Printer Friendly

Withdrawal strength of punched metal tooth plates in red oak end grain. (Wood Engineering).


Split-control end plates are toothed plates that are inserted into the ends of railroad crossties to prevent the expansion or formation of end splits. Rail vibrations combined with moisture cycling can cause the end plates to withdraw or back out from the ends of the crosstie. Two types of end plates were tested in direct, static withdrawal to determine their withdrawal strength. The Type P plate had a 20 percent greater withdrawal strength than the Type Q plate. A method of testing toothed plate products in direct withdrawal was also presented.


Railroad crossties are cut from green logs and must be air-dried (seasoned) before pressure preservative treatment. During the seasoning process, stresses caused by uneven drying of the timbers can cause checks and splits in the ends of a crosstie. These checks and splits can expand while the tie is in service and lead to failure of the tie. A split-control end plate is a device that is inserted into the end of a railroad crosstie to prevent the formation of new splits or the expansion of existing splits. Split-control end plates, here after referred to as end plates, must resist the lateral forces caused by the splitting and distortion of the wood cross section, and they must resist the withdrawal or "backing-out" forces caused by a combination of rail vibration and shrink/swell cycles.

Crosstie end plates are similar in appearance to metal connector punched tooth plates (truss plates) used in the wood truss industry. End plates are stamped from galvanized structural steel sheets and have teeth formed perpendicular to the plate surface. End plates are generally installed by machine into the ends of crossties using hydraulic pressure. Usually, the tie is hydraulically squeezed on all four faces near the ends just before the plate is inserted. This squeezing closes any existing checks and splits; additionally, it is believed that the withdrawal resistance of the plates is increased when the ends of the ties are squeezed. The squeezing of the tie "prestresses" the wood fibers, thereby maintaining the friction forces on the teeth even after the tie has dried below the fiber saturation point.

The same perpendicular-to-grain drying stresses that cause checks and splits subject end plates to lateral forces. These forces can cause failure of the end plate if they exceed the tensile strength of the steel or the lateral resistance strength of the teeth; these failure modes are similar to some of the failure modes for truss plate connectors. Generally, failures of these types will occur before the tie is placed in service (installed in the track). That is, if a particular tie will split with enough force to cause a steel or lateral resistance failure, it will occur before the tie is installed. Thus, failures of these types are rarely seen on ties in-service.

The most common way that an end plate will fail in track is to back out from the end of the tie; that is, the plate surface does not remain flush with the tie end, and partially or even completely withdraws from the tie. Plate backout is affected by the considerable vibration of the tie caused by the passing trains, and the shrink/swell mechanism caused by exposure to widely varying moisture conditions. Thus, it is important that a split control end plate be able to resist withdrawal forces as well as lateral forces. While it has not been proven by research, we believe that static withdrawal strength of end plates from ties is a logical and meaningful predictor of the withdrawal behavior of end plates in-service.


The objective of this study was to determine the withdrawal strength of two types of end plates having the same number of teeth but different tooth patterns, tooth angles, and tooth widths. All other end plate properties and timber samples were matched to the extent possible so as to produce a set of paired samples for statistical comparisons.


While there has been no previously published research on the withdrawal strength of end plates installed in the end grain of timber crossties, there have been several studies of the fastener backout phenomenon. Fastener backout is characterized by the movement of a driven-type fastener in the opposite direction from which it was driven. Nail backout from siding, trim, deckboards, and other exterior wood is a well-known phenomenon, and is caused by fiber relaxation after driving followed by swelling of wetted wood. In some cases, the nail can be completely withdrawn.

To illustrate the typical mechanism that causes fastener backout, consider a typical nailed connection such as a nail driven through a patio deckboard into a joist. Immediately after driving the nail into the deckboard and joist, the wood fibers surrounding the nail in both the deckboard and the joist create friction force by pressing against the nail shank. As the lumber dries, the wood fibers shrink away from the nail shank, resulting in less friction force. When the joist is exposed to moisture, the joist expands in depth, drawing the nailhead downward against the deckboard and slightly drawing the nail back out of the joist. Later, the deckboard dries out and the thickness of the deckboard decreases. With a small gap between the deckboard and the joist present, it is easily closed by gravity and occasional foot traffic. When again exposed to moisture, the wood fibers of the deckboard expand, gripping the nail shank and at the same time, the joist swells in depth, causing the nail to be pulled from the joi st. Upon redrying, the deckboard can move down on the nail causing the head to protrude yet more. Thus, the nail is drawn farther out of the joist upon each cycle of wetting and drying.

In the case of the end plate on a tie, lateral forces in the plane of the end plate can be relieved to some extent by bending of the teeth when the teeth are "worked out" of the end grain from swelling of the end grain parallel to the grain. (This action is analogous to "nail pop" discussed later.) Even though parallel-to-grain swelling is small compared to the perpendicular directions, it will occur frequently due to the high permeability of end grain to water absorption. Upon subsequent end grain drying, it is unlikely that lateral forces induced in the end plate by crosstie splitting and checking will drive the teeth back into the end grain.


Fiber relaxation and changes in moisture content greatly affect withdrawal resistance of nail shanks (FPL 1999). When pulled soon after driving, the withdrawal resistance of a nail is about the same, whether it is driven into green wood or seasoned wood.

Suddarth and Angleton (1956) presented a theory to explain the mechanics of nail pop (backout), and conducted four experiments to test the proposed theory. The depth of penetration of the nail tip and the amount of wood shrinkage was found to be directly proportional to the amount of nail pop. Also, exposure to repeated moisture cycling was found to have a cumulative effect on nail backout. It should be noted that their theory and experiments involved nails driven into the side-grain of lumber, not the end grain.


According to Quaile and Keenan (1979), truss plate backout can be caused by shrinkage across the grain of the lumber. This shrinkage compresses the truss plate across its width, causing the plate to arch upward thus pushing the center portion of the plate out slightly from the wood.

Groom (1994) subjected wood truss joints to eight moisture cycles from 5 to 19 percent moisture content (MC), and found that truss plates backed out a distance of about 17 percent of the tooth length. Gangnail GN2O 20-gauge truss plates (3 in. wide by 4 in. long, with 0.360-in, long teeth) were used with 2 by 4 No. 2 southern pine lumber at 12 percent MC to fabricate 300 tensile test joints. The joints were divided into three groups; one group was stored at a constant 12 percent MC, one was cycled between 9 and 15 percent MC, and the third was cycled between 5 and 19 percent MC. Plate backout was slight for the 9 to 15 percent MC joints (about 2% of tooth length), and was 1 percent for the constant 12 percent MC joints. Groom (1994) attributed the plate backout to shrinking and swelling stresses in the lumber.


Two types of end plates were investigated in this study; these plates were the only commercially available end plates of which the authors were aware. The end plates, referred to as Type P and Type Q in this paper, were very similar; both were stamped from 18-gauge (0.047-in. thick) steel sheet of the ASTM A653 Grade 40 designation, with ASTM A924 G60 galvanized coating. Although the Type P and Type Q end plates are made from the same strength grade of steel, the actual strength of a steel sheet coil can vary significantly from the nominal strength (Skaggs et al. 1995). The nominal strength is the minimum acceptable strength for that grade, so it is possible to have one coil that meets the minimum strength and another that exceeds the minimum by 20 percent or more. Thus, an attempt was made to select plates made from steel coils with similar properties. For the Type P plates used in this study, the yield point was 49.5 ksi, the ultimate strength was 63.3 ksi, and the elongation was 37.9 percent. For the Type Q plates used in this study, the yield point was 53.5 ksi, the ultimate strength was 62.5 ksi, and the elongation was 33.6 percent. The strength data for each coil were obtained from the metallurgical test reports provided by the steel coil manufacturer.

Table 1 lists the properties of the end plates used in this study, and Figures 1 and 2 show the Type P and Type Q plates, respectively. The Type P plates are 6 inches wide and 7 inches long, while the Type Q plates are 6.2 inches wide and 7.125 inches long. Both plate types have 180 teeth per plate, although the tooth patterns are different; the Type P plates have a slightly staggered tooth pattern, while the Type Q plates have an aligned pattern. Also, the Type P plates have 10 rows with 18 teeth per row, and the Type Q plates have 9 rows of 20 teeth each. The teeth of both plate types are approximately 0.563 inch long; however, the teeth of the Type P plate are slightly wider than those of the Type Q plate. The teeth of the Type P plate are angled at nearly 90 degrees to the plate surface, while the Type Q plates have teeth angled at about 85 degrees to the plate. The fact that Type Q plates are at a slight angle may cause some slight damage to the end grain fibers when the plates are embedded. However, in this experiment, it was not possible to separate out or partition the effects of differences in tooth angle and tooth alignment on the ultimate tooth withdrawal strength of the Types P and Q end plates.

Twelve end plates of each type were cleaned with mineral spirits to remove any oil that may have remained on the plates from the stamping process. Each plate was washed with clean mineral spirits for about 30 seconds, then allowed to drip-dry for several minutes. Finally, the plates were dried with compressed air to remove all traces of the mineral spirits.


Ten red oak crossties were obtained from a tie manufacturer in Alabama. The tie manufacturer randomly selected the ties from normal production stock. The ties were 7 by 9 inches by 8 feet 6 inches, incised, green, and of the Grade 5 designation. The ties were shipped to the test facility in Florida about i week after manufacture.

Six of the 10 ties were selected for testing. Five 18-inch sections were cut from each tie using a horizontal bandsaw. The tie samples were cut as shown in Figure 3. Because lumber loses moisture fastest at its ends, the ends of the ties were discarded. All of the tie samples were in the green moisture condition, as evident by the free moisture on the sample ends immediately after cutting.

Immediately after cutting, each sample was wrapped in a plastic bag to maintain the green moisture condition. The wrapped tie samples were stacked onto a pallet and the pallet was wrapped with plastic stretch-wrap. The pallet was stored in the lab facility for 1 month, after which the plastic wrap and bags were removed. Because there was some free water on the surface of the ties, the ties were stacked so that the surface moisture could dry before inserting the end plates. The end plates were inserted 2 days after unwrapping the tie samples.

Table 2 shows how the test blocks were cut from the tie samples. The six ties were numbered 1 through 6, and the five test sample sections cut from each tie were labeled A through E. Thus, the second sample cut from the first tie was labeled 1B. From each tie, one sample was used to make a test sample with Type P plates, and one sample was used to make a test sample with Type Q plates. The remaining tie samples were used for another study.

For this study, it was not feasible to test the end plates in full-length crossties, which are typically 8 feet, 6 inches long. The only difference between the test samples and actual ties was the length, and this difference should have minimal or no impact on tooth withdrawal test results because the teeth interact with only the first inch or so of the tie end. Further, the squeezing mechanism only squeezes the ends of the ties (within 6 in. of the end). If the ties were put in service and subjected to drying stresses and vibration, then the length of the tie could possibly interact with the plate type and have an effect on the results of a mechanical test.


A hydraulic test machine was designed and built for the purposes of installing end plates into tie samples and testing the tie samples. This machine is somewhat similar in appearance to a typical universal test machine, except that it is divided into an upper portion and a lower portion. The upper portion is used to install the end plates into the tie sample, and the lower portion is used to test the tie sample. The two portions are divided by a movable crosshead. A main hydraulic cylinder moves the crosshead upward and downward relative to a fixed test base near the bottom of the machine and a fixed base at the top of the machine.

To install end plates into a tie sample, the sample was placed in the upper portion of the machine, and the crosshead was moved up, thereby pressing the end plates into the sample. A squeeze mechanism located in the upper portion of the test machine uses another hydraulic cylinder to close the squeeze mechanism around the tie sample. The squeeze mechanism is very similar to that on a commercial end-plater machine. The squeeze mechanism squeezes both ends of the tie sample at the same time, and maintains the squeeze pressure while the end plates are inserted.

The test machine was controlled by a hydraulic control system that included adjustable pressure relief and flow control valves. A pressure transducer was used to measure the pressure during plate insertion, and a strain gage load cell was used to record the withdrawal force during testing. The movement of the crosshead was measured with an LVDT displacement transducer. A data-acquisition card installed in a personal computer recorded the hydraulic pressure, force, and displacement data.


To test the end plates in withdrawal, some means of applying a withdrawal force parallel to the teeth must be achieved. For this project, the best solution was to securely attach a base plate to the "back" of the end plate (that is, the plate face without teeth), and then attach the base plate to the test machine. The base plate is a 7- by 9- by 1-inch steel plate with a threaded bolt hole in the center. An end plate with attached base plate was installed into each end of a crosstie sample; when the tie sample was placed into the test portion of the test machine, a bolt was connected to each base plate so that withdrawal force can be applied to each end plate, as shown in Figure 4.

Several methods of attaching the end plates to the base plates were investigated. The first method was to solder the end plate to the base plate. This method was rejected for several reasons. For the solder to melt and flow, the end plate must be subjected to high temperatures, which caused the teeth to glow red-hot. Also, the flux used to facilitate soldering left residue on the plate and teeth. Finally, the method required considerable time and labor.

The second method was to epoxy the end plate to the base plate. This method was marginally successful, but required significant surface preparation. Furthermore, the epoxy seeped up through the tooth slots of the end plate and prevented full insertion of the end plate.

Another method involved drilling holes in a piece of sheet steel to fit the tooth pattern of the end plate. This sheet was then placed over the end plate and base plate, and the edges of the sheet were folded over the sides of the base plate and attached to the base plate with machine screws. This method worked, but the withdrawal force was not applied uniformly to all of the teeth; that is, the teeth at the outer perimeter of the plate were loaded and withdrawn first, followed by the inner teeth withdrawing. This method was rejected because we wanted to load all of the teeth evenly at the same time.

The final method investigated was to use small machine screws to attach the end plate to the base plate. These screws were spaced throughout the end plate and were inserted through the tooth slots of the end plate into tapped holes in the base plate. Originally, some prototype base plates were drilled and tapped for 80 screws, but it was determined that 70 screws were sufficient to hold the end plate to the base plate during withdrawal.

The screws used were stainless steel truss head machine screws, with 5-40 thread and a 3/8-inch length. Truss head screws have a low, rounded head with a flat bearing surface, and have the largest head diameter of all machine screw head styles. The head height of these screws was approximately 0.078 inch, and the head width was approximately 0.289 inch.

Because the heads of the screws protruded above the surface of the end plate as can be seen in Figure 4, the teeth could not be fully inserted. However, it was found during preliminary testing that the screw heads were partially embedded into the end of the tie sample, and that the amount of gap was uniform across the plate area.

The screws were inserted with an adjustable-torque cordless screwdriver; the same torque setting was used for all screws. It was found that if the screws were tightened very securely, the angle of the teeth tended to change as the teeth "leaned in." This angling was more evident in the Type Q plates. Thus, the selected torque setting resulted in tightening the screws enough to keep them secure but not enough to cause significant change in the tooth angle. The screwdriver used was a Black & Decker model VP760, and the torque setting was 6.


The test samples were fabricated about 1 month after receiving and cutting the ties. About I week before installing the end plates, they were attached to the base plates using the machine screws. To fabricate a test sample, a tie sample was first placed into the upper portion of the test machine. Then an end plate with attached base plate was placed under the bottom end of the sample, and another end plate with base plate was placed on the top end of the sample. Two bolts were installed into each base plate to help maintain the proper position of the base plate relative to the tie sample. The crosshead was moved upward until the tie sample was centered in the squeeze mechanism. The hydraulic pressure was set to 2,300 psi and the squeeze was applied until the pressure reached the maximum setting. With the squeeze held, the crosshead was moved upward to press in the end plates. As soon as the maximum pressure was reached, the crosshead was moved down enough to release the pressure, and then the squeeze mechanis m was opened. The data-acquisition computer recorded the pressure and crosshead position during the installation process.

Figure 5 shows a typical graph depicting the rate of loading in terms of cylinder pressure used to embed the end plates for the two red oak test samples cut from tie 6. The insertion pressure increased at a similar rate for both samples, evident from the similar shapes of the curves. The change in the slope of the curves at approximately 1,500 psi cylinder pressure indicates that the teeth were nearly fully inserted at that pressure. Both end plate types, P and Q, required about the same cylinder pressure to embed the teeth based on observing the pressure versus time plots for all 12 test specimens.

After fabrication, the samples were stored on pallets in the lab facility. The samples were stood on end, and spaced several inches apart to allow for good air circulation.


The samples were tested 15 days after inserting the end plates. Each sample was placed in the test area of the hydraulic test machine. A bolt was inserted through the lower fixed platen of the machine into the threaded hole in the center of the base plate attached to the bottom of the tie sample. Another bolt was inserted through the load cell and the movable crosshead into the base plate on the top of the sample. A set of spherical washers was used under each bolt head. This allowed for some misalignment and prevented eccentric loading of the test sample.

The flow rate of the hydraulic system was controlled throughout the test. Initially, load was applied relatively quickly until about 3,000 pounds was reached, after which point the flow rate was decreased so that the cylinder moved at approximately 0.002 inch per minute. Once the ultimate load was reached and the load level began decreasing, the flow rate was increased until one of the end plates was fully withdrawn from the tie sample. Figure 6 is a typical graph of withdrawal load versus crosshead displacement; the loading rates are constant and similar for both plate types.

All test samples failed in the same manner. As the ultimate load was approached, some gap was evident between the end plates and the tie ends. After the load passed the ultimate level and began dropping, one of the end plates would be visibly withdrawing from the tie sample. The loading continued until that plate was completely withdrawn.


The average ultimate withdrawal load for the Type P samples was 12,502 pounds, which is 20 percent greater than the average ultimate withdrawal load of 10,425 pounds for the Type Q samples. To determine if this difference was statistically significant, two statistical tests were conducted; the paired-sample t-test and the Wilcoxon signed rank test (Hollander and Wolfe 1972). These tests were appropriate because two paired samples were selected from each tie (for example, test samples P-1B and Q-1A), resulting in sample pairs with nearly identical wood grain. This experimental design reduces the overall variance of the experiment because the variability stemming from lumber property variation between crossties is not present, and variability stemming from spatial lumber property variation within a single crosstie is reduced. Both tests showed that Type P end plates had a statistically significant greater withdrawal strength than the Type Q plate (the p-values were 0.017 and 0.018, respectively). The coefficien t of variation for the ultimate withdrawal strength was 13 percent for the Type P plate and 9 percent for the Type Q plate.

There are several possible explanations for the greater withdrawal strength of the Type P plates. The teeth of the Type P plate are wider than those of the Type Q plate (0.156 in. vs. 0.140 in., respectively). Withdrawal of common nails is proportional to nail diameter, which is related to surface area (FPL 1999). Because of the wider tooth, the Type P tooth had more surface area than the Type Q tooth; this may explain the greater withdrawal resistance. Another possible explanation for the difference in tooth withdrawal strengths is the different tooth angles. The Type Q tooth has a greater angle with respect to the plate surface than the Type P tooth, which may have caused some damage to the wood fibers during embedment. Finally, the staggered tooth pattern of the Type P plates may be responsible for the increased strength.


Two types of end plates used to control splits in the end grain of railroad crossties were tested in static withdrawal to determine and compare their withdrawal strength. Several methods of pulling the end plates from the crosstie samples were investigated; the best method was to use machine screws to attach the end plate to a steel base plate. The end plates had similar properties but different tooth patterns, tooth widths, and tooth angles. Crosstie samples were matched to produce a set of paired samples for statistical comparisons. The Type P plates had a 20 percent greater withdrawal strength than the Type Q plates, and the difference was statistically significant. The differences in tooth width, tooth pattern, and tooth angle between the plate types are likely responsible for the difference in withdrawal strength. One possible extension of this research may be to install Types P and Q end plates into opposite ends of a sample of crossties, place the ties into service, and periodically observe the amount of plate backout.



Properties of Type P and Q end plates.

Property Type P Type Q

Steel gauge 18-gauge 18-gauge
Minimum steel thickness (in.) 0.0470 0.0470
Steel grade ASTM A653 Gr. 40 ASTM A653 Gr. 40
Galvanized coating ASTM A924 060 ASTM A924 660
Plate width (nominal) (in.) 6 6.2
Plate length (nominal) (in.) 7 7-1/8
Tooth length (approx.) (in.) 0.563 0.563
Tooth width (approx.) (in.) 0.156 0.140
Tooth pattern Staggered Aligned
Teeth per row 18 20
Rows of teeth 10 9
Teeth per 6 x 7 plate 180 180
Actual steel yield point (ksi) 49.5 53.5
Actual ultimate strength (ksi) 63.3 62.5
Actual elongation of steel (%) 37.9 33.6

Labelling of test samples by end plate type (P or Q) and source of red
oak samples taken from six crossties.

Tie no. Section A Section B Section C Section D Section E

 1 Q-1A P-1B
 2 Q-2B P-2C
 3 Q-3C P-3D
 4 Q-4D P-4E
 5 P-5A Q-5E
 6 P-6B Q-6C

Ultimate withdrawal load matched specimens from the same crosstie.

End plate type Sample ID Ultimate load Ultimate load ratio (a)

 P lB 13,526 1.23
 Q 1A 10,978
 P 2C 14,088 1.32
 Q 2B 10,687
 P 3D 12,174 1.10
 Q 3C 11,067
 P 4E 13,017 1.17
 Q 4D 11,156
 P 5A 12,754 1.27
 Q 5E 10,075
 P 6C 9,450 1.10
 Q 6B 8,588

(a) Ratio of ultimate load for matched specimens P and Q. Specimens
were matched by crosstie lumber.


Groom, L.H. 1994. Effect of moisture cycling on truss-plate joint behavior. Forest Prod. J. 44(1):21-29.

Hollander, M. and D.A. Wolfe. 1972. Nonparametric Statistical Methods. John Wiley & Sons, Inc., New York.

Quaile, A.T. and F.J. Keenan. 1979 Truss plate testing in Canada: Test procedures and factors affecting strength properties. In: Proc. Metal Plate Wood Truss Conf., St. Louis, MO. Forest Prod. Res. Soc., Madison, WI. pp. 105-112.

Skaggs, T.D., F.E. Woeste, and S.L. Lewis. 1995. Steel properties used to manufacture wood truss metal connector plates. Forest Prod, J. 35(1):187-195.

Suddarth, S.K. and H.D. Angleton. 1956. Nail popping, a result of wood shrinkage. Sta. Bull. 633. Purdue Wood Res. Lab., West Lafayette, IN.

USDA Forest Service, Forest Products Laboratory (FPL). 1999. Wood Handbook: Wood as an Engineering Material. Gen. Tech. Rept. FPL-GTR-113. Forest Prod. Soc., Madison, WL.


* Forest Products Society Member.

The authors are, respectively, Research & Development Engineer, Robbins Engineering, Inc., 13025 North Nebraska Ave., Tampa, FL 33612; and Professor, Biological Systems Engineering Dept., Virginia Tech, Blacksburg, VA 24061. This paper was received for publication in June 2001. Reprint No. 9329.

[c]Forest Products Society 2002.

Forest Prod. J. 52(10):82-88.
COPYRIGHT 2002 Forest Products Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2002 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:O'Regan, P.J.; Woeste, F.E.
Publication:Forest Products Journal
Date:Oct 1, 2002
Previous Article:Vibration testing of timber floor systems. (Wood Engineering).
Next Article:Torsional strength of dowel joints constructed of plywood and oriented strandboard. (Wood Engineering).

Terms of use | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters