U. S. FOREST SERVICE RESEARCH PAPER FPL 67 NOVEMBER 1966

MOISTURE

RAFTER JOINTS

CYCLING OF TRUSSED

U. S. DEPARTMENT OF AGRICULTURE FOREST SERVICE FOREST PRODUCTS LABORATORY MADISON WISCONSIN

FOREST PRODUCTS LABORATORY U.S. DEPARTMENT OF AGRICULTURE FOREST SERVICE ··· MADISON, WIS.

In Cooperation with the University of Wisconsin

SUMMARY

Prefabricated wood trussed rafters are widely used in small building con­struction, and various connector systems are used in preassembling the wood members, including nailed o r glued wood and plywood plates and newly developed metal gusset plates of various designs. Trussed rafters installed in a building are subjected to many conditions that could affect the load-carrying capacity of the joints. To illustrate, changes in relative humidity and temperature could cause dimensional changes in truss members and this could cause a loosening of the connector; cyclic live-loading, resulting from changes in the amount of snow wind pressure on the roof, may cause the joints to “work” and thus alter their performance; or a constant load, such as a dead load, could cause creep. It was considered to determine actual effect of these con­ditions, by means of accelerated tests, so that the information could be made readily available to aid in the design of wood trussed rafters.

This Research Paper presents the results of the first phase of a longtime performance study of trussed rafter joints, namely, the effects of initial mois­ture content and moisture cycling under load on strength and rigidity of trussed rafter joints. Included in the investigation were nailed wood and plywood joints; phenol-resorcinol and casein glued joints; and nailed. barbed, and toothed metal-plate joints. Joints tested in tension and bending were fabricated at 10, 17, and 25 percent moisture contents and half of the specimens were subjected to moisture cycling while under load. These specimens were then conditioned to 10 percent moisture content, destructively loaded, and the results compared with those of matched, uncycled control specimens.

The results of this study showed that the moisture-cycled specimens had from I to 3-1/2 times more elongation o r deflection than the control specimens. Losses in maximum load for the cycled specimens ranged from 0 to 30 percent greater than for the controls. The initial moisture content had a significant effect upon the elongation and deflection of the mechanically fastened joints but not on the maximum load, with the exception of the barbed metal-plate joints. Wet material had an adverse effect upon the maximum loads of the glued joints.

The results of this research provide information and data essential to design of trussed rafter joints to insure their satisfactory longtime performance under varying conditions of loading, moisture content, and moisture cycling.

i

CONTENTS

Page

INTRODUCTION . . . . . . . . . . . . . . . . . 1 REVIEW OF PAST RESEARCH . . . . . . . . . 2 OBJECTIVES . . . . . . . . . . . . . . . . . . . 2 SPECIMENS . . . . . . . . . . . . . . . . . . 2 MOISTURE CONTENT OF MATERIAL . . . . . 5 EXPERIMENTAL PROCEDURES . . . . . . . . 5 RESULTS AND DISCUSSION . . . . . . . . . . . 9

Moisture Cycling . . . . . . . . . . . . . . . . 9 Destructive Loading . . . . . . . . . . . . . . 16

Nailed-wood joints . . . . . . . . . . . . . . 18 Nailed-plywood joints . . . . . . . . . . . . 22 Nailed metal-plate joints . . . . . . . . . . 24 Barbed metal-plate joints . . . . . . . . . . 25 Toothed metal-plate joints . . . . . . . . . 27 Phenol-resorcinol glued joints . . . . . . . 29 Casein-glued joints . . . . . . . . . . . . . 31

STATISTICAL ANALYSIS . . . . . . . . . . . . 33 “t” Values . . . . . . . . . . . . . . . . . . . 33 “F” Values . . . . . . . . . . . . . . . . . . . 35

SUMMARY OF FINDINGS . . . . . . . . . . . . 37 LITERATURE CITED . . . . . . . . . . . . . . 38

FPL 67 ii

MOISTURE CYCLING OF

TRUSSED RAFTER JOINTS

by THOMAS LEE WILKINSON, Engineer

Forest Products Laboratory1

Forest Service U. S. Department of Agriculture

INTRODUCTION

The preassembled wood trussed rafter, a a decade ago, has become widely accepted

and used in the small building industry. An esti­mated one-quarter billion dollars is spent annually in the United States on trussed rafters.

Trussed rafters are made in many shapes and sizes, the most popular types being the king-post and Fink or W-truss. Many different connection systems are used to assemble the wood members, including nailed wood and plywood plates, glued plates, and the newly developed metal gusset plates, available in various configurations. These connectors resist axial forces, shear forces, and moments in varying degrees. Newly developed design procedures consider all of these forces to secure a balanced design. Many full-size trussed rafters have been tested soon after fabrication and have been found to be adequately designed.

Trussed rafters installed in a building may be subjected to many conditions that could affect the load-carrying capacity of the joints. For example, changes in relative humidity and temperature could cause dimensional changes in the truss members or connectors resulting in a of the connectors. This same condition could result from in moisture content of the truss members as they come into equilibrium with their surroundings. Also, cyclic live-loading, resulting from changes in the amount of snow or wind pressure on the roof, may cause the joints to “work” and thus alter their performance; or a load, such as a dead load, could cause creep. of these could affect the longtime service performance of wood trussed rafters.

Prior to this particular research study, little

1Maintained at Madison, Wis., in with the University of

was known concerning the permanence long­time service performance of the newly developed metal gusset plates or glued joints under load. Available information the longtime performance of nailed joints was obtained mainly from service records. Yet, a large number of trussed rafters being fabricated and used without knowing how well they will continue to perform during the expected life of the buildings in which they are installed. If the trussed rafters now in service give satisfactory per­formance, it will prove that their design was adequate to meet in-service conditions. but this will take 20 to 30 years to determine. It was con­sidered important, therefore, that the effect of these conditions be investigated by means of accelerated tests, so that the results could be made readily available to aid in the design of trussed rafters.

This Research Paper presents the results of the first phase of a longtime performance study of trussed rafter joints, namely, the effects of initial moisture content and moisture cycling under load on the strength and rigidity of trussed rafter joints.

