United States

Department of
Agriculture Strength and
Stiffness of
Forest Service

Forest
Products

Light-Frame
Laboratory
Research
Paper
FPL 471
Sloped Trusses
Ronald W. Wolfe
Donald H. Percival
Russell C. Moody
Abstract Contents

Bending tests conducted on 42 wood trusses provided Page

information on the effects of truss slope and lumber modulus Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 1
of elasticity (MOE) on truss strength and stiffness. Two
28-foot-span Fink truss configurations with slopes of 3/12 Literature Review . . . . . . . . . . . . . . . . . . . . 2
and 6/12 were tested. Component lumber was all No. 2
grade southern pine separated into low, medium, and high Objectives and Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
MOE categories. All trusses exhibited linear load vs.
deformation behavior up to and beyond twice their design Research Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
load range. Variations in load carrying capacity were small Truss Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
within a slope/stiffness category, and lumber stiffness Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
appeared to be closely correlated with load capacity, failure Truss Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
mode, and truss stiffness. Information in this report is an TrussTests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
important first step toward formulating improved structural Analysis Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
analysis procedures for roof systems.
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Keywords: Roof, Trusses, Light-frame, Strength, Stiffness, Lumber Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Test, Truss plates TrussStiffness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Truss Strength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Analysis of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
TrussStiffness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Truss Strength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Literature Cited . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Appendix-Summary of Truss Member MOE

Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

April 1986

Wolfe, Ronald W.; Percival, Donald H.; Moody, Russell C. Strength and
stiffness of light-frame sloped trusses. Res. Pap. FPL 471. Madison, WI:
U.S. Department of Agriculture, Forest Service, Forest Products
Laboratory; 1986. 16 p.

A limited number of free copies of this publication are available to the public
from the Forest Products Laboratory, One Gifford Pinchot Drive, Madison, WI
53705-2398. Laboratory publications are sent to over 1,000 libraries in the
United States and elsewhere.

The Laboratory is maintained in cooperation with the University of Wisconsin.

United States Errata
Department of
Agriculture

Forest Service Strength and Stiffness

of Light-Frame Sloped
Forest Trusses
Products
Laboratory

Research
Paper
FPL 471 Errata

April 1986

In Figure 10, page 12. t h e u n i t o f m e a s u r e f o r t h e v e r t i c a l a x i s

should be
"Thousand lb"
not
"Thousands"

as written.
2
"Million lb/in. ”
should be the unit of measure for the horizontal axis.

Wolfe, Ronald, W.; Percival, Donald H.;

Moody, Russell C. Strength and
Stiffness of Light-Frame Sloped
Trusses. Res. Pap. FPL 471. Madison, WI:
U.S. Department of Agriculture.
Forest Service, Forest Products
Laboratory; 1986. 16 p.
 Strength and
Stiffness of
Light-Frame
Sloped Trusses1

Ronald W. Wolfe,* Research General Engineer

Donald H. Percival,** Research Professor of Wood Technology
Russell C. Moody,* Research General Engineer

*Forest Products Laboratory, Madison, WI

**Small Homes Council, University of Illinois, Champaign, IL

Introduction

Roof systems used in residential construction have an Another concept which should receive greater attention is
outstanding record of structural performance. This is due in system design. Current roof design procedures are based on
part to the use of conservative design assumptions. Current single member design and give little credit to the advantage
design procedures make no attempt to account for this of repetitive member systems. In light-frame trussed roof
conservatism; therefore, the inherent margin of safety in systems, each truss is designed as though all its chord and
these roof systems cannot be quantified. This report covers web members have a near minimum clear wood strength
the first phase of a research program to evaluate the relative and the maximum allowable strength reducing characteristic
reliabilities of various roof system designs. Specifically, the permitted for their grade. In addition, it is assumed that each
objective of this study is to characterize the strength and truss in the system must be designed to carry its tributary
stiffness performance of two light-frame truss configurations. area of design load. Limited recognition is given to load
sharing mechanisms, either within the truss or between
Designers of steel and concrete structures have taken the trusses. In order to improve roof system design, analysis
lead in applying ‘reliability theory’ to structural design. They methods must be developed to minimize unrecognized
have published standards which improve design efficiency conservatism and provide a more accurate assessment of
by taking into account variability of material as well as loads. the true margin of safety. Methods are needed a) to
The trend toward reliability-based design, evident for these accurately characterize the distribution of truss stiffness and
materials, points the direction for improvements to strength, and b) to predict how loads are distributed among
engineered wood design. Most of the work required to rafters within a roof system. This report presents individual
characterize structural loads has already been done. The truss test results which will be used to evaluate the accuracy
major concern of researchers dealing with reliability-based of theoretical models developed to characterize roof system
design of wood structures, therefore, is to characterize the performance.
probability distributions associated with structural capacity.

1
This study was conducted in cooperation with the Southern Pine Inspection
Bureau and the Truss Plate Institute.
Literature Review

