Printer Friendly

Behavior of metal-plate-connected wood truss joints under wind and impact loads.

Abstract

The objective of this research was to understand the behavior of metal-plate-connected (MPC) wood truss heel and tension splice joints subjected to dynamic loads that simulated hurricane wind and impact loads. The general loading procedure was to ramp load the joints to dead/design load level, apply dynamic load, unload to zero, and finally apply ramp load to failure. The stiffness of the joints before and after the dynamic loads was compared. Hurricane wind loads caused an average increase in stiffness of 300 percent, likely due to wood densification near the metal teeth. The impact load caused a stiffness increase similar to that produced by the hurricane wind loads. Dynamic loadings on heel joints caused no significant strength degradation. For tension splice joints, the accelerated ramp load produced the same results as the static ramp load in one-tenth the time. It is suggested that the static tests of MPC joints may be conducted in 1-minute to failure, like lumber, instead of 10-minute to failure as recommended by the current standard.

**********

Metal-plate-connected (MPC) wood trusses are frequently used to construct roof systems for many types of structures, including residences, apartments, and light commercial buildings. Although the MPC truss is considered an engineered product, the dynamic behavior of both the joints and the overall truss is largely unknown. This paper describes the behavior and failure modes of joints loaded with dynamic forces that simulate hurricane wind and impact loads on residential structures.

Most research on MPC joints has focused on tension splice joints tested with static loads (Gupta et al. 1996), although some studies have investigated the behavior and property changes that occur when MPC joints are loaded dynamically (Dagher et al. 1991, Emerson and Fridley 1996, Freilinger et al. 1997, Kent et al. 1997). Kent et al. (1997) tested MPC heel and tension splice joints with a variety of dynamic loads, including a historical earthquake simulation, an artificially generated earthquake simulation, sequential phased displacement (SPD) loading, and cyclic loads. They found no strength degradation related to the earthquake simulations, but they observed a slight difference in the heel joint axial stiffness. The damage that accumulated in the connection during the SPD load depended on the level of displacement. Large cyclic loads caused significant strength loss in MPC joints. Freilinger et al. (1997) used cyclic loads to evaluate the duration of load factor for wind and earthquake loads on MPC heel and tension splice joints and found the factor (1.6) to be adequate. They also found significant stiffness degradation for both joints, although they observed no strength degradation.

Materials and methods

Materials

In this study, we tested two types of joints: heel joints (HJs) and tension-splice joints (TSJs). Table 1 lists the groups of tests performed and the sample sizes for each group. The TSJs were connected with 76- by 102-mm plates, whereas the HJs were connected with 76- by 127-mm plates. All test joints were fabricated from machine-stress-rated Douglas-fir (1800f-1.6E) lumber conditioned to an equilibrium moisture content of approximately 12 percent. After fabrication, the joints were stored in the standard environment room for at least 7 days before testing (TPI 1995). The modulus of elasticity of the wood for the MPC joints was measured with an E-computer (Model 390, Metriguard, Pullman, Washington) and randomly assigned to various test groups listed in Table 1.

Apparatus

To apply loads to the joints, a horizontal testing frame (Fig. 1), developed by Gupta and Gebremedhin (1990) was used along with a 49-kN-capacity dynamic hydraulic actuator. Deflection sensors and load cells were used to measure joint displacements and member forces, respectively. HJs were tested with tension or compression in the top chord. When HJs were tested with tension in the top chord, they were tested upside down, as shown in Figure 1a so that uplift could be prevented by a pin/roller support. For compression in the top chord of the HJ, the joint was tested right side up (not shown here) and the bottom chord was supported by the pin/roller support. TSJs were tested as shown in Figure 1b.

[FIGURE 1 OMITTED]

Determination of joint forces

The truss examined in this study was a 9.1-m span, 4/12-slope Fink truss (2.4 m aboveground) composed of 38- by 89-mm Douglas-fir lumber. Each truss had a 290 N/m dead load on both top and bottom chords. The simulated location of the truss was a 240-km/hr. wind zone. A linear elastic finite element analysis software (Computers and Structures Inc. 1991) was used to determine forces at the HJs and TSJs. The truss was analyzed for both components and cladding and main wind force resisting system loads. We included six load combinations in the analysis. We used each load case twice for each truss by placing TSJs on the windward and leeward sides.