REVIEW OF PAST RESEARCH

studies have been conducted by other researchers on full-size trussed rafters. In practically all of the studies. however, the tests were conducted shortly after fabrication of the rafters. These included by: Angleton (1)2

on roof trusses using nailed-plywood gusset plates; Luxford and Heyer (3) on glued and nailed roof trusses; Pneuman (5) on king-post type trussed rafters with plywood gussets on one aide; and Radcliffe, and others (6,7,8) on nail-glued type trussed rafters. In addition. various manu­facturers have tested trussed rafters using the new metal-gusset plates of the more popular con­figurations and constructions. In all of these studies, trussed rafters were found to be adequately designed.

Luxford (2) studied the results of changes in relative humidity on strength and rigidity of trussed rafters with glued joints and also with

nailed joints. In this study, small trussed rafters were subjected to high and low humidities in a controlled atmosphere. Results showed that the glued trussed rafters suffered some loss in stiff­ness and considerable loss in strength, while the nailed trusses were less affected. After exposure, both types had ample stiffness and strength for normal service requirements, but the glued trussed rafters were considerably stiffer and stronger than the nailed ones. Other at the Forest Products Laboratory showed little loss in strength and of some glued-nailed king-post and W-type trusses after 3 years of outdoor exposure. The trusses, however, were not loaded during exposure.

The foregoing studies have proved practica­bility of trussed rafters, but their longtime per­formance has not been determined.

OBJECTIVES

Because of the broad scope of the longtime performance study of truss joints, the research to be conducted by the Forest Products Laboratory

into six phases,has been namely (1) mois­ture cycling, (2) temperature effects, (3) creep at constant load, (4) cyclic loading, (5) geometry effects, and (6) exposure of full-size trusses.

Phase I of this study has been completed and the results are presented in this Research Paper. The specific objectives of Phase I were to deter­mine the effect of moisture cycling under load and the effect of initial moisture content of the truss material on the final strength rigidity of nailed wood and plywood joints, three types of metal-plate connector joints, and phenol­resorcinol- and casein-glued gusset joints.

SPECIMENS

Two types of specimens were chosen for this study. One was a simple tension specimen and the other a simple bending specimen. These were selected mainly because of their simplicity and

2Underlined numbers in parentheses refer to Literature Cited at the end of this Paper.

67 2

Figure 1.--Method of fabricating specimens for cycling and control. A, general configuration; B, selection of joint members to obtain matched specimens.

M 127 609

the ease with which they could be tested. They also facilitated the application of forces to the connectors during testing that were similar to those found in an actual trussed rafter.

Both types of specimens had the same general configuration (fig. 1A). They were made of straight-grained Douglas-fir 2 4’s were free of defects in the area of the gusset plates. The specimens were 3 feet long with a joint at the midlength. The halves butted together so that no gap was present,

Three general groups of fastenings were eval­uated in this study, namely, nailed joints, metal-plate joints, and glued gusset joints.

The nailed joints included plywood and solid-wood gusset plates. The gussets were 20 inches long and fastened sixpenny common nails. The nail pattern is shown in figure 2. Clear straight-grained Douglas-fir 1 by 4’s were used for the wood gussets, and 1/2-inch Exterior grade Douglas-fir plywood was used for the ply­wood gussets.

Although metal-plate connectors are available in numerous configurations and constructions, they were divided into three basic groups and one representative type selected from each group for evaluation in this study. The three basic groups were: (1) nailed metal plates, which rely on the nails alone to carry the loads transmitted through the joint; (2) barbed metal plates, which rely on the barbs and also some nails to carry load;

and (3) toothed metal plates, which depend on teeth punched in the plate to carry the load with­out any nails. The plates selected for this study from each of the three groups are shown in figure 3, All plate connectors selected were of galvanized sheet metal--the nailed and barbed types being 20-gage material and the toothed type gage. Joints with toothed barbed metal-plate connectors were fabricated with the aid of a hydraulic testing machine, as shown in figure 4. Toothed plates were pressed one face at a time and the barbed plates were first positioned on each face and then pressed.

For the glued joints, a phenol-resorcinol and a casein glue were selected as being the more popular glues in the manufacture of trussed rafters. The gussets were made of 1/2-inch Exterior grade Douglas-fir plywood and were

inches in length. Pressure was applied to the glued joints in a glue press until the glue had set, so that nails were not needed. Gluing was done at room temperature and a period of 2 to 3 weeks allowed for curing after the removal of pressure.

A control specimen was matched with each cycled specimen. This was done by taking every other piece as one-half of a specimen (fig. 1B). The term “side,” as shown in figure 1, refers to one of the wood members used in a specimen. No attempt was made to match speci­mens of different groups.

Specimens simulated a tension splice in the

3

2.--Face edge views of naiIed-wood (top) and nailed-plywood (bottom) joints. Mem-bers and wood gussets are of Douglas-fir. The plywood gussets are of 1/2-inch Exterior grade. Common sixpenny nails were used. M 126 001

Figure 3.--Three general types of metal truss plates selected for the moisture cycling study. Left, barbed metal plate (20-gage); center, toothed plate (18-gage); and right, nailed metaI pIate (20-gage).

M 124 773

67 4

Figure 4.--Method of fabricating toothed and barbed metal-plate joints. Wood members were held in position by a jig whi le the hydraulic testing machine pressed in t h e plate. Toothed plates were pressed one face at a t i m e . as shown. Barbed plates were first positioned on each face and then pressed.

lower chord of a trussed rafter and were designed for a 2,000-pound tensile load of 2-months’ duration. This value represented a typical design stress and the duration was representative of the time the specimens were loaded. Since the lower chord splice usually carries bending moment as well as direct stress, the same joint. con­figuration was used in bending as in the tension specimens,

MOISTURE CONTENT OF MATERIAL

Three general moisture content groups were established to simulate the range in moisture content of material from which a trussed rafter may be made, namely. (1) wet (23 to 27 percent); (2) air-dry (15 to 19 percent); (3) kiln-dry (6 to percent). Specimens of the seven types of joints studied were made from each of these three initial moisture content groups. The mois­ture content of the gusset material was approxi­mately 11 percent for all specimens using wood or plywood.