Industry design standards for metal plate connected trusses Along with the development of computer models, data have
were first presented by the Truss Plate Institute (TPI) in also been collected on loads, member strength properties,
1960. The approach was deterministic, connections were and truss connector plate moment and axial load resistance.
assumed to transfer only axial loads, and chord elements Thurmond et al. (1983) provided a summary of roof loads
were analyzed as simple beams with superimposed and derivation of distribution parameters. Truss lumber
moments. strength and stiffness properties have been collected
(Gerhards and Percival 1979) and characterized (Hoyle et
Increased popularity of computers in the 1960’s resulted in al. 1979) to indicate correlations between stiffness and
significant improvements to truss design through the strength. The In-grade test program being conducted as a
development of computer programs capable of providing cooperative effort between lumber grading agencies and the
indeterminate analyses. The most notable of these programs Forest Products Laboratory (FPL) is supplying full-size
for the truss industry was the Purdue Plane Structures member strength and stiffness data for bending, tension and
Analyzer (PPSA) (Suddarth 1972, Suddarth and Wolfe compression (Galligan et al 1980).
1983). PPSA gave a more accurate analysis of member
stresses than the TPI procedure, however, it was considered A number of studies have been conducted to characterize
too costly to run and too complicated to use for everyday the load-slip behavior of several truss connector plates
truss design. PPSA was therefore used by the truss industry (Foschi 1979; Perkins et al. 1962; Quaile and Keenan 1979).
as a means of checking and improving determinate analysis Also, work conducted by Zahn (1982, 1984) at FPL, and by
methods described in TPl’s design specifications (TPI 1978). Buchanan at the University of British Columbia (Buchanan
1984) on the analysis of the effects of combined
Other truss analysis programs have also been developed for axial-bending loads provide a data base for the analysis of a
the analysis of wood structures. A program called SADT, stress limit state for the chord and web members. Necessary
developed by Foschi (1977a & b), provides options to material property information either is or soon will be
account for sheathing contributions and plastic behavior in available to begin development of a reliability-based design
truss plate connectors. This program permits procedure for roof systems. Before these design procedures
characterization of displacement behavior beyond the design can be implemented, however, available material properties,
range whereas PPSA is a design tool, oriented to the analysis models, and load information must be refined and
analysis of trusses and frames within the elastic range. decisions must be made regarding limit state criteria.

The truss industry design methods have evolved following

methods recommended in the National Design Specification
for Wood Construction (NDS), (Nat. For. Prod. Assoc. 1982).
The NDS recommends use of lumber stress values derived
according to standards published by the American Society
for Testing and Materials (ASTM) Standard D 245 (ASTM
1983c). These values are based on a clear wood strength
(ASTM D2555-83), which is estimated to be at the
5th percentile of the distribution of all clear wood pieces.
Strength reductions for grade characteristics assume knots
and/or slope of grain are the maximum permitted. Thus each
truss is designed as though each member is of minimum
strength and quality for its species and grade.

Egerup (1975, 1979, undated) used Monte Carlo simulation,

along with programs similar to those developed by Suddarth
and Foschi, to assess the distribution of stiffness and load
capacity of roof trusses. Results of his studies suggest that
the 5th percentile value for truss load carrying capacity is
greater than the load capacity predicted using the
5th percentile value for strength of individual members. He
attributes this to several factors:

(a) loads are distributed away from weak members

(b) there is a low probability that a member with the
5th percentile strength will be located at the point of
maximum stress, and
(c) two or more members may show some form of plastic
deformation before a truss actually fails.

2
Objectives and Scope Research Methods

The objective of this study is to characterize the strength The two truss configurations considered were designed by
and stiffness performance of two light-frame truss representatives of the TPI. Prior to manufacturing the
configurations. Results will be used in the development and trusses, the lumber was sorted into three modulus of
evaluation of a truss strength/stiffness model. Undamaged elasticity (MOE) categories. For each combination of truss
trusses from this study will also be used in a subsequent configuration and MOE category, seven trusses were
study to evaluate the influence of system interactions on the manufactured using metal plate connectors; three trusses
load-deflection response of trusses used in a full-scale roof. were tested to 1.25 x design load and four trusses were
tested to failure. A description of the truss test variables is
Forty-two full-size trusses were evaluated. Twenty-four were given in table 1.
tested to failure and the other 18 were tested to 1.25 times
their design load to determine their stiffness characteristics. Truss Design
The truss configurations were selected to be typical of
commonly used residential roof trusses. Both designs were
for 28-foot-span Fink trusses to be spaced 24 inches on
center. Both were designed for top chord load only. The
3/12 sloped truss (fig. 1) had a design live load of 17.5 Ib/ft2,
and the 6/12 sloped truss had a design live load of 23 Ib/ft2.
The dead load was 10 Ib/ft2 in each case. Total design loads
for the 3/12 trusses (1,540 pounds), and for the 6/12 trusses
(1,850 pounds), were for 10 years duration and were
controlled by stresses in the top chord. The ratio of actual
stress to allowable stress (sec. 303.6 of TPI 1978) at some
location in the top chord of each truss design was equal to
1.0. The bottom chord had no vertical load, thus the ratio of
actual to allowable was much lower (0.46 for the 3/12 and
0.20 for the 6/12).

In both cases, the metal plates were designed to be larger

than normal (i.e. either thicker or larger area) to force failure
to occur in the wood. The rationale for this “over plating”
was that the allowable stresses for lumber include an
adjustment for load duration and those for steel do not. A
short-term test to determine truss strength would have a
greater chance of resulting in metal failure than wood failure,
if both were selected on the basis of the same ratio of
imposed stress to allowable stress at the design load.
Industry cooperators recommended that it was important to
force failure to occur in the wood in order to supply data to
evaluate the ability of a computer model to predict the
probability of wood failure.

Table 1 .—Description of southern pine truss tests

Number of trusses
MOE 3/12 Slope 6/12 Slope
Total
category Loaded to Loaded to Loaded to Loaded to
1.25 x DL1 1.25 x DL1
only failure only failure

Low (L) 3 4 3 4 14
Medium (M) 3 4 3 4 14
High (H) 3 4 3 4 14
TOTAL 9 12 9 12 42
1
DL = design load.

3
Figure 1.—Test trusses with 3/12 and 6/12 top chord slopes were
designed for this study. In all cases, the heel and bottom chord
splice joints were made using 16-gauge metal connector plates.
Remaining plates were 20 gauge. Plate dimensions (in inches) are
shown near each joint. (ML85 5612)

Material
The southern pine lumber used for this study was No. 2
2 x 4 in lengths of 12, 14, and 16 feet. Each piece was
visually graded to meet the requirements for No. 2 and
marked with an identification number. It was then passed Figure 2.—Truss plates used in the fabrication of the test trusses.
through a Continuous Lumber Tester (CLT) machine to The 16-gauge steel plate (top) had a tooth density of 4-1/2 teeth/in.2
measure an average MOE value along its length. Additional and the 20-gauge steel plate (bottom) had a tooth density of
evaluations of the lumber are discussed in Appendix A. Part 8 teeth/in.2. In both cases, the individual teeth are approximately
of the lumber, obtained from Alabama, had MOE values 3/8-inch long by 1/8-inch wide. (ML85 5613)
ranging from 1.4 to 2.6 million lb/in 2. A second portion,
obtained from Arkansas, had MOE values ranging from
0.8 to 1.4 million lb/in 2.
Various lengths were randomly selected within each of the
A 16-gauge metal truss plate supplied by one manufacturer three MOE categories and assigned to 1 of 14 trusses
was used at the heel and bottom chord splice joints, and a (7 with 3/12 slope and 7 with 6/12 slope). Lumber for each
20-gauge plate by another manufacturer was used at all truss was marked with a four part identification which gave
other joints. In both cases, the teeth were approximately the slope of the truss (3,6), its MOE category (L,M,H), a
3/8-in. long by 1/8-in. wide (3/8 in. by 5/32 in. for the sequence number (1-7), and a code related to the location in
16 gauge, and 5/16 in. by 1/8 in. for the 20 gauge, see the truss (fig. 3). Hydraulic presses were used to install the
fig. 2). truss plates during manufacture. The trusses were then
delivered to the testing laboratory where they were stored
Truss Fabrication inside for at least 7 days prior to testing.

The lumber was divided into three MOE categories Truss Tests
according to the average MOE values determined by the
CLT machine. Lumber in the low stiffness (L) category had The trusses were evaluated in a horizontal position using a
MOE values ranging from 0.8 to 1.4 million lb/in 2. The range test facility that consisted of parallel reaction brackets laid
for the medium stiffness (M) was 1.4 to 2.0 million lb/in 2, into a concrete floor (fig. 4). Hydraulic loading rams, lateral
and for the high stiffness (H) was 2.0 to 2.6 million lb/in.2. restraint braces, and reaction pads were attached to these
The few pieces with MOE values outside of these ranges brackets and could be moved in one direction in the plane of
were not used. the floor to accommodate any truss configuration with a
span of up to 35 feet and a height of up to 10 feet.

4
The ends of the truss were placed on 4-inch-wide reaction
pads spaced 28 feet apart (outside to outside). Thus, the
span from center-to-center of each reaction was 27 feet
8 inches. Twenty-eight concentrated loads were applied at
12-inch intervals along the top chord of each truss to
simulate a uniform load application. These loads were
applied using 14 hydraulic cylinders spaced 24 inches apart.
Each cylinder was centered on a distributor beam which
transferred the load to the top chord through two load points
spaced 1 foot apart. Figure 3.—Identification of truss members and deflection points.
Truss members are identified by use and location: top chord (TC),
Deflection readings were taken using a taut wire stretched bottom chord (SC), and web (WB). Member numbers are assigned
between points on the centroidal axis of the bottom chord left to right corresponding to the orientation of the truss in the test
above the reactions. Wood lath strips extending from seven apparatus. Deflection points are circles and are also numbered left
points along the top chord held mirrored gauges under the to right. (ML85 5614)
wire to enable measurement of top chord deflections.
Mirrored gauges were also placed at five points along the
bottom chord. These positions are shown as the circled
locations numbered 1-12 on figure 3.

Calibrated proving rings positioned at the reactions were

used to measure the relationship between the hydraulic
pump pressure gauge and the applied load on the truss.
Two pilot trusses of each slope were tested to develop
pressure versus load curves for the two truss configurations.
These calibration procedures were also checked using
electronic load cells. Measurements taken from the pressure
versus load curves were found to be within 2 percent of
those measured directly using the load cells. For the
majority of tests, end grain Douglas-fir blocks were inserted
at the reaction points and loads were read from the pressure
gauge.

An initial load of 1,000 pounds was applied to each truss

and held for 5 minutes to check for alignment, settlement,
final adjustment, and then removed. Loads were then
applied in increments of 0.25 x design load, and deflection
readings were taken at each load increment after the load
had been maintained for 5 minutes. Total time of test varied
from about 1/2 hour for those tested to determine stiffness
only to nearly 2 hours for the stronger trusses tested to
failure.

For each combination of top chord slope and MOE category,

three trusses were loaded to 1.25 x design load and four
trusses were loaded to failure. For the trusses loaded to
failure, deflection readings were taken at each increment
until either failure appeared imminent or four times design
load was reached. Ultimate load and failure locations were
recorded for each truss along with comments regarding
apparent influencing factors and failure progression.

Analysis Methods
Figure 4.—Truss test facility at the Small Homes Council, University
Data obtained included lumber stiffness, and the stiffness, of Illinois, Champaign-Urbana with a 6/12 truss. (M85 0005)
strength, and failure mode of each truss. Linear regression
analyses were used to show how lumber MOE, truss slope, strength and failure mode of those trusses tested to
and load level affected truss response to increased load. destruction, it was possible to estimate the load capacity of
Evaluations of the effect of lumber MOE on truss strength those trusses tested to only 1.25 x design. This information
and failure mode were conducted to aid the prediction of will be used in the planning stages of full-scale roof tests
truss failure. By correlating lumber and truss stiffness to which will use the nondestructively tested trusses.
Results

Individual trusses will be referred to by their three-part Table 2.—Summary of modulus of elasticity values for lumber
identification code number, as discussed previously under in each truss category
truss fabrication. Truss members have an additional code 3/12 Slope 6/12 Slope
related to their location in the truss (fig. 3). MOE Coefficient Average Coefficient
category Average of of
In all cases, the discussion of truss stiffness in the load MOE MOE variation
variation
range from 0 to 1.25 x design load involves measurements
on all seven trusses in each slope-MOE category. Million lb/in.2 Pct Million lb/in.2 Pct
Discussion of truss strength, however, refers only to the four
Low (L) 1.23 9.4 1.24 9.2
trusses in each category which were tested to failure. Medium (M) 1.67 9.1 1.73 10.7
High (H) 2.30 6.7 2.31 7.2
Lumber Properties
Table 2 gives a summary of MOE values measured for
lumber used in the trusses in each slope/stiffness category.
MOE values for each truss member are given in the
Appendix.

Moisture content readings (resistance meter) taken at the

time of testing indicated that all members of each truss had
close to the same moisture content. Trusses 6H1 through
6H5 had moisture contents of 9 percent. All other trusses
were at 10 percent.

Truss Stiffness
Within the range of design load, deflections appeared to be
symmetric and linear with load. Figure 5 shows typical
deflection profiles for a truss loaded to 1, 2, and 3 times
design load. A parabolic curve connecting the panel point
deflections is indicative of the global deflection of the truss,
while deviations from this curve represent localized bending
deflections in the loaded top chord. Significant localized
bending deflection is evident in the top chord while all points
on the bottom chord appeared to fit the expected parabolic
pattern.

Figure 5.—Deflection profile of a low stiffness 3/12 sloped truss

(3L7) loaded to 1, 2, and 3 times the design load of 1,540 lb. Loads
were applied at l-foot intervals along the top chord, and deflections
were measured at all web connections and midway between web
connections for both top and bottom chords. (ML85 5615)

6
Figure 6 is a typical set of load versus deflection curves for
one side of a symmetric truss. Deflections of node points 3,
4, and 6 represent global deflection of the truss. Deflections
measured at mid-panel points (1 and 5) are a combination of
localized bending of the top chord and global deflection of
the truss.

Due to its close proximity to the heel reaction, the global

deflection of point 1 is much smaller than that for point 5. In
fact, for every truss, point 5 (or its symmetric point 6) in the
panel adjacent to the peak deflected more than any other
point measured. If the node point deflections are used as an
indicator of truss stiffness, figure 6 indicates that a linear
load deflection model would be an adequate predictor of
truss global deflections up to twice design load.

Truss Strength
Table 3 provides a summary of ultimate loads, and gives
failure for each truss tested to destruction. These trusses
failed during the 5-minute holding period between load
applications. Of the 24 trusses tested to failure, 17 had
lumber failures in the top chord, 3 failed due to knots or
brash wood in the tension chord, and 4 failed at web
connections. Examples of all types of failures are shown in
figure 7.

The 17 top chord failures included 12 associated with knots,

and 5 in wood free of knots. All these failures were due to
combined bending and compression stresses. Failures in
knot-free wood included three which were initiated over the
compression web connection and two unusual failures.
Truss 3M3 failed when a piece of wood flew out of the end
panel of the top chord (fig. 8 top). Inspection of the break in
truss 3M3 suggested that some ring shake may have been Figure 6.—Typical plot of truss deflection versus the design load
present in a region of two edge knots. When compressive ratio. Points 3, 4, and 6 correspond to web-chord connections.
load on the remaining effective section exceeded the Point 1 is mid-panel of the top chord section adjacent to the truss
threshold required to cause it to buckle, it suddenly pushed heel and point 5 is mid-panel adjacent to the peak. (ML85 5616)
a 2-foot-long piece away from the surface. The other
unusual failure occurred in truss 6M1. The top chord
member (TC1) seemed to fail in horizontal shear (fig. 8
bottom). Lengthwise splits in the area of the neutral axis of
this member ran from the web connection to the splice joint
between members TC1 and TC2.
Table 3.—Ultimate load and description of failure
Truss Design load Failed Description of failure
Ultimate load ratio member1
identification
Lb

3/12 SLOPED TRUSSES

3L1 4,620 3.00 TC1 Edge knot

3L3 5,000 3.25 TC1 Edge knots
3L5 4,240 2.75 TC4 Edge knot
3L7 5,000 3.25 BC2 Brash failure at edge of heel plate
Average 4,720 3.06

3M1 6,540 4.25 BC1 Centerline knot

3M3 5,780 3.75 TC4 Shake and edge knots
3M5 6,540 4.25 TC1 Edge knots
3M7 7,320 4.75 TC4 Splinter in clear wood
Average 6,550 4.25

3H2 7,320 4.75 WB2 Peak plate connection

3H4 8,080 5.25 WB2, WB3 Peak plate connection
3H6 7,320 4.75 BC1 Edge knot
3H7 7,320 4.75 TC1 Centerline knot
Average 7,510 4.88

6/12 SLOPED TRUSSES

6L2 6,010 3.25 TC1 Brash failure over web connection

6L3 6,010 3.25 TC1 Edge knot
6L5 6,470 3.50 TC2 Edge knot
6L7 6,010 3.25 TC1 Brash failure over web connection
Average 6,120 3.31

6M1 10,630 5.75 TC1 Shear at top chord splice

6M2 9,240 5.00 TC2 Edge knot
6M4 9,240 5.00 TC2 Centerline knot
6M7 9,240 5.00 TC1 Splinter over web connection
Average 9,590 5.19

6H1 11,090 6.00 WB3 Wood failure in web bottom chord

connection
6H2 7,390 4.00 WB3 Wood failure in web bottom chord
connection
6H6 8,780 4.75 TC1 Centerline knot over web connec-
tion
6H7 9,240 5.00 TC4 Centerline knot over web connec-
tion
Average 9,120 4.94
‘Primary piece of lumber in the truss (fig. 3) involved in failure.
Figure 7.—Examples of different failure types. Top: Top chord failures resulting from combined bending and compressive stresses. Middle: Bottom
chord failures resulting from tensile stresses. Bottom: Connection failures in a bottom chord joint (left) and a peak joint (right). (M85 0003)

9
 Analysis of Results

Table 2 gives the mean values and coefficient of variation of

MOE values for the lumber in each truss category. There
was little difference between the MOE of lumber used in the
3/12 and 6/12 trusses.

Truss Stiffness
Table 4 gives three values of the average global stiffness for
trusses in each slope-MOE category. These values were
derived by using a linear regression analysis to determine
the slope of the load versus deflection curve of each truss
up to a load of 1.25 times the design load. Within each
category, these slopes were averaged for the top chord
node points (3 and 10), the bottom chord node points
(4 and 9), and the peaks (6) (fig. 3). For 3/12 trusses, it is
evident that deflection increases slightly as you move toward
the center of the truss. However, this is not the case for the
6/12 trusses. For these trusses, the average peak
deflections were about the same as those of the top chord
(nodes 3 and 10).

Table 4.—Truss stiffness represented by the average slope of

the load deflection curves determined at the five web-to-
chord connections1
Top chord2 Bottom chord3 Peak4
MOE
category Average COV5 Average COV5 Average COVs

Lb/in. Pct Lb/in. Pct Lb/in. Pct

3/12 TRUSSES

Low 3,460 8.8 3,270 9.1 3,220 9.6

Medium 5,030 6.9 4,570 6.2 4,670 8.4
High 6,020 5.3 5,600 6.5 5,550 6.6

Figure 8.—Unusual failures showing a buckling type failure (top) 6/12 TRUSSES
and shear type failure (bottom) in top chord. (M85 0004)
Low 9,820 6.3 9,040 5.1 9,740 3.5
Medium 13,920 9.7 13,010 7.0 14,540 6.8
High 18,690 9.4 17,820 8.0 18,740 9.1
1
Values are based on averages of seven tests up to 1.25 times design
The three tension chord failures occurred in the 3/12 slope load.
trusses. Two of these were associated with knots, and one
2
occurred in brash wood close to the heel joint (middle of Averages of points 3 and 10 shown in fig. 3.
fig. 7). 3
Averages of points 4 and 9 shown in fig. 3.
All four connection failures were in the high MOE 4
Point 6 shown in fig. 3.
trusses-two in each slope category (bottom of fig. 7). For
the 3/12 trusses these failures occurred at the web-to-peak 5
Coefficient of variation.
joint. In these cases, the peak plates were misplaced by as
much as 5/8 inch resulting in reductions in the number of
teeth connecting the plate to the web. Estimated reduction in
plate holding capacity was 13 percent for truss 3H2 and
29 percent for 3H4. For the two 6/12 trusses it was the
bottom chord-tension web connection that failed. In both
these cases, a layer of wood as thick as the tooth length
was torn out of the web member.

10
The effects of lumber MOE on truss stiffness can be Table B.-Linear regression parameters for deflection of 3/12
evaluated by comparing values given in tables 2 and 4. sloped trusses. Values correspond to average web-chord
Table 2 shows that the average lumber MOE increased connection deflections at three increments of design load
about 40 percent with each change in lumber MOE category (DL);
DEFLECTION = A (Design Load Ratio) + B
(L-M and M-H). Table 4 shows that the 6/12 trusses exhibit
about a 40 percent stiffness increase for both steps in Load increment
lumber MOE category. For the 3/12 trusses, the low to Truss 0-DL DL-2xDL 2xDL-3xDL
medium step is also close to 40 percent: however, the No.
medium to high step is closer to 20 percent. This reduction A B A B A B
in stiffness increment may be due to the higher member 3L1 0.454 -0.002 0.487 -0.036 1.004 -1.070
stresses that develop in the low sloped, high MOE lumber 3L3 0.471 0.007 0.545 -0.067 0.850 -0.677
trusses. These higher member stresses in turn place greater 3L5 0.462 0.005 0.548 -0.081 0.578 -0.141
stress on the truss plate connections which may have 3L7 0.438 0.002 0.573 -0.134 0.825 -0.637
Average 0.456 0.000 0.541 -0.085 0.814 -0.631
caused significant slip in the truss joints. This slip could
cause these trusses to have less apparent stiffness. Table 4 3M1 0.318 0.006 0.336 -0.012 0.507 -0.353
also shows that the 6/12 trusses were about three times as 3M3 0.334 0.009 0.366 -0.022 0.523 -0.336
stiff as the 3/12 trusses. 3M5 0.292 0.001 0.308 -0.017 0.466 -0.332
3M7 0.330 0.000 0.345 -0.008 0.437 -0.191
Linear regression techniques were also used to characterize Average 0.318 0.000 0.345 -0.026 0.483 -0.303
the load deflection response of trusses beyond the design
load. This analysis, which was based only on those trusses 3H2 0.262 0.004 0.229 0.038 0.349 -0.203
tested to failure, resulted in a trilinear curve to represent the 3H4 0.247 0.000 0.344 -0.097 0.303 -0.016
3H6 0.289 -0.002 0.293 -0.005 0.390 -0.200
average performance of trusses in each category. Slopes 0.271
3H7 0.003 0.259 0.015 0.353 -0.173
were derived for three segments of a load vs. deflection Average 0.267 0.000 0.282 -0.015 0.349 -0.148
curve which represents the average deflection measured at
five truss node points (Points 3, 4, 6, 9, and 10, fig. 3).
These segments were selected as multiples of design load
(DL) beginning at zero and ending at 3 times design load. Table 6.—Linear regression parameters for deflection of 6/12
Tables 5 and 6 give the parameters derived to fit straight sloped trusses. Values correspond to average web-chord
line segments to these load intervals. These parameters connection deflections at three increments of design load
were averaged for each stiffness category to give the (DL);
trilinear curves shown in figure 9. These regression DEFLECTION = A (Design Load Ratio) + B
parameters were derived to model deflection as a function of Load increment
the design load ratio (applied load/design load). In order to Truss 0-DL DL-2xDL 2xDL-3xDL
convert the slope (A) of the first segment (zero to design No.
load) shown in tables 5 and 6 to a truss stiffness, one must A B A B A B
multiply the truss design load by 1/A. 6L2 0.178 -0.006 0.191 -0.019 0.222 -0.082
6L3 0.183 -0.003 0.242 -0.061 0.345 -0.268
Truss Strength 6L5 0.198 -0.004 0.259 -0.066 0.368 -0.283
6L7 0.197 0.001 0.263 -0.065 0.404 -0.349
Lumber MOE had an effect on both the failure modes and Average 0.186 0.000 0.239 -0.053 0.335 -0.245
strength of the trusses. Most of the low and medium MOE 6M1 0.124 -0.001 0.121 0.002 0.108 0.028
trusses failed due to bending in the end panels of the top 6M2 0.134 0.002 0.141 -0.006 0.113 0.050
chord which would be expected based on the truss design. 6M4 0.123 -0.002 0.137 -0.016 0.138 -0.018
For the high MOE trusses, there was a mixed mode of 6M7 0.121 -0.004 0.170 -0.053 0.149 -0.010
failure with half of them failing in the web-to-chord Average 0.124 0.000 0.142 -0.018 0.127 0.012
connection.
6H1 0.110 -0.003 0.119 -0.013 0.112 0.002
The effect of lumber MOE on strength is shown in figure 10 6H2 0.109 -0.002 0.116 -0.009 0.106 0.012
6H6 0.091 -0.005 0.103 -0.017 0.138 -0.087
which contains plots of the maximum load of each truss
6H7 0.109 -0.002 0.116 -0.009 0.106 0.012
versus the average MOE of its component lumber. Within Average 0.102 0.000 0.114 -0.012 0.115 -0.015
most MOE-slope categories, the trusses show little strength
variation. An exception is the high MOE 6/12 trusses. The
highest and lowest strength trusses in this group were the
first two tested, and some difficulties in setup and calibration
of the equipment may have influenced the failure loads.
Also, the two types of failures (both bending-compression
and web-to-chord connection) may have contributed to the
higher variability for this group.

11
 A regression analysis was used to determine the relationship
between lumber MOE and truss strength. The regression
lines shown in figure 10 for the 3/12 trusses includes the
mean and both the upper and lower 95 percent confidence
limits on the prediction curve for individual values.

A quadratic model gave a significantly better fit than either a

linear or logarithmic model for the 3/12 trusses:

P = 7953 MOE - 1522 MOE2 - 2650

where
P = maximum load (Ibs)
MOE = average lumber modulus of elasticity (million
Ib/in.2).

For the 3/12 trusses, 91 percent of the variation in strength

could be explained by the model.

A satisfactory regression relationship could not be obtained

Figure 9.—Trilinear curves fit to load-deflection data of three MOE
for the 6/12 trusses; mainly due to the high MOE trusses
categories of 3/12 sloped trusses (left) and 6/12 sloped trusses
which had both high variability and lower-than-expected
(right). Parameters for these curves are given in tables 5 and 6.
strength. Possible reasons for this have been previously
discussed and this group was omitted from further analysis.
(ML85 5617) Because the failure modes of the trusses in both slope
categories with both low and medium MOE were similar, the
expected performance of the 6/12 trusses was represented
as a constant multiple of the 3/12 trusses.

The plot shown for the 6/12 trusses in figure 10 is the

function used for the 3/12 trusses multiplied by 1.38 which is
the average ratio of strength of the low and medium MOE
ranges of 6/12 and 3/12 trusses. The fit of this curve
suggests that the 6/12 trusses with low and medium MOE
lumber may have been slightly more sensitive to the lumber
MOE than the 3/12 trusses. For the small amount of data
involved, however, the 38 percent increase on the 3/12
model gives a reasonable fit to the 6/12 data.

Figure 10.—Ultimate load of trusses as affected by the average

MOE of truss members. Quadratic regression and 95 percent
confidence limits were derived on the basis of 12 trusses tested to
failure for the 3/12 slope. (ML85 5618)

12
Conclusions Literature Cited

These test results demonstrate truss performance American Society for Testing and Materials. Standard
characteristics which will be useful in modeling the stiffness Method for Establishing Structural Grades and Related
and strength of light-frame wood trusses to be used in tests Allowable Properties for Visually Graded Lumber. ASTM
to evaluate a full-scale roof system model. Specifically: D245-81. Philadelphia, PA: American Society for Testing
and Materials; 1983.
1. All trusses tested showed linear deflection performance American Society for Testing and Materials. Standard
up to twice the design load and only slight deviations Methods for Establishing Clear Wood Strength Values.
from linearity up to three times design. ASTM D2555-81. Philadelphia, PA: American Society for
Testing and Materials; 1983.
2. Test results show little variability in truss strength and Buchanan, A. H. Strength model and design methods for
stiffness within a given slope-MOE category. bending and axial load interaction in timber members.
Vancouver, BC: University of British Columbia,
3. There appears to be a definite correlation between Department of Civil Engineering; 1984. Ph. D. thesis.
lumber stiffness, load capacity, and failure mode for Egerup, A. R. Monte Carlo simulation of the ultimate load
these trusses. of timber trusses. Structural Research Laboratory,
Technical University of Denmark; undated.
Caution should be used in applying these results to Egerup, A. R. Theoretical and experimental determination
production trusses because: of stiffness and ultimate load of timber trusses. Rep. R62.
Structural Research Laboratory, Technical University of
1. All contained larger truss plates than required in order to Denmark; 1975.
force failure to occur in the wood member. Egerup, A. R. European practice and future development in
the design of metal plate connected wood trusses. In:
2. No loads were applied directly to the bottom chord as Proceedings, Metal plate wood truss conference; Forest
would be expected in most construction. Products Research Society; 1979: 117-122.
Foschi, R. O. Analysis of wood diaphragms and trusses.
Part I: Diaphragms. Canadian Journal of Civil Engineering.
4(3): 345-352; 1977a.
Foschi, R. O. Analysis of wood diaphragms and trusses.
Part II: Truss plate connections. Canadian Journal of Civil
Engineering. 4(3): 353-362; 1977b.
Foschi, R. O. Truss plate modeling in the analysis of
trusses. in: Proceedings, Metal plate wood truss
conference; Forest Products Research Society; 1979:
1557-1574.
Galligan, W. L.; Green, D. W.; Gromala, D. S.;
Haskell, J. H. Evaluation of lumber properties in the
United States and their application to structural research.
Forest Products Journal. 30(16): 45-51; 1980.
Gerhards, C. C.; Percival, D. H. Species grades and
mechanical properties of lumber sampled from truss
fabricators. In: Proceedings, Metal plate wood truss
conference; Forest Products Research Society; 1979.
pp. 21-31.
Hoyle, R. J.; Haskell, J. H.; Galligan, W. L.
Characterizing lumber properties for truss research. In:
Proceedings, Metal plate wood truss conference; Forest
Products Research Society; 1979. pp. 32-64.
National Forest Products Association. National design
specification for wood construction. NFPA,
1619 Massachusetts Avenue, Washington, DC; 1982.
Perkins, R. H.; Suddarth, S. K.; Dale, A. C. Rotational
resistance of three-membered nailed joints subjected to
bending moment. Purdue University Research Bulletin
No. 753, Lafayette, IN. 1962.
Quaile, A. T.; Keenan, F. J. Truss plate testing in Canada:
Test procedures and factors affecting strength properties.
In: Proceedings, Metal plate wood truss conference;
Forest Products Research Society; 1979: 105-112.

13
Suddarth, S. K. A computerized wood engineering system:
Purdue plane structures analyzer. Res. Pap. FPL 168.
Madison, WI: U.S. Department of Agriculture, Forest
Service, Forest Products Laboratory; 1972.
Suddarth, S. K.; Wolfe, R. W. Purdue plane structures
analyzer II—A computerized wood engineering system.
Gen. Tech. Rep. FPL-40. Madison, WI: U.S. Department
of Agriculture, Forest Service, Forest Products Laboratory;
1983.
Thurmond, B. M.; Woeste, F. E.; Green D. W. Roof
loads for reliability analysis of lumber properties data.
Wood and Fiber Science. 16(2): 278-297; 1983.
Truss Plate Institute. Design specification for metal plate
connected wood trusses. TPI 78. 1978 TPI, 583 D’Onofrio
Drive, Suite 200, Madison, WI 53719.
Truss Plate institute. Design specification for metal plate
connected parallel chord wood trusses. PCT 80, 1980 TPI,
583 D’Onofrio Drive, Suite 200, Madison, WI 53719.
Zahn, J. J. Strength of lumber under combined bending
and compression. Res. Pap. FPL 391. Madison, WI: U.S.
Department of Agriculture, Forest Service, Forest Products
Laboratory; 1982.
Zahn, J. J. Strength of southern pine 2 x 4 beam columns.
NTIS ADA 143-138; 1984.

14
Appendix
Summary of Truss Member
MOE Measurements

Two different values of MOE were obtained from the CLT Table A2.—Modulus of elasticity for each truss member of the
machine for each piece of lumber. One was the average 6/12 trusses. These values were determined using a CLT
value which is an average of many short span values machine.
measured along the length. The other is called the “low Truss TC1 TC2 TC3 TC4 BC1 BC2 WB1 WB2 WB3 WB4
point” which is the minimum value measured along the No.
length. More details on span length and method of
measuring are available from the equipment manufacturer.1
Prior to truss fabrication, a third value was measured for LOW E
each piece using a flatwise bending test having a center 6L1 1.27 1.26 1.24 1.08 1.29 1.08 1.14 1.35 1.26 1.40
point load on a 12-foot simple span. In the main text of this 6L2 1.03 1.24 1.38 1.31 1.23 1.28 1.14 1.34 1.00 1.17
report, only the CLT average values were used. Tables Al 6L3 1.18 1.17 1.38 1.16 1.31 0.98 1.17 1.25 1.36 1.34
and A2 give these values for each 3/12 and 6/12 truss 6L4 1.30 1.19 1.16 1.21 1.28 1.17 1.28 1.39 1.05 1.34
member respectively. 6L5 1.31 1.26 1.28 1.24 1.11 1.37 1.02 1.40 1.06 1.34
6L6 1.18 1.30 1.30 1.19 1.23 1.12 1.02 1.39 1.36 1.28
Table A3 compares all three values of MOE for each of the 6L7 1.09 1.17 1.19 1.30 1.37 1.38 1.34 1.30 1.05 1.40
MOE categories. This table includes more lumber than was
MEDIUM E
used in manufacturing the trusses, thus the data may not
agree with table 2. 6M1 1.96 1.87 1.79 1.94 1.97 1.71 1.66 1.50 1.70 1.45
6M2 1.92 1.79 1.87 1.89 1.74 1.41 1.61 1.66 1.87 1.62
6M3 1.94 1.58 1.46 1.99 1.46 1.74 1.87 1.85 1.81 1.81
Low point CLT values were about 3/4 of the average CLT 1.87 1.64 1.43 1.93 1.98 1.63 1.93 1.61 1.59 1.68
6M4
values. For the high MOE lumber, the flatwise and CLT 6M5 1.95 1.46 1.46 1.91 1.91 1.59 1.80 1.68 1.93 1.97
average values were quite close. But, for both the low and 6M6 1.93 1.43 1.46 1.98 1.91 1.74 1.70 1.97 1.62 1.85
medium MOE lumber, the flatwise bending MOE was slightly 6M7 1.95 1.43 1.64 1.96 1.54 1.61 1.50 1.60 1.45 1.59
less than the CLT average.
HIGH E
1 6H1 2.14 2.37 2.37 2.38 2.41 2.45 2.25 2.25 2.36 2.36
lrvington-Moore, P.O. Box 23038, Portland, OR 97223. 6H2 2.09 2.18 2.18 2.52 2.31 2.21 2.46 2.36 2.58 2.36
6H3 2.38 2.23 2.23 2.03 2.28 2.07 2.45 2.42 2.46 2.39
6H4 2.07 2.47 2.47 2.05 2.35 2.15 2.42 2.39 2.45 2.48
6H5 2.07 2.04 2.58 2.57 2.50 2.14 2.58 2.58 2.48 2.58
Table A1 .—Modulus of elasticity for each truss member of the 6H6 2.28 2.35 2.19 2.29 2.07 2.02 2.18 2.03 2.39 2.50
3/12 trusses. These values were determined using a CLT ma- 6H7 2.32 2.35 2.19 2.10 2.07 2.20 2.18 2.06 2.44 2.50
chine.

Truss TC1 TC2 TC3 TC4 BC1 BC2 WB1 WB2 WB3 WB4
No.

LOWE
3L1 0.97 1.31 1.33 1.23 1.37 1.26 1.09 1.21 1.23 1.09
3L2 1.30 1.23 1.32 1.26 1.31 1.25 1.32 1.14 1.14 1.11
3L3 1.06 1.03 1.29 1.34 1.17 1.11 1.32 1.34 1.39 1.32
3L4 0.96 1.33 1.39 1.27 1.01 1.35 1.11 1.37 1.34 1.09
3L5 1.24 1.24 1.29 0.97 1.39 1.19 1.29 1.03 1.24 1.11
3L6 1.13 1.32 1.37 1.33 1.23 1.34 1.32 1.25 1.25 1.11
3L7 1.37 1.33 1.33 1.13 1.33 1.15 1.09 1.31 1.21 1.28

MEDIUM E
3M1 1.76 1.69 1.72 1.85 1.55 1.79 1.83 1.89 1.57 1.83
3M2 1.63 1.56 1.89 1.47 1.49 1.41 1.58 1.58 1.58 1.58
3M3 1.60 1.65 1.72 1.87 1.80 1.63 1.52 1.85 1.52 1.52
3M4 1.55 1.74 1.74 1.87 1.59 1.49 1.55 1.54 1.61 1.55
3M5 1.77 1.65 1.69 1.60 1.75 1.93 1.55 1.89 1.90 1.55
3M6 1.46 1.75 1.29 1.42 1.91 1.92 1.52 1.57 1.52 1.52
3M7 1.71 1.56 1.90 1.74 1.81 1.81 1.75 1.54 1.85 1.88

HIGH E
3H1 2.36 2.44 2.54 2.31 2.32 2.25 2.44 2.45 2.50 2.03
3H2 2.40 2.32 2.46 2.41 2.27 2.27 2.18 2.47 2.51 2.44
3H3 2.44 2.32 2.34 2.39 2.25 2.33 2.50 2.45 2.36 2.44
3H4 2.36 2.46 2.34 2.25 2.25 1.81 2.06 2.47 2.36 2.44
3H5 2.26 2.44 2.54 2.34 2.34 2.28 2.44 2.51 2.50 2.39
3H6 2.07 2.04 2.04 2.13 2.02 2.05 2.24 2.15 2.31 2.24
3H7 2.15 2.25 2.25 2.08 2.09 2.07 2.24 2.15 2.31 2.24

15
 Table A3.—Comparison of measured MOE values for lumber used in the fabrication of the
test trusses. CLT values were measured in the mill and flatwise bending values were
measured just prior to truss fabrication.
Continuous lumber tester (CLT)
Approximate Flatwise bending
MOE Average Low-point
category number of
specimens 1 STD 2 STD
Mean COV 2 STD3 Mean COV 2 ERR 3 Mean COV ERR 3
ERR
Million Million Million Million Million Million
lb/in. 2 Pct lb/in. 2 lb/in. 2 Pct lb/in. 2 lb/in. 2 Pct lb/in. 2

Low 165 1.25 9.5 0.009 (4) (4) (4) 1.12 12 0.02
Medium 150 1.74 10.6 0.015 1.26 16.3 0.017 1.65 14.3 0.019
High 120 2.29 6.9 0.014 1.72 11.9 0.019 2.27 9.8 0.01
1
Sample size for various methods of MOE measurement may vary by ±3.
2
Coefficient of Variation for sample.
3
Standard error of the estimate of mean value.
4
Data not obtained on a significant part of the sample.

2.5-3/86

16