Table 2 shows the maximum forces in HJs for two different load cases. Magnitude of the forces in TSJs was very small compared with the published strength (Gupta et al. 1996). Therefore, we did not test TSJs under wind loads. To test HJs, we transformed the forces given in Table 2 into dynamic time histories, which we accomplished using real-time wind speed data from Hurricane Bob, provided by the U.S. Army Corps of Engineers (1991). Additional details of the wind analysis are given in Redlinger (1998).

[FIGURE 2 OMITTED]

[FIGURE 3 OMITTED]

Heel joint loads

For this project, we defined HJ tests by the type of force exerted on the top chord of the HJ. Compression tests apply compression forces to the top chord and tension forces to the bottom chord. Tension tests apply tension to the top chord and compression forces to the bottom chord. We defined slip deflection as the distance the top chord moves with respect to the bottom chord, measured in the longitudinal direction of the top chord. Rotation measures how much the top chord rotates with respect to the bottom chord.

Static ramp load. -- A static ramp load of 4 kN/min. was applied to cause failure in 5 to 10 minutes.

For the compression static ramp load test, we subjected a group of 10 joints to compression static ramp load in the top chord to establish a control group for all dynamic tests in which the top chord was subjected to compressive load.

For the tension static ramp load test, we subjected a group of 10 joints to tension static ramp load in the top chord to establish a control group for all dynamic tests in which the top chord was subjected to tensile load.

Hurricane wind load. -- We performed hurricane wind load simulation tests in three stages: beginning, wind, and ending (Fig. 2). During the beginning stage, we applied a ramp load to the HJ at 4 kN/min. When the load in the joint was equal to the design load (one-third of the average ultimate load from the static tests), we stopped the ramp. This stage was used to determine the beginning dead load and beginning design load stiffnesses (Fig. 2). The beginning dead load stiffness is the secant stiffness of the joint before the wind load simulation is applied, measured at the dead load (6.5 kN in top chord). The beginning design load stiffness is the secant stiffness of the joint, before the wind simulation is applied, measured at the design load of individual joints.

During the wind stage, the hurricane wind load simulation was transferred from computer memory to the hydraulic actuator controlling system. The wind stage, the dynamic portion of the loading, lasted for approximately 2 hours and 15 minutes for each joint. The purpose of this stage was to apply the hurricane loading and measure the "dynamic deflection" (Fig. 2), which is the difference in the deflection before and after the wind load. Both of these deflection readings were taken at the design load, which is the beginning and ending load of the wind load simulation. The ending stage began after the wind load simulation finished. After the load returned to zero, the ramp load was restarted at 4 kN/min. and continued until failure occurred.

For the compression hurricane wind load tests (top chord in compression), the compressive load ranged from approximately 7.1 kN to 9.6 kN. Figure 3 shows a graphical representation of the wind loading stage. Only a small part of the hurricane loading function is included here because of the extremely large size of the full plot.

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

For the tension hurricane wind load tests (top chord in tension), the load in the wind stage ranged from approximately 0.2 kN to 17.4 kN (Fig. 4). Again, only a small part of the hurricane loading function is included in the plot.

The tension test had a higher maximum load and a lower minimum load than the compression test. Although the average force for the tension test was slightly lower than the average load for the compression test, the larger load range made the tension test potentially more damaging to the HJ.

Impact load. -- For the double-design tension impact load, we performed the impact tests in tension (top chord in tension) because the results from the compression wind load tests showed little difference when compared to the static group results. We based the dynamic load for this test on design factors found in the National Design Specification (NDS) for Wood Construction (AF & PA 1997). We set the maximum value of the spike at double the design value, having found the design value by dividing the average strength from the static test HJ group by 3.0 (TPI 1995). For HJ double-design impact load tests, the peak load was 21.3 kN. Figure 5 shows an idealized load deflection plot, along with properties measured during the test.