M 129 928

EXPERIMENTAL PROCEDURES

Twelve bending and tension specimens at each initial moisture content (wet, airdry, and kiln-dry) were evaluated for each of the seven joint types. Six specimens of each sample were subjected to moisture cycling and six were used as uncycled controls.

The moisture cycling consisted of three cycles of high and low relative humidity. The first phase of each cycle was at 90° F. and 95 percent rela­tive humidity. This gave the wood an equilibrium moisture content of about percent. The second phase of each cycle was at 160° F. and 45 per­cent relative humidity, which gave an equilibrium moisture content of about 6 percent. Each phase lasted for 1 week, At the end of the three cycles, the specimens were conditioned at 160° F. and 73 percent relative humidity for 3 to 4 days. which resulted in a final moisture content of about 10 percent., After conditioning they were stored at 74° F. and 50 percent relative humidity.

During this entire time, the uncycled control specimens were stored at 74° F. and 50 percent relative humidity to attain an equilibrium mois­ture content of approximately 10 percent.

5

Figure 5.--General arrangement used for loading bending specimens during moisture cycling. Total load on each specimen was 500 pounds, which produced a bending moment on the joint similar to that caused by a 10-pound-per-lineal-footceiling load.

A total dead load of pounds was applied to all moisture-cycled specimens during the entire time of cycling and to all uncycled control speci­mens during the time of storage. Bending speci­mens were loaded at the quarter points as shown in figure 5, to produce a moment at the joint equivalent to that caused by a uniform ceiling load of 10 pounds per lineal foot. Tension speci­mens were loaded as shown in figure 6, which resulted in a stress on the joint equivalent to that caused by a dead load of pounds per lineal foot on a 28-foot-span trussed rafter.

To obtain some idea of the moisture content the cycled specimens reached at the end of each phase of cycling, a moisture determination was made on a piece of nominal 2 by 4 lumber which had been placed in the cycled atmospheric con­ditions and was. representative of the material used for the specimens. It was end-coated to prevent moisture penetration from the ends and, after each phase of cycling, a moisture sample

taken from it. The sample was divided into

M 125 941

outer, intermediate, and inner sections, with the outer and intermediate sections each 1/4 inch thick, and the moisture content of each section was determined. This gave the gradient of mois­ture penetration as well as the overall moisture content.

At the end of each phase of cycling, deflection elongation readings were made on the cycled

specimens. This was done to determine the effect of moisture cycling upon the rate and amount of creep. Figure 7 shows the method used to meas­ure the deflection of the bending specimens and figure 8 shows the method used. to measure elon­gation of the tension specimens. For comparison, similar readings were taken of the uncycled con­trol specimens at the same time intervals.

After the specimens were conditioned to 10 per­cent moisture content, elongation and deflection readings were made. The load was then removed and the immediate recovery recorded. Next, all specimens were loaded to destruction at room temperature. Figure 9 shows the general arrange-

FPL 67 6

Figure 6.--General arrangement used for tension specimens during moisture cycling. The load on the joint was 500 pounds, which is a force similar to that produced by a dead load of 20 pounds per lineal foot on a trussed rafter.

M 124 776

Figure 7.--Method used to measure deflection of bending specimens at end of each phase of moisture cycling. M 126 003

7

Figure 8.--Method used to measure elongation Figure 9.--Destructive loading arrangement of tension specimens at the end of each for tension specimens. Shear plates were phase of moisture cycling. Also shown a r e used to apply load to the specimen and C-t h e round shear plates used in applying clamps were used to close splits and final destructive load to the specimen. increase the capacity of shear plates.

Elongation was measured over a 6-Inch gage M 126 004 length.

M 124 777

FPL 67 8

Figure 10.--Destructive loading arrangement for bonding specimens. The span of the join? was 33 inches and loads were applied at the quarter-mints.

ment used for destructive loading of the tension specimens. Shear plates, 2-1/2 inches in diam­eter, were used to apply load to the specimens and were located 5 inches from the ends of the members. This gave a maximum capacity of approximately 12,000 pounds. When the shear plates were placed in wet material and the mate­rial allowed to dry, there was a tendency to split

members; therefore, to obtain loads large enough to break the joints, C-clamps between the shear plates and ends of the members to close the splits. Load was applied at a con­stant machine-head movement of 0.010 inch per minute and elongations were measured over a 6-inch gage length on both sides of the joints.

Figure 10 shows the destructive loading arrangement for the bending specimens. The specimens were loaded at the quarter-points and deflections were measured at the load points. The deflection equipment permitted readings beyond design-load levels, but was removed before maximum load was reached. Load was applied at a constant machine-head movement of 0.010 inch per minute.

The moisture content and specific gravity of

M 124 775

both halves of each specimen were determined immediately after the destructive loading of the joints.

RESULTS AND DISCUSSION

Moisture Cycling

The data obtained during cycling consisted of moisture determinations of sample material located with the cycled specimens and creep measurements at the end of each phase of cycling,

Table 1 gives values of moisture content attained at various sections of the specimens during each phase of moisture cycling as deter­mined from sample material. The sample pieces were divided into three sections, as described under “Experimental Procedures,” to obtain a moisture gradient. Values presented in the table are the averages for all seven types of joints at each of the three initial moisture contents for both tension and bending specimens. The. values obtained fox the whole section compare favorably

9

Table 1. --Moisture contents attained at various sections of t he specimens during each phase of moisture cycling as determined from sample mater ia l 1

with the range that could be encountered in trussed rafters located in buildings. The largest range of moisture contents was noted in the outer 1/4 inch of the material, but a large vari­ation was also obtained between phases of cycling: within the second 1/4 inch. This outer 1/2 inch is the portion of the material in which fastener was located and where the greatest variation in moisture was desired. The initial moisture con­tent of some of the “wet” sample pieces selected

higher than that desired for the specimens.

This was probably caused by incomplete condi­tioning of the sample material, as it was taken directly from a saturated condition and used after a short period of drying.

Table 2 gives the average elongation and deflection measurements of initially wet, air-dry, and kiln-dry cycled and control specimens of each type of joint evaluated in tension and bending. These data show a substantial increase in the amount of creep when the specimens were sub­jected to cycling. The cycled tension specimens

FPL 67 10

Table 2 .--Elongation and deflection2 measurements of initially wet, air-dry, kiln-dry cycled and control specimensof each type of joint evaluated in tension and bending3

had from 2.0 to 37.0 times greater elongation than the controls and the cycled bending speci­mens had 2.1 to 6.9 times greater deflection than the controls, The larger ratios were for the kiln-dry group in which the control specimens had less creep than the controls of the wet or air-dry groups. For the glued joints. there was less variation in the ratios of elongation deflection of cycled and control specimens for the different initial moisture contents than for the other joint types. For specimens with mechan­ical connectors, the ratios increased the initial moisture content decreased, primarily because of the smaller deformation of the controls.

Figures 11-14 show typical elongation and deflection curves of various types of joints during moisture cycling. At the end of condition­ing while under load, the control specimens had ceased creeping, while creep was still occurring in the cycled specimens. The largest increases in deformation generally occurred during the dry phases of cycling. During this phase the members shrank, leaving a slight gap between the members and gussets, thus possibly allowing the speci­mens to deform by removing friction. Another contributing factor could be the expansion of the metal gussets because of the increase in temper-

160° F.ature from 90°

11

Figure It.--Typical elongation curves for kiln-dry, toothed metal-plate joints in tension during moisture cycling. M 131 598

Figure 12.--Typical elongation curves for wet, nailed-wood joints in tension during moisture cycling. M 131 599

FPL 67 12

Figure 13.--Typical deflection curves for kiIn-dry, nailed metal-plate joints in bending during moisture cycling. was measured at the quarter-points.

M 131 600

Figure 14.--Typical deflection curves for wet, barbed metal-plate joints in bending during moisture cycling. Deflection was measured at the quarter-points.

M 601

13

Table 3.--Percent elongation and deflection recovery of initially wet, air-dry, and kiln-dry cycled and control specimens of each type of joint evaluated in tension and bending1

The percentage of elongation and deflection recovery at the end of moisture cycling is shown in table 3. The bending control specimens recov­ered approximately 60 percent of their total deflection, and the tension controls recovered approximately 35 percent of their total elongation, with values ranging from 0 to 100 percent, The cycled specimens recovered approximately 20 percent of their total deflection elongation. The larger amount of creep of the cycled speci­mens and the smaller percentage recovery resulted in a large residual set in the cycled specimens.

Deterioration of the cycled joints was noted at the end of moisture cycling. Members quite often were split or checked. Wood and plywood gussets had checks and were often discolored. Corrosion

of metal fasteners was common, with rust appear­ing on the nailed joints and oxidation of the zinc coating occurring on the metal gusset plates. The glued joints had a small amount of glue-bond failure, especially with those made of “wet” material. Partial failure of the plywood gussets,

of unequal shrinkage of the gussets and members, was noted. This occurred in both the cycled and control specimens. Gaps were often present between gussets and members of nailed joints. Shrinkage caused the barbed metal plates to buckle out of the wood, and were held only by the positioning nails along edge of the plate. A comparison of the general appearance of various cycled specimens and their controls is shown in

FPL 67 14

Figure 15.--Comparison of cycled specimens with controls for nailed-wood joints (A, B) and for nailed-plywood joints (C, D). Joints A and C were cycled and B and D were controls.

M 126 002

Figure 16.--Comparison of cycled and control specimens for na i led metal-plate joints. Lower spec i men was cycled .

M 128

15

Figure 17.--Comparison of cycled and control specimens for barbed metal-plate joints (A, B) and toothed metal-plate joints (C, D). Specimens A and C were cycled and B and D were controls. M 124 774

Figure 18.--Comparison of cycled specimen (top) with control (bottom) for glued joints

Destructive Loading

General.--After cycling was completed, the specimens were destructively loaded to obtain values for and maximum load. The rigidity of mechanical fasteners is important in determining design loads, and thus the rigidity values obtained were of primary interest for most

M 129 984

of the mechanical fasteners. For the glued joints and barbed metal plates, the maximum loads were of primary interest. Tables 4 and 5 summarize the results of the destructive loading in tension and bending. The values of specific gravity shown in these tables indicate that the method of matching control and cycled specimens was very good. Specific gravity

FPL 16

Table 4.--Summary of moisture content arid specific gravity values, elongation measurements, and maximum loads far the various moisture cycled and control specmens of each type of

subjected to destructive loading in tension1

values were determined for both halves of each specimen. The final moisture contents of the cycled and control specimens were so close that correction for the difference was unnecessary.

The joints were designed for tensile force and thus the specimens loaded in tension can be com­pared to each other in determining their ability to meet design requirements. In bending, however, the different connector types should not he corn-pared unless they have comparable gusset lengths.

The values of elongation for the tension speci­mens are given at a load of 2,000 pounds, which was the assumed design load for the joints. The

values of deflection for the bending specimens are given at a load of 1,000 pounds. All values of elongation and deflection due to destructive loading are in addition to the permanent set that resulted from dead loading and moisture cycling of the cycled specimens and the dead loading of the control specimens. The total distortion of any particular joint specimen is equal to the sum of the permanent set and elongation o r deflection values shown for that specimen. These values are given in table 6. In general, the moisture cycling caused permanent sets that were much larger than the deformations resulting from destructive load­

17

Table 5.--Summary of moisture content and specific gravity values, deflection measurements, and maximum loads for the various moisture cycled and control specimens of each type of joint to destructive loading in bending1

ing. The permanent set in the control specimens was generally less than the deformation resulting from destructive loading.

The results of destructive loading in tension indicate that the greatest effect of moisture cycling was on the rigidity of most of the mechan­ical fastener joints. For the glued joints and the barbed metal-plate joints, however. the maximum tensile loads were affected. For the bending specimens, moisture cycling caused similar

losses in both rigidity and maximum load. A detailed analysis of the results of destructive

loading on the individual types of joints studied is presented in the following portion of this Research Paper.

Nailed-wood joints.--Design values for mechanical connectors are based on the load at a given elongation, usually 0.015 inch, or a portion of the maximum load. The lower of these two values determines the allowable load. For nailed

FPL 67 18

Table 6.--Permanent set resulting from dead loading and moisture cycling, and additional elongation and deflection due to destructive loading of the various types of joints evaluated tension and bending

joints, the load at a given elongation generally governs the allowable load. The effect of moisture cycling upon the elongation and deflection of the nailed wood joints, therefore, was of primary concern.

The cycled tension joints had 1.06 to 2.00 times greater elongation than the controls, the smaller ratio being for the “wet” initial moisture content group and the larger ratio for the “kiln-dry” group. The kiln-dry specimens, however, were still more rigid than the higher moisture content groups. This greater percentage increase in elongation can be accounted for partially by the gap left between the gussets and members in the cycled specimens. No gap was present in the kiln-dry control specimens, while for the other con­trols a gap was generally present due to shrinkage as the specimens dried to 10 percent moisture content. These gaps caused the control specimens in the higher moisture content groups

to be two to three times less rigid than the kiln-dry control specimens. Typical load versus elongation curves for kiln-dry, nailed-wood joints in tension are shown in figure 19.

The cycled bending specimens had 1.09 to 1.42 times greater deflection than the controls. Again, the largest ratio was for the kiln-dry specimens because of the smaller deflection of the controls. The effect of the gap left between the gussets and the members of the cycled specimens can be seen in figure 20. The curve for the cycled specimen starts at a flatter slope than the curve for the control. When the gap becomes closed, the two curves have nearly the same slope.

Losses in maximum load due to moisture cycling were not as great as the losses in rigidity. They ranged from 0 to 15 percent loss for the tension specimens and 11 to 19 percent for the specimens. The greater losses were for the initially wet specimens.

19

Figure 19.--Typical load versus elongation curves for kiln-dry, nailed-wood joints in tension.

M 131 602

Figure 20.--Typical load versus deflection curves for kiln-dry, nailed-wood joints in bending. M 131 603

67 20

Corrosion of the nails had an effect upon the maximum loads. At small deformations, the corrosion had practically no effect, since the lateral of the nails was dependent on the bending of the shank, but at maximum load there was some withdrawal of the nail and resist­ance to withdrawal was increased by the rust. All of the cycled specimens, as well as the con­trol specimens from the initially wet moisture content group, had nails that were rusted. Nails from the controls of the other groups had shiny bright shanks. The rusting resulted in higher maximum loads for the control specimens of the initially wet moisture content group and, thus, a greater percentage loss in maximum load due to moisture cycling. This does not mean, however, that joints made of wet material will be more satisfactory in over a long period of time. Over a period of years, the nails will

corrode to the point of being loose and, if extremely corroded, may even break. This will cause a greater loss in and, of more importance, a greater loss in rigidity. Since design loads are based on rigidity, the initially wet group is not as satisfactory as the drier groups, even though its maximum strength is greater because of the initial corrosion of the nails.

Figures and 22 show typical failures. The joints usually failed by splitting of the members or gussets with nearly simultaneous withdrawal of the nails.

The results for the tension specimens com­designpared quite well with the usually

values. There was still a ratio of about 4 to 1 in the maximum loads based on a 2,000-pound design load. The elongation values for the tension con­trol specimens nearly equalled or were less than

Figure 21.--Typical failures in tension for the nailed-wood and nailed-plywood joints. The plywood gusset (bottom) broke in tension. The nailed-wood specimen (upper) split in the gussets and members as nails were withdrawn. M 126 099

21

Figure 22.--Typical failures in bending for the nailed-wood and nailed-plywood joints. The plywood joint failed in the gusset, while the nailed-wood joint (bottom) failed in the member and gusset.

the value of 0.015 inch used to establish design loads. The wet specimens showed the greater elongation but, in design, the allowable load would be multiplied by a factor of three-quarters to compensate for the moisture content (4). The cycled specimens, both in tension and bending, showed greater deformation than the uncycled controls.

Nailed-plywood joints.--As with the nailed-wood joints. the effect of moisture cycling upon the elongation and deflection of the joints was of primary concern, since deformation governs the allowable design load.

The cycled tension specimens had to 3.50 times greater elongation than the controls,

larger ratio being for the kiln-dry group and the smaller ratios for the wet specimens. The cycled specimens made of kiln-dry material, however, were stiffer than the controls made of

M 126 098

wet material. The greater percentage increase in elongation can be accounted for, in part, by the much greater stiffness of the kiln-dry con­trols that had no gap between the gussets and the members, while the other specimens did. The effect of this gap can be seen from the load versus elongation curves shown in figure 23. The curve for the cycled specimens had a much flatter slope under initial loading than the curve for the controls.

The cycled bending specimens had 1.44 to 1.61 times greater deflection than the controls. Again, the larger ratio for the specimens made of kiln-dry material. This can be accounted for by the gap between the gussets and members of all specimens except the kiln-dry controls, which had a greater stiffness. This effect can be seen in figure 24, which shows typical load versus deflection curves.

FPL 22

Figure 23.--Typical load versus elongation curves for kiln-dry, nailed-plywood joints in tension. M 131 604

Figure 24.--Typical load versus deflection curves for kiln-dry, nailed-plywood joints in bending. 131 605

23

Losses in maximum load due to moisture

pro-

and bending, showed greater deformation than the cycling ranged from 0 to 16 percent in tension uncycled controls. and from 10 to 30 percent in bending. As with the The nailed-plywood joints had nearly same nailed-wood joints, specimens with rusty nails maximum loads in tension as the nailed-wood had higher maximum loads. As corrosion joints. The maximum loads in bending, however, ceeds, however, these joints will become weaker were about 500 pounds less. This can be accounted than those made of dry material. for the fact that the plywood gussets were not

Figures 21 and 22 show typical failures. The as thick as those made of wood and only three-joints usually failed in the plywood gusset with fifths of the plywood thickness was effective in very slight withdrawal of the nails. resisting bending moment. This is also true for

The results for the tension specimens compared the tension specimens may account for the quite well with the usually accepted design values. greater effect of moisture cycling on the plywood There was still a ratio of about 4 to 1 in the joints than on solid-wood joints. The plywood ultimate loads on the basis of a 2,000-pound gussets developed checks in the surface veneers design lead. The elongation values for the tension from moisture cycling. This caused a greater control specimens nearly equalled or were less increase in deformation for the plywood joints than the value of 0.015 inch used to establish than for the solid-wood joints. design loads, except for the initially wet speci- Nailed metal-plate joints.--The allowable loads mens. The wet showed greater elonga- for this type of joint are usually based on load tion but, in design, the allowable load would be at a given elongation. The effect of moisture reduced one-quarter to compensate for moisture cycling upon the elongation and deflection of the content (4). The cycled specimens, both intension joints. therefore, was of primary concern.

Figure 25.--Typical load versus elongation curves for kiln-dry, nailed metal-plate joints tension. M 131 606

FPL 67 24

The cycled tension specimens had from 1.00 to 2.67 times greater elongation than the controls. The greatest difference between cycled and con­trol specimens was for kiln-dry group, in which the controls had approximately one-half the elongation of any other group. All values of elongation nearly equalled or were less than the value of 0.015 inch used to establish allowable loads.

The cycled bending specimens had 0.98 to 1.49 times greater deflection than the controls. Here again, the values of deflection were practi­cally the same for all groups. The reason there was practically no difference in rigidity between the different groups, either in bending or tension, is probably because the nail shanks were entirely in the wood members. This would cause the shanks to bend less than with the nailed-wood and nailed-plywood joints. Typical load versus elongation or deflection curves are shown in figures 25 and 26.

Typical failures of these joints were caused

either by popping of the nailheads with a slight withdrawal or tearing of the plate, as shown in figure 27. Thus, there was no effect of moisture cycling upon the maximum loads. The wetter moisture content groups in tension, however, did have slightly higher loads, This again was due to the rusting of the nails, as previously explained.

Barbed metal-plate joints.--These joints were relatively stiff, as will be noted from the typical load versus elongation or deflection curves in figures 28 and 29. The tensile load at 0.015-inch elongation was about 4,200 pounds, which is greater than one-half the maximum load. The Truss Plate Institute, Incorporated (9) requires that the allowable load shall not be greater than one-third of the maximum load; thus, the allow­able design load for the barbed metal-plate joints is governed by maximum load and, therefore, the effects of moisture cycling and initial moisture content upon maximum load were of primary concern.

The cycled tension specimens had from 0 to

Figure 26.--Typical load versus deflection curves for kiln-dry, nailed metal-plate joints in bending. M 131 607

25

Figure 27.--Typical failures for the nailed metal-plate joints. The tension specimen (left) generally failed by popping of the nailheads with slight withdrawal, and splitting of the wood members. The bending specimen (right) generally failed by tearing of the plate.

M 129 986

Figure 28.--Typical load versus elongation curves tor kiln-dry, barbed metal-plate joints in tension. M 131 608

FPL 67 26

Figure 29.--Typical load versus deflection curves for kiln-dry, barbed metal-plate joints in bending.

18 percent loss in maximum load due to moisture cycling. The greatest loss was in the kiln-dry group, where the controls had a maximumload of about pounds more than any other group. The wet specimens had no loss in maximumload; however, the loads for the control and cycled specimens were below the 6,000 pounds needed to establish an allowable design load of pounds. This was due to shrinkage of the members caus­ing the plates to bow out of the wood, leaving the joint held together by only a few of the barbs plus the positioning nails along the outside edges. This same. effect also occurred with the air-dry speci­mens, but to a smaller extent.

The cycled bending specimens had from 2 to 9 percent loss in maximum load due to moisture cycling. The smaller effect on the bending speci­mens, as compared to the tension specimens, is explained by the type of failures that occurred (fig. 30). The bending specimens usually failed by tearing of the plates, while the tension speci­mens failed by withdrawal of the barbs from the wood members.

The barbed metal-plate joints were relatively stiff and moisture cycling had no effect on elonga­tion or deflection. There was some effect due to

M 131 609

initial moisture content, the initially wet control joints having about 4 times greater elongation than the kiln-dry specimens in tension and about 1.24 times greater deflection in bending.

Toothed metal-plate joints.--For this type of connector, the deformation and maximum load values were such that the allowable design load could be governed by either value, Thus, the effects of moisture cycling initial moisture content en elongation, deflection. and maximum load were of equal concern.

The cycled tension specimens had from 1.00 to 2.00 times the elongation of the controls, the greatest ratio being for the kiln-dry specimens, in which the controls had one-half to two-thirds the elongation of the other moisture content groups. All values of elongation were below the value of 0.015 inch used to establish allowable loads.

The cycled bending specimens had 1.12 to 1.30 times greater deflection than the: controls. Again, the greatest percentage increase in deflec­tion was for the kiln-dry specimens, which had the stiffer controls.

Typical load versus elongation or deflection curves are shown in figures 31 and 32.

27

Figure 30.--Typical failures for the barbed metal-plate joints. The tension specimen (left) generally failed by withdrawal of the barbs, while the bending specimen failed by tearing of the plates.

M 129 985

Figure 31.--Typical load versus elongation curves for kiln-dry, toothed metal-plate joints in tens ion. M 131 610

FPL 67 28

Figure 32.--Typical load versus deflection curves for kiln-dry, toothed metal-plate joints in bending.

Tension specimens had from 0 to 9 percent loss in maximum load, the larger loss being for the kiln-dry specimens. Both the cycled and con­trol specimens of this group, however, had higher maximum loads than those specimens made of wet material. Shrinkage of the wet material caused the members to split slightly, thus reduc­ing their lead capacity. Values of maximum load still had a ratio of about 4 to 1 after cycling, based on a design load of 2,000 pounds.

Bending specimens had from 2 to 20 percent loss in maximum load. The larger loss was for specimens made with wet material. Values were nearly the same except for the wet cycled speci­mens, which were about 500 pounds less.

Typical failures are shown in figure 33. Tension specimens generally failed by withdrawal of teeth from the members and splitting of the members. Bending specimens usually failed by splitting of the members along the bot-.

row of teeth. Phenol-resorcinol glued joints.--Allowable

loads for glued plywood joints depend primarily on the strength of the plywood if an adequate bond can be made between the wood members and gussets. For glued joints, therefore, the effect of

M 131 611

initial moisture content upon the bonding of the plywood to the members was of first concern and then, the effect of moisture cycling upon both the plywood and the bond.

No difficulty was encountered in bonding the gussets to the members with the phenol-resorcinol glue at any of the initial moisture contents. AS

the wetter material dried to 10 percent moisture content, however. partial failures occurred in the plywood because of unequal shrinkage between the gussets and wood members. Partial glue-bond failure was also observed in some instances. These partial failures caused about a 200-pound loss in maximum load for the tension specimens and about a 600-pound loss for the bending specimens.

The moisture cycling accentuated the partial failures due to shrinkage. This, coupled with checking of the plywood veneers, caused losses in maximum load of 5 to 21 percent for the tension specimens and 7 to 20 percent for the bending specimens. The greater losses were for speci­mens made of wet material,

Elongation values were small (0.001 or 0.002 inch) and appeared to be unaffected by moisture cycling. Similarly, there was hardly effect

29

Figure 33.--Typical failures for toothed metal-plate joints. Tension specimen (left) gener­ally failed by withdrawal of teeth, while bending specimen (right) generally failed by splitting of the alongmembers the bottom row of teeth. M 129 988

Figure 34.--Typical load versus elongation curves for kiln-dry, phenol-resorcinol glued joints In tension. M 131 612

FPL 67 30

Figure 35.--Typical load versus deflection curves for kiln-dry, phenol-resorcinol glued joints in bending.

upon deflection. Typical load versus elongation or deflection curves are shown in figures 34 and 35.

Typical failures are shown in figure 36. Both the tension and bending specimens generally failed in rolling shear in the plywood. Occasionally there would be partial glue-bond failure in the wet specimen group.

Casein-glued joints.--As with the phenol­resorcinol glued joints, the effect of initial mois­ture content upon the bonding of the plywood to the members was of first concern and then, the effect of moisture cycling upon both the plywood and the bond.

No difficulty was encountered in bonding the gussets to the members for the kiln-dry and air-dry material. For the wet material, the bond appeared to be adequate at the end of fabrica­tion, but during the dead loading for moisture cycling, three of the bending specimens failed. Examination showed complete glue failure. The glue was still damp, even though 3 weeks had been allowed for curing. Since casein glue is a water-base adhesive, the excess moisture in the wood tended to keep the glue soft. The joints made of wet material had about 2,000 pounds less maximum load in tension and 700 pounds less

M 131 613

maximum load in bending than those joints made of drier material.

The cycled tension specimens had from 11 to 23 percent loss in maximum load. The smallest loss was for the wet specimen group. The con­trol specimens for this group, had maximum loads approximately 2,000 pounds leas than the maximum loads of the other controls. Some complete glue failure was noted for the wet specimens due to moisture cycling, and partial glue failures were noted for the rest of the speci­mens. There were also some partial failures of the plywood gussets due to unequal shrinkage,

Cycled bending specimens had from 1 to 14 percent loss in maximum load due to moisture cycling. The smaller loss was for the wet speci­mens. the wet control specimens were about 700 pounds weaker than the other controls and were even less mens. Again, some noted for the wet cycling and partial mens. There were

than the other cycled speci­complete glue failures were specimens during moisture

failures in the other speci­also some partial plywood

failures due to shrinkage. The elongation values were relatively

and showed no effect from moisture cycling. deflection was unaffected by moisture

31

Figure 36.--Typical failures of glued-plywood joints. Both tension (left) and bending (right) specimens generally failed in rolling shear in the plywood.

M 129

Figure 37.--Typical load versus elongation curves for kiln-dry, casein-glued joints in tension. M 131 614

FPL 67 32

Figure 38.--Typical load versus deflection bending.

cycling. Typical load versus elongation o r deflec­tion are shown in figures 37 and 38.

Typical failures are shown in figure 36. Those specimens that held together until destructive loading generally failed in rolling shear in the plywood. Some partial glue failures also occurred, especially in the wet specimen group.

STATISTICAL ANALYSIS

“t” Values

A statistical analysis was made to determine if the differences between cycled and control specimens were significant.

To do this, the hypothesis was made that there was no difference between the cycled and control specimens. The statistical relation

curves for kiln-dry, casein-glued joints in 131 615

mean (in this instance it was assumed to be zero); s = standard deviation of the differences; and N = sample size. The calculated t was then com­pared to a t value associated with a 95 percent confidence level. This value comes from the t distribution and was equal to 2.015 for all of the sample groups except those that had only five specimens. For groups having five specimens, the t value was 2.132.

The calculated t values for tension speci­mens are shown in table 7. For all values larger than the 2.015 or 2.132 reference: values, the hypothesis that there is no difference would be rejected and a significant difference shown, while for values less than the selected values, the hypothesis would be accepted. It is evident from the calculated t values that moisture cycling had significant effects upon the elongation of all the mechanical-connector joints made of kiln-dry material. This was expected, since the kiln-dry controls were much stiffer than the other controls. For the Specimens made of air-dry material, the elongation of the nailed-wood, nailed-plywood, toothed metal-plate, and casein-glued joints showed a significant difference between cycled and control specimens. The fact that the difference in elongation for the air-dry casein-glued joints

33

was calculated for the difference between control and matched cycled specimens. In this equation, X = mean of the differences; X' = the population

Table 7.--Calculated t values for differences between cycled and control specimens in elongation and maximum load for

various types of joints eva lua ted in tension1

was significant should not be given too much importance, since the values of elongation were very small. This is also true of the wet casein-glued joints, which also showed moisture cycling to have a significant effect upon elongation. The only other joint showing a significant effect on elongation from moisture cycling was the nailed-plywood type. With the wet specimens, the con­trols had larger values of elongation than the drier controls, and thus their calculated t values were smaller.

The calculated t values for maximum load in tension showed less significant effects of mois­ture cycling than on elongation. The specimens most affected were wet and air-dry phenol­resorcinol joints and the air-dry casein joints. The reason the wet casein joints did not show a significant difference was because of the low value of maximum load for the controls. Of the mechanical connectors, only the air-dry nailed-wood, kiln-dry barbed metal plate, and kiln-dry toothed metal plates showed a significant differ­ence between cycled and control specimens. In the case of the two metal plates. the kiln-dry controls had maximum loads that were higher

FPL 34

than the maximum loads of the other controls. The effect of moisture cycling on the wet nailed-wood joints would probably have been significant, as it was for the air-dry specimens, except for the temporary effects of corrosion.

The calculated t values for the bending speci­mens are shown in table 8. Deflections of all the mechanical-connector joints made of kiln-dry material were significantly affected by moisture cycling, except the barbed metal plate. Other groups with significant effects of moisture cycling on deflection are the air-dry nailed-wood joints, wet and air-dry nailed-plywood joints, and wet and air-dry toothed metal-plate joints. The rea­son for the kiln-dry specimens having a more significant difference between cycled and control specimens was the smaller deflection of the kiln-dry controls than the deflections of the other con­trol specimens.

The calculated t values for maximum load in bending showed a significant difference between cycled and control specimens for the glued joints made of wet and air-dry material. Of the mechanical-connector joints, the nailed-wood and nailed-plywood joints made of wet and air-dry

Table 8.--Calculated t values for d i f fe rences between cycled and control specimens in deflection and maximum load for the various types of joints evaluated in bending1

material had significant effects on maximumload from moisture cycling. Others with significant differences in maximum load were the kiln-dry nailed-plywood joints, air-dry barbed metal-plate joints, and wet toothed metal-plate joints.

The fact that the more significant effects were fox joints made of drier material does not mean that they are inferior to the joints made of wet material. In practically all instances, the kiln-dry controls were much stiffer and stronger than the controls of the other moisture content groups; thus, they had a greater percentage increase in deflection or elongation and a greater loss in strength. even though the cycled kiln-dry speci­mens were stronger than the other cycled specimens.

“F” Values

or elongation, an analysis of variance was made on the control specimens for the three initial moisture content groups. F values were calcu­lated (table 9) and compared with a value of 3.68, which is associated with a 95 percent confidence level. This analysis showed a significant differ­ence in maximum load in tension due to initial moisture content for the three metal-plate joints [the calculated values were greater than The only joints showing significant differences in maximum load in bending were the toothed metal-plate and glued joints. Maximum loads of all other joints were not significantly affected by initial moisture content. The differences in deflec­tion or elongation were quite significant, except fox the glued joints in tension and the nailed metal plates in bending. Since design loads for mechanical joints are generally on deforma­tion, there is obviously an advantage in using dry lumber.

To determine if initial moisture content had a significant effect upon the strength, deflection,

35

Table 9.--Calculated F values from an analysis of variance between initial moisture-content groups in deflection, elongation, and maximum load for control specimens of the various types of ,joints evaluated in bending and tension1

FPL 67 36

SUMMARY OF FINDINGS

1. Moisture cycling caused deflection of bending specimens to be from 2.1 to 6.9 times that of uncycled specimens and elongation of tension specimens to be from 2.0 to times that of uncycled specimens at the end of moisture cycling.

2. Moisture cycling caused residual deforma­tion that was approximately two to three times that of uncycled specimens.

3. Results of destructive loading showed that cycled tension specimens had from 1.00 to 3.50 greater elongation than the control specimens at a design load of 2,000 pounds.

4. Results of destructive loading showed that cycled bending specimens had from to 1.61 times greater deflection than the control speci­mens at a load of 1,000 pounds.

5. Losses in maximum load ranged from 0 to 23 percent for the cycled tension specimens.

6. Losses in maximum load ranged from 0 to 30 percent for the cycled bending specimens.

7. The maximum loads of cycled specimens were adequately larger than the design loads, except for the glued joints and barbed metal-plate joints made of wet material.

8. Elongation values for control specimens of the mechanical-connector type joints compared favorably with the value of 0.015 inch used to establish design loads.

9. Wet material had an adverse effect upon the glued joints, causing partial glue failures and partial plywood failures due to unequal shrinkage.

10. Wet material had an adverse effect upon the barbed metal-plate joints where shrinkage caused the plates to bow, thus removing the barbs from the wood.

37

LITERATURE CITED

1. Angelton, H.D. Nailed-plywood gusset roof trusses, 4/12 and greater slopes. maximum span-28feet 8 inches. Mimeo. F-40. Purdue Univ., Agr. Exp. Sta., Lafayette, Ind.

2. Luxford, R.F. 1958. Light wood trusses. U.S. Forest Serv.,

Forest Prod. Lab. Rep. 2113. 3. , and Heyer, O.C.

1954. Glued and nailed roof trusses for house construction, U.S. Forest Serv., For­est Prod. Lab. Rep. 1992.

4. National Lumber Association, 1962. National design specification for

stress-grade lumber and its fasten­ings. Washington, D. C.

5. Pneuman, F.C. 1960, King-post trusses with plywood gussets

one side. Tech. Bull. Douglas Fir Ass., Tech. Dep., Tacoma, Wash.

6. Radcliffe, B.M., and Granum, H. 1955. A new low-pitched roof truss with nail-

glued connections. Agr. Exp. Sta. Bull. 617, Wood Res. Lab., Purdue Univ., Lafayette, Ind,

7. , Granum, H., and Suddarth, S.K. 1955. The Purdue-Illionis nail-glued roof

truss with pitch of 3:12 and 4:12 for spans of feet 8 inches and 28 feet 8 inches. Agr. Exp. Sta. Bull 629. Wood Res. Lab., Purdue Univ., Lafayette, Ind.

8. , and Suddarth, S.K. 1955. The Purdue-Illinois nail-glued truss

with a pitch of 2:12 for spans of 24 feet 8 inches and 28 feet 8 Agr. Exp. Sta. Bull. Wood Res. Lab., Purdue Univ., Lafayette, Ind.

9. Truss Plate Institute, Inc. 1965. Design specifications far light metal

plate connected wood trusses TPI-65. Miami, Fla.

FPL 38 1.2-45