material costs, and conserve wood fiber supplies.
Based on the results of the pilot study, Virginia Tech embarked on a larger study to fully examine the strength and stiffness of hardwood pallet parts assembled using both finger jointing and metal connector plate splicing technologies. This study has been completed and was funded by the USDA Forest Service, Wood Education Resource Center (WERC) in Princeton, WV.
The objectives of the Virginia Tech research were the following:
1. To compare the relative flexural strength (Modulus of Rupture – MOR) and stiffness (Modulus of Elasticity - MOE) of spliced hardwood pallet parts with whole non-spliced hardwood pallet parts. Spliced pallet part components were manufactured from short wood segments using both metal connector plates (unnotched stringers only) and finger jointing splicing procedures (unnotched stringers and deck boards).
2. To determine the influence of moisture content, wood species, and the number of splices per component on the performance of spliced hardwood pallet parts.
3. To determine the effect of finger joint and metal connector plate design on the performance of spliced pallet parts.
Pallet part segments were cut from randomly selected full sized parts and then spliced together using either finger jointing or metal connector plates. The original full-size parts were cut into either two or three equal length segments depending on the number of spliced joints required per component in the experimental design (one or two splices per component). All pallet stringers prepared and tested in this study were 2-way style (without notches). Also, the effect of tree growth wood related defects (i.e. knots) on joint integrity was not studied, and defects were not permitted at joint locations.
Mixed eastern U.S. red and white oak species and yellow-poplar pallet parts were selected because they are commonly used by U.S. pallet manufacturers and represent a range in wood density typical of the hardwood varieties used for pallet manufacturing. To minimize material properties variation within species, all pallet part components were purchased at the same time from one location and one production run. The parts were maintained in the green condition (dead-stacked, wrapped, and refrigerated). For one group of test samples, it was necessary to dry wood segments before splicing. For another group of test samples, it was necessary to dry pallet parts that had been spliced green before testing. For these two groups, a dry kiln was used to achieve an average MC of 12 percent (approximately 4 days at 140oF for deckboards, and 7 days at 140oF for stringers).
The tables showing experimental design and results of testing can be found in the Internet version of this article. The independent treatment variables are splicing technique used, number of joints per connection, moisture content (assembled versus tested), and wood species. Two equal length segments were used for parts spliced with one joint, and three equal length segments were used for parts with two joints. It is important to note that green MCs’ for both oak and yellow-poplar samples in this study were in excess of 60 percent. This was substantially higher than the MC of specimens evaluated in the previous pilot study.
For each spliced pallet part type (deckboard or stringer) there exist 18 treatment combinations (2 spliced connection groups x 3 MC groups x 3 species groups = 18 possibilities). Where mixed oak and yellow-poplar were used in combination with two joints per component, oak segments on each end of the component were joined with a yellow-poplar segment in the center.
Utilizing 10 replications for each test group, a total of 180 finger jointed deckboards, 180 finger jointed stringers, and 180 metal plate connected stringers were prepared and tested.
Finger Jointed Samples
All finger jointed samples used in this study were manufactured at Michael Weinig, Inc. located in Mooresville, NC. with a Grecon Profijoint D-110 finger jointer. Two possible finger joint orientations were used for both deckboards and stringers: horizontal and vertical. In the current study, the fingers were oriented vertically in both deckboards and stringers and were cut 10-mm deep with a 3.5mm tip-to-tip span. Two adhesive types were used based on the moisture content at the time of assembly.
Metal Connector Plate Spliced Samples
The metal connector plates (MCP) used to splice stringer segments into full size 48-inch stringers were supplied by Eagle Metal Products, Inc. (see Figure 1).
Metal connector plates were applied to pallet stringer segments with a portable hydraulic truss-chord plater (See Internet article for more details). Metal plate connectors were applied to each joint on each side of a pallet stringer test sample.
Control Samples (Non-Spliced Solid Wood)
Non-spliced solid wood pallet part samples were not prepared or tested in this study. Flexural strength (MOR) and stiffness (MOE) results for the solid wood classifications were obtained from previous research. To determine the effect of splicing on the performance of pallet parts, the flexural MOR and MOE of spliced parts were compared to the flexural MOR and MOE of non-spliced pallet parts of the same size, species, quality, and moisture levels.