During the initial stage of the test, we loaded the joint to 6.5 kN (simulating the dead load) at 4 kN/min. This part of the test determined the beginning dead load stiffness. We then spiked the load from dead load to double the design value in 1 second, held the load at double the design load for 1 second, and returned it to the dead load in 1 second. This 3-second duration of loading is consistent with the definition of "impact" given by the NDS (AF & PA 1997). We determined the "dynamic deflection" as the deflection reading taken immediately after finishing the impact load minus the deflection reading taken immediately before applying the impact load.

After the impact load, we unloaded the joint and applied a static ramp load (4 kN/min.) until failure occurred. The ending stage of the impact test allowed us to determine the ultimate load, ultimate deflection, ending dead load stiffness, and ending design load stiffness.

Tension splice joint loads

Static ramp load. -- For the tension static ramp load test, we applied a static ramp load of 4kN/min. to nine TSJs until failure occurred. Ten joints were originally assigned to this group, but one was inadvertently destroyed before the test began. The static ramp load group served as the control group to which all other TSJ groups were compared. We set the ramp load rate to cause failure in 5 to 10 minutes.

In the accelerated tension static ramp load tests, we applied a tensile ramp load that increased linearly at 40 kN/min to 10 TSJs until failure occurred. We set the ramp load rate to cause failure in approximately 30 to 60 seconds.

Impact load. -- The double-design tension impact load test for TSJs was similar to that for HJs, as explained earlier. The ultimate tension impact load test for TSJs was identical to the double-design impact load test, with two exceptions. First, the maximum load in the spike was the average strength (30.4 kN) found from the static ramp load tests. Second, since the 3-second duration was maintained, the load rate for the ultimate impact tests was also higher than the load rate for the double-design impact tests.

Statistical analysis

To test our hypothesis that wind and impact loads decrease the strength and change the stiffness of MPC joints, we compared the strength, ultimate deflection (defined as the deflection at the ultimate load), and stiffness values from the experimental groups to the corresponding property values in the control group. We compared the experimental group to the control group by means of a Student's t-test with a 95 percent confidence interval and assumption of unequal variances.

Determination of joint stiffness

We determined TSJ stiffness for each joint using a secant stiffness approach, defined as the axial load divided by the corresponding axial deflection (TPI 1995). HJ stiffness utilized load in the top chord and the corresponding slip deflection. Because the load deflection plot is nonlinear, we measured the stiffness at two different load levels: 1) dead load; and 2) design load. Stiffness at each load level was determined twice, before and after dynamic load, resulting in four stiffnesses for each experimental group: 1) beginning dead load stiffness; 2) beginning design load stiffness; 3) ending dead load stiffness; and 4) ending design load stiffness (Fig. 5).

We calculated only three stiffnesses for the impact tests because in the beginning phase, the impact load was applied after reaching dead load. Therefore, the beginning design load stiffness was not determined for all impact load tests. Dead load level was found by analyzing the Fink truss loaded with 290 N/m on the top and bottom chords (plus member self weight) using a finite element model (Computers and Structures, Inc. 1991). The design load level used to determine the design load stiffness was the average ultimate load from the static load group divided by 3.0 (TPI 1995).

We compared multiple stiffness measurements to determine if there were any differences between the static load group and the impact load group before and after the dynamic load. If there are no differences before the dynamic load, comparisons of static load and impact load ending stiffnesses can help determine if the dynamic load affected the joint stiffness.

Results and discussion

Heel joints

Redlinger (1998) provides all the properties of HJs calculated and/or measured for this study. Table 3 summarizes strength, stiffness, and deflection results.

We identified no dominant failure mode for HJs. Wood shear failure occurred 11 times, tooth withdrawal 15 times, and a combination of tooth with-drawal and wood shear 19 times. For one sample, both metal plates were sheared into two pieces along the interface between the two chords. Combinations involving plate failure occurred four times. Gupta and Gebremedhin (1990) observed similar failure modes. Load deflection plots for HJs exhibited nonlinear behavior, which is typical of wood members and wood joints (Gupta and Gebremedhin 1990).