All spliced pallet parts were tested to failure in static third point bending using a 10-kip servo-hydraulic test machine. Figure 2 shows the test setup for stringers (similar for deckboards). The procedures and fixtures that were used generally conform to ASTM D-198, “Standard Test Methods of Static Tests of Lumber in Structural Sizes.” Green spliced specimens were tested immediately after appropriate adhesive cure or plate applications (assembled green / tested green). A group of matched specimens were kiln dried first and then tested (assembled green / tested dry). A third group was kiln dried before splicing and tested dry (assembled dry / tested dry). A test span of 44 inches was used for all 48-inch stringers, and 36 inches for all 40-inch deckboards. For both deckboards and stringers tests, the rate of deformation was 0.5-in./min. After failure, the MC and specific gravity (SG) (oven dry basis) were determined. Flexural strength (MOR) and stiffness (MOE) were calculated for each piece.
The results of this study are presented in tables available in the Internet article. These tables contain descriptive statistics, maximum load, and average flexural MOR and MOE for all test treatments. These tables also contain the average specific gravity and moisture content of the test samples at the time of testing.
The Effect of Splicing Method on Bending Strength and Stiffness of Deckboards and Stringers
Finger jointed pallet components (deckboards and stringers) were stronger and stiffer than stringers spliced using metal connector plates. The finger jointed components performed as well as non-spliced solid wood. The average MOR for MCP spliced unnotched stringers was 66 percent of finger jointed components and 70 percent of non-spliced solid wood. The results also show that pallet parts spliced using metal connector plates had approximately 50 percent of the stiffness of solid wood and finger jointed parts. The average strength and stiffness of the stringers spliced with metal connector plates were the same at all moisture levels and for all species that were tested. The MOR and MOE were limited by the tensile strength and tooth withdrawal resistance of the plates themselves. Previous research has shown that notched stringers spliced with metal connector plates perform as well as notched solid wood stringers. In the case of notched stringers—spliced or non-spliced—the notch is the strength limiting defect. In the current study using unnotched stringers, the strength and stiffness of stringers were limited by the plates.
With few exceptions, MCP spliced stringers and finger jointed deckboards and stringers failed at the spliced joint. For MCP test samples, there were two modes of failure: 1) tooth withdrawal—plate teeth pulling out of the wood, and 2) plate failure—tension fractures in the plate initiated at the bottom edge of the plate web when stressed in bending. The ratio of tooth withdrawal to plate failure was approximately 50:50. Figures 3 and 4 illustrate these two MCP failure modes. These failure modes occurred in dry and green, as well as oak and yellow-poplar test samples. A longer plate with more and longer teeth would improve the resistance to failure caused by tooth withdrawal. A heavier gauge steel plate, or fewer teeth in the region of the splice, would improve the tensile strength of MCPs used to splice pallet parts. New plate designs should be tested.
The failure mode for finger jointed components was either glue line joint failure or a combination of wood and glue failure. As mentioned, all parts that were assembled in the green condition and either tested in the green condition or allowed to dry before testing (Green-Green and Green-Dry groups) failed in the glue line of the splice. For parts that were assembled dry and tested dry (Dry-Dry group), failure also mostly occurred at the splice location. However, these failures were a combination of wood and glue line failure. The activated PVA adhesive used to finger joint dry pallet parts provided a strong finger joint. A few Dry-Dry samples failed at defect locations (knots) other than the spliced joint.
Most spliced pallet parts had relatively low variation for strength and stiffness. This is due to the characteristic of the splice as the critical defect. Therefore, the MOR and MOE of spliced pallet parts are more predictable than for solid wood of similar size, MC, and species. However, the implications of this on the design load criteria for spliced pallet parts are beyond the scope of this study.
The Effect of the Number of Spliced Connections on Flexural MOR and MOE of Spliced Deckboards and Stringers
The number of splices (one or two) had little effect on the flexural MOR and MOE of spliced pallet components.
The Effect of Moisture Content on Flexural MOR and MOE Finger Jointed Deckboards and Stringers
Moisture content at the time of testing (green versus dry) had a significant effect on the average strength of finger jointed deckboards—the higher the wood MC the lower the MOR. The average bending strength of deckboards tested green was approximately 50 percent of the strength of deckboards that were tested dry. The flexural strength of the two finger jointed MC groups that were tested dry are similar to non-spliced deckboards.
Finger jointed deckboards tested in the dry condition were significantly stiffer than those tested green. Deckboards tested in the green condition had an average of 83 percent of the MOE of those tested dry. The moisture content at the time of splicing finger jointed deckboards (green or dry) did not have a significant effect on strength and stiffness when they were tested in the dry condition.
The average flexural strength of finger jointed stringers was significantly different between all three moisture content groups. Finger jointed stringers that were assembled and tested in the green condition (Green-Green) had 59 percent of the strength of stringers assembled green and tested dry (Green-Dry), and 33 percent of the strength of those assembled dry and tested dry (Dry-Dry). The strength of finger jointed stringers that were assembled dry and tested dry (Dry-Dry) was comparable to solid wood. The average strength of finger jointed stringers assembled green and tested dry (Green-Dry) was 57 percent of the strength of those assembled dry and tested dry (Dry-Dry).
Moisture content significantly influences the stiffness of finger jointed stringers. The average stiffness of finger jointed stringers in the Dry-Dry MC group was comparable to that of solid wood. The average stiffness for stringers in the Green-Dry group was 33 percent of the stiffness of stringers in the Dry-Dry group. Finger jointed stringers that were assembled green had significantly lower average MOR and MOE than those that were assembled dry.
The results indicate that full size finger jointed pallet deckboards and stringers that are assembled in the dry condition (average 8 to 20 percent MC) are as strong and stiff as non-spliced pallet parts at the same MC, for the same species, and of the same quality. The results also indicate that the Resorsabond® 4214/4552 adhesive systems chosen for this study was not effective for gluing finger jointed stringer segments at moisture levels in excess of 60 percent, although finger jointed deckboards assembled green performed well when allowed to dry. It is unclear why the finger jointed splices in deckboards were less sensitive to moisture at the time of assembly. It is likely that the deckboards dried faster than the stringers resulting in better bonding at the joint.
Examination of failed stringers revealed many joints with starved glue lines. This was likely caused by excessive moisture present in the joint when the adhesive was applied. While this adhesive system did not bond joints at high wood moisture levels, the previous pilot study showed that this adhesive system is effective for finger jointing wood below 40 percent MC.
The flexural strength and stiffness of samples spliced with metal connector plates were not affected by moisture content. The strength and stiffness of these samples were limited by the performance of the plates themselves.
The Effect of Wood Species on the Flexural Strength and Stiffness of Spliced Stringers and Deckboards
There were no differences in the average flexural strength and stiffness between spliced oak, yellow-poplar, and mixed species samples. However, the finger jointed oak specimens in the MC subgroup Green-Green performed consistently lower than all other test specimens. The relatively porous and dense structure of oak wood likely contributed to the lack of adhesive in these finger jointed specimens.
• Unnotched stringers spliced using metal connector plates are not as strong as non-spliced, unnotched, solid wood stringers. The racking strength of 2-way pallets using stringers spliced with metal connector plates will not be as strong as pallets made with solid wood. However, the performance of such pallets may be adequate when used in other conditions.
• For dry pallet part segments, finger jointing provides a stronger and stiffer spliced joint than metal connector plates. Dry finger jointed pallet parts are as strong and stiff as non-spliced solid wood. Finger jointing is a viable method for splicing short dry pallet part segments into full size and longer pallet parts.
• Finger jointing green segments (at moisture contents above 60 percent) to make full size and longer pallet stringers was not successful using Georgia Pacific’s Resorcinol adhesive, even when allowed to dry before testing. The results of the previous pilot study show that the strength and stiffness of finger jointed pallet parts at moisture contents below 40% approach those of solid wood.
• The number of splices, one or two, had no effect on the average flexural strength and stiffness of spliced pallet parts.
• The wood species tested had no effect on the average flexural strength and stiffness of spliced pallet parts.
Pallets manufactured using finger jointed or MCP spliced components should be tested to verify the implications of this research. For more information or to request copies of Virginia Tech reports, contact Peter Hamner at the Virginia Tech – Center for Unit Load Design (phone: 540-231-3043; email: email@example.com).