Static ramp load. -- The average strength for compressive static ramp load tests was 26.5 kN coefficient of variation (COV) = 9 percent, which is comparable to results found by other researchers (Gupta and Gebremedhin 1990, Freilinger et al. 1997, Kent et al. 1997). No data are available in the literature with which to compare the tension static ramp load strength (32 kN).

There is convincing evidence, however, of differences between tension and compression strength and ultimate deflection values (two-sided p-values of 0.002 and 0.004, respectively). HJs tested in tension have average strength 21 percent higher and average ultimate deflection 33 percent lower than HJs tested in compression. The higher friction forces between the chords and the wood crushing that occurred in the tension tests might have resulted in higher strength. Lower ultimate deflection values for tension tests could also be attributed to the high friction forces.

Evidence is not sufficient to allow us to conclude that tension stiffnesses differ from compression stiffnesses (two-sided p-values 0.93 and 0.88 for dead load stiffness and design load stiffness, respectively). Differences might exist between the stiffnesses, but because of the high COV, no differences can be determined. The COV for stiffness is high, in part because the amount of accidental eccentricity varies from joint to joint. Vatovec et al. (1996) also found a higher COV for stiffness than for strength. In addition, the COV for dead load stiffness is much higher than the COV for design load stiffness. There is an initial lack of deflection, which has a greater effect on dead load stiffness, because it is measured at a lower load and thus at lower deflection readings. The lower deflection readings are affected more by LVDT noise, which increases the COV.

Hurricane wind compression load. -- Neither beginning stiffnesses nor ending stiffnesses of the experimental group differed from those of the control group. The lack of difference was probably related to the relatively low amplitude dynamic load used in the compression tests. The average load in the wind stage was approximately 8.1 kN [+ or -] 2.5 kN.

There is moderate evidence of a difference in strength between the wind simulation load group and the control group, but the cause of this difference in strength is not clear.

Hurricane wind tension load. -- There is suggestive, but inconclusive, evidence of a difference in strength between the control and experimental groups (one-sided p-value of 0.06). Because the static strength is lower than the strength following the wind loads and some evidence of a difference exists, we could conclude that dynamic wind loading increases the strength of MPC HJs. Because only suggestive evidence of a difference exists and the apparent increase is rather small (8.4%), chance is assumed to have caused the difference between the two groups.

Results of the t-test indicate that the beginning dead load and design load stiffnesses did not differ between the experimental and control groups (two-sided p-values of 0.99 and 0.08, respectively). Therefore, we concluded that the joints came from the same population. Following application of the dynamic load, the difference between the two groups was very large (two-sided p-values of 4.E-6 and 4.E-5 for dead load stiffness and design load stiffness, respectively). Increases were approximately 390 percent for dead load stiffness and approximately 428 percent for design load stiffness. Dynamic deflection for wind load was approximately three times greater in the tension group than in the compression group. The higher dynamic deflection in the tension case densified the wood fiber near the teeth enough to increase the stiffness.

The increase in stiffness after application of the dynamic load can be explained if we closely examine the deflection of the joint. The slip deflection of the joint has three causes: 1) the metal plate connecting the two chords deflects in shear; 2) the teeth of the metal plate bend and withdraw just prior to failure; and 3) wood fibers in direct contact with the teeth are crushed. The third source of deflection is the best explanation for the increase in stiffness after application of the dynamic load. As the dynamic load progressed, the wood fibers close to the teeth were plastically crushed and densified. After the dynamic load ended and the load was removed, the wood fibers rebounded only slightly. As the joint was reloaded, the densified wood fibers did not deflect as much as they did initially.

Another possible explanation might be that when an HJ is loaded, the metal teeth are less inclined to bend, because the force and the flat side of the teeth are not aligned. The teeth must withdraw farther for a given deflection, thus they are more secure in the wood holes. Also, in an HJ, the angle between the teeth and the applied force results in distribution of the applied force over a smaller area than in a TSJ. The resulting higher stress on the wood fibers causes more wood densification. The combination of wood densification and higher plate tooth security makes the joint stiffer after application of the dynamic load.

Double-design tension impact load. -- Beginning dead load stiffness did not differ between the impact group and the control group (two-sided p-value of 0.67). After we applied the spike, however, ending dead load and design load stiffnesses for the impact group increased dramatically compared with that of the control group (two-sided p-values of 0.01 and 0.001 for ending dead load and design load stiffnesses, respectively). The ending dead load stiffness p-value of 0.01 shows only moderate evidence of a difference (because of large COVs), but the ending design load stiffness p-value of 0.001 shows convincing evidence of a difference.

Not only is the evidence compelling, but the difference between the group averages is very significant. Ending dead load and design load stiffnesses increased by 250 and 254 percent, respectively. This stiffness change is similar to the stiffness increase caused by the tension wind load (where the stiffnesses increased an average of 409%). Therefore, it is possible that if the spike load were increased to a load higher than double the design load, the stiffness change would probably be even closer to the stiffness change seen in the tension wind simulation. This suggests that a low-amplitude dynamic load of long duration (e.g., a hurricane) could be simulated with a large-amplitude dynamic load of very short duration (e.g., impact spike). The magnitude of the force, however, would have to be much greater in the short duration dynamic load than in the long duration dynamic load.

More research is needed to determine what the magnitude of that load might be. The maximum force in the tension wind load simulation was only 17.6 kN, whereas the double-design impact load maximum force was 21.4 kN.

We observed no differences in strength or ultimate deflection between the control group and impact load group (one-sided p-values of 0.32 and 0.27, respectively). Nor did we see any differences in strength or ultimate deflection between the control group and the tension load wind simulation group.

Tension splice joints

Table 4 summarizes all TSJ tests. Failure modes were: wood shear (11 times), tooth withdrawal (7 times), and plate failure (twice). Twenty joints experienced a combination of these failure modes. The load deflection curve was nonlinear, which is typical of wood joints (Gupta and Gebremedhin 1990). In TSJ, both the metal plate and the wood fibers typically affect the shape of the load deflection plot. When the load is low, the wood fibers in contact with the metal teeth deform elastically and the plate elastically deflects in tension. Increased load causes the wood nearest to the metal teeth to crush and deform. If the yield strength of the metal plate is exceeded, a plate failure mode generally follows plastic deformation of the plate. If a weak shear plane in the wood exists near the connection, a wood shear failure generally occurs. Tooth withdrawal failure is probably caused by crushing of the wood fibers in contact with the metal teeth and bending of the metal teeth. Loading a TSJ induces bending moment about the weakest axis in the metal teeth, because the force direction and the flat side of the teeth are aligned; the teeth therefore bend considerably.

[FIGURE 6 OMITTED]

Tension static ramp load. -- The average strength of TSJs found here (Table 4) is comparable to the results of previous researchers (Kent et al. 1997).

Accelerated ramp load. -- No properties differed between the static ramp load group and the accelerated ramp load group (two-sided p-values of 0.20, 0.20, 0.51, and 0.80 for dead load stiffness, design load stiffness, strength, and ultimate deflection, respectively). Although the accelerated ramp load rate was approximately 10 times higher than the static ramp load rate, the accelerated rate still took approximately 45 seconds to cause a joint to fail. The accelerated load rate used in this test produced static results. This is similar to the results obtained by lumber. Therefore, it is recommended that TSJ can be tested in 1-minute to failure instead of 10-minute to failure as recommended by the current standards.

Double-design impact load. -- The beginning dead load stiffness for this group is the same as that of the static group (p-value = 0.36). But the ending dead load stiffness differed significantly (p-value = 0.007) from the static group. The ending dead load stiffness dropped by approximately 33 percent after the double-design impact. Because of the low p-value (two-sided p-value of 0.007), it is unlikely that the stiffness decrease is due completely to chance. The ending design load stiffness was not significantly different from the control group (p-value = 0.25). Therefore, the double-design impact affected the dead load stiffness, but not the design load stiffness.

One explanation for this joint behavior is the relative magnitude of the spike and the design load. For the double-design impact group, the maximum load during the spike was approximately 20.2 kN, which was twice as much as the design load, but almost five times the dead load. A larger spike magnitude would likely cause a decrease in design load stiffness also. The double-design impact load had little effect on the strength and ultimate deflection (one-sided p-values of 0.16 and 0.47, respectively).

The load deflection curve of a TSJ after being subjected to the double-design impact load (Fig. 6) was linear until the maximum spike load was exceeded. After the maximum spike load, the joint behaved approximately like a joint tested only with a ramp load.

For joints subjected to the double-design impact load, stiffness (ending dead load and ending design load) increased for HJs but decreased (ending dead load stiffness) for TSJs, in part because of the different amounts of dynamic deflection that occurred during the impact loads.

Ultimate impact load results. -- The data did not show sufficient evidence to support the conclusion that the ultimate impact load group and the static ramp load group differ in ultimate deflection and beginning dead load stiffness (one-sided p-value of 0.17 and two-sided p-value of 0.11, respectively). The end dead load stiffness, however, differed considerably (two-sided p-value of 0.004). The dead load stiffness for the ultimate impact load group dropped by approximately 49 percent. The data also show that the ultimate impact load group and the static load group differed moderately in ending design load stiffness (two-sided p-value of 0.009), which was about 22 percent lower in the ultimate impact load group.

The load deflection plots for TSJs reveal several similarities among the double-design impact group (Fig. 6) and the ultimate impact group (Fig. 7). The beginning portions of the plots are almost identical. The general shape of the load deflection plot during the load spike is the same for the double-design impact group and the ultimate impact group. For both groups, the spike caused a nonlinear curve with positive change in slope during the increasing portion of the spike, inelastic deformation when the spike was at the maximum, and a nonlinear curve with negative change in slope during the decreasing portion of the spike.

[FIGURE 7 OMITTED]

Both groups show that the joints behaved in a much more linear fashion after application of the spike load. The ending portion of the load deflection curve is very straight and does not change slope appreciably until the load exceeds the spike maximum, after which point both groups show a large decrease in stiffness.

Dynamic deflection

Evidence of a difference in dynamic deflection (for double-design impact load) between the HJs and TSJs was very significant (two-sided p-value of 2.E-06). We suspect that in HJs, the large amount of dynamic deflection caused wood fiber densification near the metal teeth, which resulted in increased stiffness. After we applied the dynamic load in HJ tests, we saw small gaps between the metal teeth and the wood. Based on this observation, we assumed that wood densification had occurred on the opposite side of the teeth. For TSJs, however, the dynamic deflection was only enough to slightly enlarge the holes in which the teeth sit, which caused a decrease in stiffness. This difference in stiffness change between HJs and TSJs subjected to the same dynamic load probably is related to the fundamental differences in joint geometry and loading.

Dynamic deflection is a measure of the permanent deflection caused by the spike load. Large dynamic deflections suggest that the spike load had a large effect on the joint, whereas the absence of dynamic deflection implies that the spike load had little or no effect. We saw evidence of a difference in the dynamic deflection values between the double-design impact group and the ultimate impact group (two-sided p-value of 3.E-5).

We expected larger dynamic deflection for the ultimate impact group than for the double-design impact group because the spike maximum was 50 percent larger for the ultimate impact test than for the double-design impact test. The dynamic deflection, however, was approximately 360 percent larger for the ultimate impact group than for the double-design impact group. For this group, a 50 percent increase in load produced a 360 percent increase in permanent deflection, because joints subjected to the ultimate impact experienced large inelastic deformations while the load was at the spike maximum. For the double-design impact maximum, some joints had very small inelastic deformations, while others did not yield at all.

Conclusions

Static ramp load tests revealed that HJs tested with the top chord in tension had average strength 17 percent higher than HJs tested with the top chord in compression. We observed no significant difference in stiffness between HJs tested in compression and those tested in tension. Stiffness of HJs subjected to the hurricane wind tension load increased by 325 percent as a result of densification of the wood closest to the metal teeth. The same load caused no significant change in HJ strength or ultimate deflection. HJs subjected to the hurricane wind compression load showed no change in stiffness or strength. HJs subjected to the double-design impact load behaved similarly to joints subjected to the wind tension load.

TSJs subjected to the accelerated ramp load produced the same results (strength and stiffness) as joints subjected to the static ramp load. Using the accelerated ramp load as a control group for MPC TSJs would produce "static" results in one-tenth the time. Therefore, TSJ can be tested in 1-minute to failure instead of 10-minute to failure as recommended by the current standards. Neither the double-design impact load nor the ultimate impact load caused strength or ultimate deflection changes in TSJs.

Due to the small sample size and large coefficient of variation, more testing is needed before more conclusions could be drawn that are related to duration of load for MPC joints.
Table 1. -- Sample size and test types for MPC joints.

 Test Sample size

Heel joints
 Static ramp load
 Compression static ramp load (control group)
 (top chord in compression) 10
 Tension static ramp load (control group)
 (top chord in tension) 10

Hurricane wind load test
 Compression hurricane wind load 10
 Tension hurricane wind load 10

Impact load test
 Double-design tension impact load 10

Tension splice joints
 Static ramp load test
 Tension static ramp load (control group) 9
 Accelerated tension static ramp load 10

Impact load test
 Double-design tension impact load 10
 Ultimate tension impact load 10

Table 2. -- Controlling forces in heel joints.

 Load case Joint side Force Top chord in
 (kN)

Main wind force resisting
 system with positive
 internal pressure and wind
 normal to ridge Windward 16.30 Tension
Components and cladding
 case with negative
 internal pressure Windward 9.46 Compression

Table 3. -- Heel joint test results. (a)

 Average Ultimate Dynamic
Test description strength deflection deflection
 (kN) (mm)

Static ramp load
 (control group)
 Compression 26.5 6.0 --
 (10, 9) (9, 19) --
 Tension 32.0 4.0
 (10, 12) (10, 28)

Hurricane wind load
 (experimental group)
 Compression 28.6 7.0 0.05
 (10, 5) (10, 16) (10, 48)
 Tension 34.7 4.0 0.15
 (10, 11) (9, 20) (10, 39)

Double-design impact load
 (experimental group)
 Tension 31.2 4.0 0.64
 (10, 10) (9, 18) (10, 24)

 Stiffness
 @ dead load @ design load
Test description Begin End Begin End
 (X [10.sup.5] N/mm)

Static ramp load
 (control group)
 Compression 0.60 0.60 (b) 0.57 0.57 (b)
 (9, 51) (9, 51) (9, 40) (9, 40)
 Tension 0.58 0.58 (b) 0.39 0.39 (b)
 (8, 63) (8, 63) (8, 28) (8, 28)

Hurricane wind load
 (experimental group)
 Compression 0.49 0.45 0.47 0.45
 (10, 20) (10,10) (10,25) (10,16)
 Tension 0.58 2.26 0.56 1.67
 (10, 35) (10, 28) (10, 44) (10, 34)

Double-design impact load
 (experimental group)
 Tension 0.51 1.45 -- 0.99
 (10, 56) (10, 55) (10,42)

(a) Numbers in parentheses are numbers of observations and coefficients
of variation in percent.

Table 4. -- Tension splice joint test results.

 Average Ultimate Dynamic
Test description strength deflection deflection
 (kN) (mm)

Static ramp load
 Tension 30.4 2.0 --
 (9, 15) (9, 5)
 Accelerated tension 32.0 2.0 --
 (10, 18) (10, 32)

Impact load
 Double-design 32.2 2.0 0.18
 (10, 8) (10, 21) (10, 35)
 Ultimate load 33.5 2.0 0.81
 (7, 6) (7, 31) (7, 20)

 Stiffness
 @ dead load @ design load
Test description Begin End Begin End
 (X [10.sup.5] N/mm)

Static ramp load
 Tension 1.14 -- 0.77
 (9, 27) (9, 22)
 Accelerated tension 1.31 -- 0.86
 (10, 20) (10, 17)

Impact load
 Double-design 1.26 0.76 -- 0.68
 (10, 21) (10, 19) (10, 19)
 Ultimate load 1.52 0.77 -- 0.60
 (7, 47) (7, 15) (7, 5)

(a) Numbers in parentheses are numbers of observations and coefficients
of variation in percent.


[c]Forest Products Society 2004.

Forest Prod. J. 54(3):76-84.

Literature cited

American Forest and Paper Association (AF & PA). 1997. National Design Specification for Wood Construction. AF & PA, Washington, DC.

Dagher, H.J., V. Cacese, and Y. Hsu. 1991. Feasibility of metal connector plates in timber bridges: Fatigue study. Tech. Rept. Dept. of Civil Engineering, Univ. of Maine, Orono, ME.

Emerson, R. and K.J. Fridley. 1996. Resistance of metal-plate-connected truss joints to dynamic loading. Forest Prod. J. 46(5):83-90.

Freilinger, S., R. Gupta, and T.H. Miller. 1997. Cyclic performance of wood truss joints. Proc: Structures Congress XV, Am. Society of Civil Engineers, Reston, VA. pp. 939-943.

Gupta, R. and K.G. Gebremedhin. 1990. Destructive testing of metal-plate-connected wood truss joints. J. of Structural Engineering 116(7):1971-1982.

__________. M. Vatovec, and T.H. Miller. 1996. Metal-plate-connected wood joints: A literature review. Res. Contribution 13. Forest Res. Lab., Oregon State Univ., Corvallis, OR.

Kent, S.M., R. Gupta, and T.H. Miller. 1997. Dynamic behavior of metal-plate-connected wood truss joints. J. of Structural Engineering 123(8):1037-1045.

Redlinger, M.J. 1998. Behavior of metal-plate-connected wood truss joints under wind and impact loads. MS thesis. Oregon State Univ., Corvallis, OR.

Computers and Structures, Inc. 1991. SAP90: A series of computer programs for the finite element analysis of structures--user's manual. Version 5.31. Computers and Structures, Inc., Berkeley, CA.

Truss Plate Institute (TPI). 1995. National design standard for metal plate connected wood truss construction. TPI 1-1995. TPI, Madison, WI.

U.S. Army Corps of Engineers. 1991. Hurricane Bob Wind Data. (unpublished data). Field Research Facility, Kitty Hawk, NC.

Vatovec, M., R. Gupta, and T.H. Miller. 1996. Testing and evaluation of metal-plate-connected wood truss joints. J. Test. Eval. 24(2):63-72.

Rakesh Gupta*

Thomas H. Miller

Mark J. Redlinger

The authors are, respectively, Associate Professor, Dept. of Wood Sci. and Engineering, Oregon State Univ., Corvallis, OR 97331; Associate Professor, Dept of Civil, Construction, and Environmental Engineering, Oregon State Univ.; and Designer, Degenkolb Engineers, 620 Fifth Ave., Portland, OR 97204. Funding for this research was provided by the USDA National Research Initiative, Competitive Grants Program. We also wish to acknowledge Alpine Engineered Products, Inc., for donating metal connector plates; Frank Lumber Co. for donating lumber; and Milo Clauson, Research Associate, for his help and advice with the instrumentation and testing. This is paper 3462 of the Forest Res. Lab., Oregon State Univ. This paper was received for publication in October 2002. Article No. 9574.

*Forest Products Society Member.
COPYRIGHT 2004 Forest Products Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2004 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Gupta, Rakesh; Miller, Thomas H.; Redlinger, Mark J.
Publication:Forest Products Journal
Geographic Code:1USA
Date:Mar 1, 2004
Words:6012
Previous Article:The effects of species, adhesive type, and cure temperature on the strength and durability of a structural finger-joint.
Next Article:Biocide retention variation of southern yellow pine products treated with waterborne preservatives in commercial, pilot plant, or laboratory...
Topics:

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |