Design and Experimental Analysis of NLT Trussed Rafter … · Design and Experimental Analysis of...

Journal of Civil Engineering and Architecture 11 (2017) 16-28 doi: 10.17265/1934-7359/2017.01.003

Design and Experimental Analysis of NLT Trussed

Rafter in Yellow Pine

Emanuelle Graça Recco1, Jorge Daniel de Melo Moura1 and Everaldo Pletz2

1. Center of Technology and Urbanism, Department of Architecture and Urbanism, Londrina State University, Londrina CEP:

86057-970, Paraná, Brazil

2. Center of Technology and Urbanism, Department of Structures, Londrina State University, Londrina CEP: 86057-970, Paraná,

Brazil

Abstract: The roof system of social housing in Brazil generally consists of components made out of native forest lumber of high market value. Taking into account the increasing number of planted forests and the need to develop new products and to add value to this timber, this work deals with the development and structural analysis of a roof system using yellow pine plantation wood (Pinus spp), a sustainable material which however presents many defects. The NLT (laminated nailed timber) technology was chosen as it allows the use of shorter length and smaller cross section pieces, eliminating major defects. Seven samples of structural trussed rafters in NLT were tested; six made out of graded timber and one ungraded in order to verify the impact of the wood grading in the structural performance of the model. The results showed that the trussed rafter system in NLT meets the necessary structural performance requiring poor conditions of infrastructure for manufacture process, and that the graded wood samples showed better performance than the ungraded one.

Key words: Roof systems, yellow pine wood, laminated nailed timber, trussed rafter, design methodology.

1. Introduction

When considering the social circumstances of the

Brazilian housing deficit and the precariousness of

living conditions today, it appears that there is a vast

field of research of manufacturing and use of

prefabricated building systems with plantation lumber.

The construction sector has shown the increasing need

of product development and processes that promote

the reduction of costs while improving the quality of

construction systems and housing.

Plantation wood occupies increasing space in the

market, due to its rapid growth. The good quality

wood is the one that has fewer defects, whether

related to the wood itself or those from the production

chain. Although the yellow pine (Pinus spp) wood

presents good features that make it easy to transport

Corresponding author: Emanuelle Graça Recco, M.Sc.,

research fields: wood construction and wood structures. E-mail: [email protected].

and use for various purposes, it also presents aspects

that deserve further attention during the tree’s growth

to obtain a final product with better quality.

The traditional construction systems in Brazil have

not contributed to overcoming the existing housing

deficit and, consequently, prefabrication is a viable

path to be followed. It is known that in prefabricated

structures, the connections are extremely important.

According to Ref. [1], connections enable the

transportation of the parts, being responsible for the

continuity and the overall behavior of the buildings as

well as being the cause of most problems and

structural flaws.

The trussed rafter system is known and widely used

as a prefabricated subsystem for roofing in European

countries, the United States and Canada. The NLT

(nail laminated timber) is an effective alternative to

enable the use of smaller structural timber pieces,

considering its design and standard requirements.

This study aims to develop an alternative for

D DAVID PUBLISHING

Design and Experimental Analysis of NLT Trussed Rafter in Yellow Pine

17

roofing subsystem facing industrialization, enabling

cost reduction and improvement of the performance of

building components.

2. Literature Review

The main feature of the trussed rafter roof system is

the elimination of traditional trusses, the main girders

and their replacement for rafters working as light

trusses. The grid is only composed of slats bracing the

trusses. They are generally spaced 100 cm from each

other allowing spans ranging from 6 m to 12 m. Such

large spans are possible due to the large number of

light trusses, better distributing the acting forces

among the other structural elements according to Ref.

[2].

The laminated timber consists of timber laminations

connected to one another, keeping the lamination

fibers parallel to the longitudinal axis of the element.

This system allows for obtaining significant structural

elements with a variety of shapes and sizes, which

most often are limited only by transport issues

according to Ref. [3].

Among the structural elements of timber, there are

basically three types of connections: contact

connections, which act only by the contact between

timber pieces, glued connections, made from the

union of timber pieces by the action of an adhesive,

and mechanical connections consisting of the

penetration of a connecting element into the timber

piece. The study of the behavior of these connections

is important as it allows identifying their issues and

finding solutions, making it possible to improve their

load capacity, rigidity and ductility. Because of these

studies, there has been a change or even the creation

of new arrangements and/or connecting elements [1].

In the process of failure of mechanical connections

under monotonic loading, Haller [4] observes four

phases: (1) no-charge displacements due to

manufacturing adjustments; (2) elastic phase: linear

elastic behavior of the wood and the pin; (3) plastic

phase: irreversible displacements due to the

flow/twisting of the metal pin or cracks in the wood

matrix around; (4) complete failure of the connection.

From this observation, changes are proposed in bonds

in order to correct these failures.

Since wood is a fragile material under tension stress,

the connections tend to exhibit the same behavior if

the failure happens in the wood. According to Ref. [4],

the only way to increase the ductility of the

connections in wooden structures is to make use of

mechanical connecting elements that allow the

making up of plastic hinges in the connection before

failure.

Regarding the ductility of wooden structures, when

the connections are very resistant and rigid, Smith and

Foliente [5] stated that the rupture tends to happen in a

sudden and unwanted manner. Thus, the connections

should be chosen and designed in order to act as a

factor responsible for the ductility of the structures

and not aiming only at the transmission of internal

stresses with high strength and stiffness. In fact, the

ductility of the connections is characterized by the

behavior of the stress-strain curve. A ductile behavior

of connections is only possible when the failure

occurs after high plastic deformation.

The elements used in the mechanical connections

are classified into two groups; pin-like elements and

connectors. Nails are the most common pin and can be

found in the market in various shapes and sizes. They

may be fabricated with one or two heads and even

without a head. The shank nails can be of circular

section (smooth nail), square (sailor nail) or

mechanically worked yielding ringed shanks (ring nail)

or helical (ardox nail). They can be coated in order to

increase their resistance to oxidation or axial load [1].

Martin [6] developed research focused on studying

the role of nailing as a means to prevent sliding of

splice connections submitted to bending. The analysis

was done on a two-board nailed beam. A series of

tests allowed the author to define the non-linear

behavior to shear of a nailed connection through the

sliding of the laminations. Three point bending tests


18

on two-laminated beams were performed to evaluate

the nailing connection, in particular, the nailing

density, with three different types of nails: smoothed

rod nail, spiraled rod nail (ardox) and ringed rod nail.

The results showed close performance of ardox and

ringed nail types, and poorer for the smooth nailed

connection.

In addition, five variations of nailing densities (0.89,

1.26, 1.63, 2.00 and 2.49 nails/dm2) were tested with

the density increasing from the center to the edge of

the beam. The author demonstrated that with an

appropriate density of nails, the NLT can provide

rigidity similar to that observed in glulam.

Seeking to develop roofing systems for social

housing at a lower cost and easy building using

plantation lumber, Valle [2] proposed the analysis of

some models. The NLB (nail laminated beam) system

was one of those that allowed for the elimination of the

three traditional structure roof components—the slat,

the rafter and the girder—by a single nail laminated

part. The NLB proposed by Valle [2] considered the

commercial cross sections manufactured in sawmills,

resulting in the use of different length parts, avoiding

joint overlaps and splices in the middle of the span.

The pin-connection element eventually adopted for the

three-layered NLB’s was the smooth nail, longer than

the width of the three layers, having the tip bent, and

thus hampering nail withdrawal. Samples with

different types of nails (smoothed with tip bent, spiral

and ringed) and different spacing were tested.

Although the best results were observed in helical nail

connected beams, Valle [2] opted to use the tip-bent

smoothed nail due to its lower cost. The main feature

needed, (increased pullout strength), would be

achieved by folding the outer tip of the nail. Several

tests were conducted in order to reach the ideal

product. Four variations to NLBs were developed to

fit the rural house design in which the beam was

meant to be applied.

The production process was considered simple,

easily assimilated by the users/builders, and needing

small fabrication infrastructure due to the use of

simple equipment. The model has met the required

conditions to structural performance, durability and

quality, proving the system’s viability.

The nails can resist transverse load, axial load or a

combination of the two according to Ref. [7]. In

nailed connections, under the condition that large

displacements are not tolerated in the connection, the

influence of the resistance to the axial load on the

transverse load resistance can be disregarded. In ring

nails and ardox nails, which have the shafts

mechanically worked, there is an increase in the

contact surface between the nail and the wood without

an increase in nail weight. The axial load resistance of

such nails is always higher than the common smooth

nails with the same diameter. The European standard

Eurocode 5 [8] calls this behavior of metal-pin

connections under large deformations the rope effect.

Unlike the European standard, the Brazilian standard

does not consider the rope effect while designing the

resistance of nailed connections.

The design criterion proposed by NBR 7190/97

(and also various other standards) [9] is based on the

concept that the resistance of metal-pin connections

depends on the resistance of the embedment in the

wood and pin yield strength, this way determining the

behavior of the connection according to Ref. [10].

To propose an alternative roof system in timber

based on building rationalization criteria, it is

important to consider an efficient design method.

In addition to ultimate limit states related to failure,

the Brazilian standard provides for the restriction of

displacements, which prevents the aesthetic

commitment and malfunction of equipment and

facilities, avoiding damage to the finishing materials

or non-structural parts of the building according to

Ref. [11].

3. Method and Materials

3.1 Trussed Rafter Design in NLT

The trussed rafter system in NLT proposed in this


19

research consists of a three-layered yellow pine (Pinus

spp) component; 1 central layer with continuous

laminations and two external layers with

discontinuous laminations. The rafter span is 6 m and

the total height is 1.2 m. All timber of the trusses was

connected with spiral nails (ardox type) 18 × 30 to

overcome the total thickness of the three layers. The

total cross section of the NLT pieces was 6 × 14 cm,

corresponding to the three laminations.

The connections of the truss rafter were executed

according to a design based on the NBR 7190/97 and

Eurocode 5. The connection between the splices were

determined in tension tests performed on models with

20 cm minimum distance between the NLT ends

following a distribution of 9 nails (18 × 30) in a 3 × 3

mesh distanced 2 cm from the lamination edge and 5

cm between each other. This arrangement followed

the minimum spacing recommended for metal-pin

connections specified in the NBR 7190/97, as shown

in Fig. 1.

A number of six samples was adopted to the testing

following the recommendations of the NBR 7190/97

Annex B (though the standard does not address

structural dimension tests except to small specimens

testing). In addition to the six samples with graded

timber recommended by the standard, one test was

also carried out on a sample made out of ungraded

timber, totaling seven full-scale samples.

3.2 Timber

Both for the connections as for rafters testing, kiln

dried yellow pine (Pinus spp) boards averaging 12%

moisture content and approximate dimensions: 2.5 ×

15 cm cross section and 300 cm length were used.

Fig. 1 NLT trussed rafter composition.


20

The pieces were first visually graded according to a

structural visual grading manual for Pinus spp [12],

based on the Brazilian standard NBR 11700-90, the

American visual grading method issued by the SPIB

(Southern Pine Inspection Bureau) and the ASTM

D245-93 [13].

After machining and cutting, all the pieces were

grouped into similar lengths to facilitate the

organization for mechanical grading through

ultrasound NDT (non destructive testing) method.

3.3 Definition of Nail Type and Arrangement of the

Connections between Splices

First of all, tension tests on splices were performed.

Three variations of nailed were tested: smooth nail

(PA), smooth nail with bent tip (PB) and ardox-type

helical nail (PC). The test was performed according to

the NBR 7190-97 Annex C. For determining the

connection stiffness, the timber strength needed to be

estimated (ft0,est—estimated tensile strength along the

grain) by destructive testing of a twin specimen of the

same sample to be investigated. Once the estimated

strength of the sample (ft0,est) was known, two loading

and unloading cycles were applied (10% and 50% of

ft0,est). Displacement gages (LVDTs—linear variable

differential transformer) were installed at the time of

the test in each side of the specimen for

measuring the relative displacement of the

connection.

A testing apparatus had been developed whose

main concern was to keep the ends stronger to ensure

failure in the central section. The configuration of the

full-scale sample was determined by the length

required to keep the ends of the sample stronger than

the central section where the connection to be studied

was located. In addition to the 9 nails arranged

between the splices spaced 20 cm, 12 more nails were

distributed on each side of the sample, totaling 33

nails by sample (Fig. 2).

Stress-strain diagrams were drawn, and found the

strength and rigidity of the splice connection for each

sample according to the NBR 7190/97 specifications.

The analysis of stresses shows that in the central

region of the horizontal rod in the range of 20 cm

between the two splices, the transmission of the axial

force is supplied by two laminations, the central and

one external. A portion of the axial force is transmitted

by the central lamination, due to its continuity, and the

remaining portion of this stress by the nailed

connection that transfers from an external lamination

to another. Thus, by hypothesis, it is assumed that the

nailed connection has, in a cross section, a stress

solicitation equal to half the normal stress existing in

the composite bar as displayed in Fig. 3.

Fig. 2 Apparatus and tension test sample.

Fig. 3 Stress transfer mechanism in the splice connection.


21

The design based on the NBR 7190/97 determines a

strength of 0.35 kN for this connection. The results of

each sample group are shown in Table 1.

The smooth nails led to the lowest connection

strength. This can be attributed to the fact that the

folded tip nails and the ardox-type (coiled rod nail)

provide respectively, support in the folded end and

have greater side adherence to the timber.

As for the stiffness of the connection in the elastic

range, it is also observed that the smooth nails

produce more deformable connections than the other

two types of nails. Nevertheless, they all showed

ductility, a very important feature for the connections

to become more predictable and safe. The bent-tip

nails behave like a statically indeterminate

bi-pinned beam at their ends, and with large

deformations, a development of an extensional

stiffness, as well as flexural rigidity are observed. This

causes the bent-tip nails to outperform the ardox

nailed connection concerning stiffness.

Also noteworthy is the variability of results. The

strength of the connections was more variable than the

stiffness in the elastic range. Smooth nails showed a

more variable performance than the other two types of

nails (25%). In the early stages, the nails still show

low deformability, and therefore, the stiffness of the

connections is less dependent on the timber

heterogeneity. As to the ultimate load, the variability

increased for the three kinds of nails compared to the

stiffness result.

As for the failure mode, these connections initially

showed large deformation, bending the nails, and

subsequently the already plasticized nails, the central

lamination failed. In the failure, the displacements of

the connections correspond to deformations ranging

from 4% to 5%. Those are extremely high values

when compared to the design threshold.

The failure occurred in the connection, except in

one of the samples. The central lamination rupture

was always followed by excessive deformation and

rotation of the nails. There was one case of rupture by

longitudinal shear in the wood with embedment of the

nail heads in the wood, in the case of the folded-tip

nails. In this particular case, the spacing adopted

among nails was about 12 diameters that exceeded the

required minimum of six diameters. Table 1 Stiffness of the connections and resistance per nail.

PA (Smooth nail) PB (Smooth nail with bent tip) PC (Ardox-type nail)

SAMPLE Failure load (kN)

Stiffness of the joints 200 mm (kN/cm)

Resist. per nail for each cutting section (kN)

Failure load (kN)



Failure load (kN)



1 45.00 167 1.94 47.50 235 2.04 45.00 153 1.94

2 44.00 169 1.89 65.00 190 2.80 50.00 175 2.15

3 32.50 (*) (*) 36.00 210 1.55 45.00 190 1.94

4 22.50 118 0.97 30.00 195 1.29 35.00 163 1.51

5 17.50 105 0.75 39.00 175 1.68 30.00 157 1.29

6 22.50 101 0.97 24.00 249 1.03 57.50 193 2.47

Average 30.67 132 1.30 40.25 209 1.73 43.75 172 1.88 SD (standard deviation)

11.78 33 0.57 14.52 28 0.62 9.97 17 0.43

CV (coefficient of variation)

38% 25% 43% 36% 13% 36% 23% 10% 23%

*: PA 3 results were not considered due to problems during test.


22

The final analysis showed that the smooth nail

generally showed the poorest results. The best

performance was observed to the helical (ardox)

nailed connections.

3.4 Criteria for Choosing the Pieces of the Trussed Rafter Samples

Once the arrangement of the connections was

determined and the nail type was chosen, the next step

was to determine the distribution of pieces to compose

the seven samples of structural trussed rafters to be

tested.

With the piece lengths resulting from the visual

grading and the values of dynamic MOE (modulus of

elasticity) determined by nondestructive ultrasound

grading, a spreadsheet with the distribution of pieces

was made. Such distribution was thought out in such a

way that that the samples had similar characteristics,

both geometrical and concerning mechanical

properties. The splices of the outer layers must be at

correspondent locations and have uniform MOE in

order to make all the samples become comparable.

The timber showed great variability of density and

MOE, which made a uniform distribution in the

samples difficult. The grouping and distribution of

pieces were carried out in stages:

(1) Pieces and splices were distributed according to

the lengths originated in the visual grading. Since

most pieces had small lengths (ranging from 40 cm to

70 cm), it was necessary to make a great number of

splices;

(2) Once the lengths were determined and the

splices of the three layers were located, the bottom

chord pieces were set apart. These pieces should have

low MOE variation so that all the bottom chords of all

samples would be comparable, since this is the

component of a truss in which the greatest tension

stresses appear. Thusly organized, the MOE ranged

from 8,000 MPa to 14,000 MPa. The pieces were

distributed in such a way that all bottom chords of all

six samples presented the same MOE average, and

close standard deviation;

(3) The pieces of the upper chords and the webs

(diagonals and posts), showed high variation of MOE,

however, average, standard deviation and coefficient

of variation, were similar. They were distributed

according to the needed lengths.

The underlying idea of this setting was having

samples showing mechanical properties as close as

possible therefore, with similar structural behavior

and thus comparable.

3.5 Layout of Samples

The assembly of the samples was performed in the

model laboratory at the Universidade Estadual de

Londrina—UEL (Londrina State University). The first

trussed rafter was a template for the assembly of the

remaining six copies. For this reason, the setup of the

first sample must be very thorough, especially

concerning the angular cutting to allow for an accurate

fit. Regarding the right-angled end pieces, a precise

orthogonal cut was important, to allow for full contact

between the two adjacent parts.

After the setup of the first sample was finished and

all pieces were embedded, one proceeded to the

cutting of the remaining laminations. The laminations

were end cut and numbered from 1 to 6 according to

MOE grading. Thus, all pieces were ready for set up

and screened for subsequent nailing of the remaining

samples. In order to assist nailing, other templates of

the many different arrangements of connections in

MDF (medium density fiberboard) were developed.

Then pre-drilling was carried out according to NBR

7190/97 (minimum diameter of 0.85def; (def being

the actual diameter in mm), as shown in Fig. 4.

3.6 Apparatus and Test Instrumentation

The structural testing was performed by means of a

reaction porch in the Laboratory of Structures of UEL.

The sample was lifted, set up in the connected porch

and braced to guarantee lateral stability during test

(Fig. 5).


23

Fig. 4 Setup of samples, assembly template, and pre-drilling.

Fig. 5 Testing apparatus.

The Brazilian standard prescribes pinned supports.

Therefore, a circular metal rod was placed between

the sample end and the porch to allow displacement

and rotation of the sample at the time of load

application.

The loading was applied by a 300 kN central

actuator and transmitted to the sample through steel

cables as shown in Fig. 5.

To ensure that the loads on the steel cables were

evenly distributed, a load application system was set.

It was composed by small wooden joists and metal

beams that supported in eight points a greater steel

beam wrapped around by cables. On the top chords, a

metal beam was responsible for distributing the load.

On it, a wooden beam fixed the cables as Fig. 5

shows.

Before the test, the apparatus (all metal and wood

beams) was weighed with the aid of a 0.1 kg precision

scale. The weight of the whole set was 2,160 kN. This

total weight was added to the ultimate load and the

measurement of displacements during the test was

made from this initial threshold.

The load application was carried out by an Enerpac

brand hydraulic actuator, model RCH306 30 t

capacity.

The vertical displacements were measured by

means of three dial gages (Mitutoyo brand, 3058S-19

code, 0~50 mm capacity and 0.01 mm precision), one

below the central post and two below the side posts.

3.7 Loading Test

The Brazilian standard does not address structural

testing. Thus, the test followed the procedures adopted

for small specimens described in the Annex B of NBR

7190/97.

The estimated tensile strength along the grain ft0,est,

was equal to 50 kN. The loading was applied in two

loading and unloading cycles, 10% and 50% of the


24

estimated resistance. The third and last cycle led to

failure. The load was monotonically increasing with

an approximate rate of 6 MPa per minute.

4. Results and Discussions

Fig. 6 shows the evolution of the loading versus the

testing time (Sample 2) after 47.35 kN loading level.

One can observe a significant level of plasticization

with small falls on loading. After these, a gradual load

increase is still observed until the final failure. This

behavior was common to all samples and demonstrates

partial failures before reaching the ultimate load.

The strength results can be seen in Table 2. It is

possible to observe the high failure load values of all

samples averaging 60 kN and low variability of results

(COV (coefficient of variation) equals to 7.5%).

Fig. 6 Load Evolution at the end of the third cycle (after 47.35 kN loading level) on Sample 2.

Table 2 Failure to all samples.

Sample Ultimate load (kN)

1 56.38

2 65.41

3 51.86

4 63.15

5 60.89

6 58.18

Ungraded 61.80

Average 60.89

SD 4.56

CV 7.5%

Fig. 7 Displacement of the splices during the test and after failure.

(a)

(b)

Fig. 8 Loading simulation on the trussed rafter: (a) bending moments diagram (in kNm); (b) normal stresses diagram (in kNm) under 60 kN loading.

As expected, all samples failed in the connection of

the bottom chord due to tensile stresses. After that, the

top chord failed in compression as indicated in Fig. 7.

The rupture occurred at the splice nearest to the side

posts shown in Fig. 7

In the diagram of bending moments, as shown in

Fig. 8a, one can observe that although small, there is a

bending moment equal to 0.1 kNm in that location.

Initially, it was thought that the tensile stress would be

the same along the bottom chord bar and this moment

would not be as significant. However, after testing, it

was possible to observe that there was actually an

interference of the bending moment in the failure of

the bottom chord.

The results obtained in the tests of seven samples

can be compared to the results obtained in the test of

splice tensile connection specimens. The numerical

simulation using the Ftool 1 software with the

application average failure load obtained in seven

samples (about 60 kN) was performed in order to

assess the efforts at the connections of the tensioned

bar. The average failure stress observed in the

connection specimens was 43.75 kN (Table 1). The

stress recorded in the tensioned bar of the trusses

(bottom chord) was 83 kN, approximately twice as

great, showing satisfactory results (Fig. 8b).

As for the failure mode, it has been found that after

excessive deformation of the nails, failure occurred in

1www.tecgraf.puc-rio.br/ftool.


25

two ways, as was found in testing connection specimens:

(1) By crushing the wood matrix around the nails

with the embedment of the heads and loosening the

layers of the NLT;

(2) The longitudinal shear of the outer layer and/or

core layer, as Fig. 9 displays.

Some changes in the connection arrangement

among the splices could be made in order to avoid or

decrease the embedment of the nail heads and

consequent longitudinal shear of layers. Alternating

the distribution of the nails or the interchanging the

nailing sense, (nailing half of the nails on the opposite

face) could be one of those.

The comparative graph was prepared to display

load versus displacement curves in the center of the

span of the seven samples at the loading 47.35 kN. All

the curves showed the same behavior, with a

reduction of displacement after the 1st cycle

(accommodation of the structure), about 25 kN 50%

of the estimated tensile strength (Fig. 10).

Through Fig. 10, it is possible to obtain the rigidity

coefficient (K) in the elastic deformation phase,

according to Eq. (1): K = ΔP/Δδ (1)

where:

ΔP = Variation of the applied load;

Δδ = variation of deflection.

There is a similarity in the shape of curves; a

bilinear feature present in all. In the elastic range, up

to approximately 25 kN (50% of actual load), the

slope of the curve is different from the second cycle

that leads to failure. This inclination difference

indicates that there is greater rigidity in the first cycle

compared to the second after the accommodation

phase.

The K values (coefficient of stiffness) of each

sample for two cycles are shown in Table 3. They

show that in general the samples showed greater

stiffness in the first cycle.

All samples showed close K values, with low

coefficient of variation. It should be noted that the

ungraded timber sample showed the lowest value of K

in the first cycle, close to sample number 6. In the

Fig. 9 Modes of failure.

Fig. 10 Curve load X deflection the center of the span (to load 47.35 kN).


26

Table 3 Stiffness coefficient (K)—first and second cycles.

Specimen reference

First cycle Second cycle First cycle Second cycle

ΔP 1 (kN) Δδ 1 (m) ΔP 2 (kN) Δδ 2 (m) K1 (kN/m) K2 (kN/m)

1 23 0.019 24 0.042 1,211 571

2 23 0.015 24 0.031 1,533 774

3 23 0.017 24 0.031 1,353 774

4 23 0.019 24 0.028 1,211 857

5 23 0.015 24 0.028 1,533 857

6 23 0.022 24 0.038 1,045 632 Ungraded timber specimen

23 0.023 24 0.038 1,000 632

Average 1269 728

SD 214 116

CV 5.91% 6.27%

Fig. 11 Deflection curves to the seven samples—service loading (6 kN).

Fig. 12 Load (kN) when reaching the service state limit (L/200 and L/300).

second cycle, Sample 1 showed the lowest K value

and the ungraded timber sample resembled Sample 6

staying among the lowest values and thus showing

lower stiffness.

To represent the global behavior of the truss, vertical

displacement was measured through gages placed

underneath the three posts.

The NBR 7190/97 determines that the service

deflection must be less than L/200 and the revision

text of the same standard, proposes that it should be

less than L/300, L being the total span. Since the total

span considered is 6,000 mm, the limit deflection to

the application of the service loading should be less

than or equal to 30 mm and 20 mm (text under

revision). In Fig. 11, one can see that the deflections

to all samples did not exceed 14 mm.


27

Sample 1 showed the lowest deflection, less than

5 mm and the ungraded timber sample showed the

highest deflections.

In all cases, the service state limit was the

deflection limit. Fig. 12 shows the load under which

the deflections reached 30 mm (L/200) and

20 mm (L/300) (former and new standard

limitations, respectively). To the 30 mm

deflection, the load ranged from 27 to 35 kN; and the

to 20 mm deflection, the load ranged from 22 to

27 kN.

The weakest point of the structure in terms of

failure is the connections and not the wood, since all

samples displayed highly elevated failure load, and

low variability. However, concerning stiffness, the

graded timber samples performed better than the

others.

5. Conclusions

This paper presented the analysis of seven samples

of trussed rafters made out of yellow pine nail

laminated timber-NLT, and the following conclusions

could be drawn:

The trussed rafters are widely used as roof structures

in the US, Canada and some European countries, that is,

its efficiency is proven. The fabrication of structural

elements in NLT proved to be a viable option

considering the use of pine plantation timber, which

features lots of natural defects.

Before the test of the structure, the experimental

assessment of splices in NLT tensioned bars was

performed in order to check their capacity, and to

determine the nail type to be used.

The tensile test showed that concerning splices in

NLT beams, it is necessary to ensure a minimum

spacing between two consecutive splices of two blades,

determined from the applied load. The experimental

results showed that the three types of nailed

connections are different and therefore require specific

mathematical equations unlike the only model proposed

by the standard NBR 7190/97.

Connections with smooth nails showed the worst

performance on both resistance and rigidity.

Connections with nails with bent tips showed the best

results of rigidity, while connections with ardox nail

achieved the best results with respect to resistance.

These latest two connection types showed results

close to each other and above the smooth nailed

connections.

The connections presented the necessary ductility

and for testing of the full-scale trussed rafter samples

the ardox type nail was chosen due to its higher

resistance and lower coefficients of variation.

Seven samples of trussed rafters were tested, six

fabricated with graded timber (as required by the NBR

7190/97) and one with ungraded timber. All samples

showed similar behavior during the test and the failure

occurred in the tensioned bar of the bottom chord

close to the side post. The tensile strength results were

quite high, with average of 60 kN, and low coefficient

of variation.

With the application of service load (6 kN),

deflections were below 1.4 cm lower than that required

by the Brazilian standard (30 mm current requirements

and 20 mm to the new under-revision text respectively).

The service state limit occurred under the load ranging

from 27 to 35 kN for 30 mm deflection and from 22 to

27 kN to 20 mm deflection.

The deflections were higher in the ungraded timber

sample, showing that grading benefits the performance

of the structure. However, in this case, it did not

invalidate the structure since the load values, when

reaching the service state limit, were above that which

is expected and necessary for the proper functioning of

the system.

The results of failure showed the system is highly

dependent on the quality of the connections. Regarding

the rigidity, the quality of wood had greater

influence, since the ungraded sample showed higher

deflections.

The study of tensioned connection with dimension

timber was very helpful and led to improvements in the


28

connecting system of the splices of NLT beam, and

consequently allowed enhancement in the

stiffness performance and strength of the NLT trussed

rafter.

The results could also confirm that the provision of

an adequate number of nails is able to convey

effectively the stresses and significantly increase the

performance of the structure.

The trussed rafter focused on here, achieved a

performance highly above the requirement (10 times

the service load), allowing resizing parts with

consequent saving of timber and nails.

The trussed rafter developed in this study was

supposed to be mostly applied to social housing

however, due to the excellent performance observed, its

application to several situations in civil construction is

possible.

Acknowledgments

We are very grateful to the Fundação Araucária for

the scientific and technological development of Paraná

state for funding this research.

References

[1] Duarte, R. S. 2004. “Avaliação do Comportamento de Ligações com Parafusos Auto-Atarraxantes em Vigas de MLCC.” M.Sc. Dissertation, Universidade Federal de Minas Gerais, Belo Horizonte. (in Portuguese)

[2] Valle, I. M. R. 2011. “A Pré-fabricação de dois Sistemas de Cobertura com Madeira de Florestas Plantadas: Estudos de Casos: Os Assentamentos Rurais Pirituba II e Sepé Tiaraju.” Dr. thesis, São Carlos: Escola de Engenharia de São Carlos, USP. (in Portugese)

[3] Bono, C. T. 1996. “Madeira Laminada Colada na Arquitetura: Sistematização de obras Executadas no Brasil.” M.Sc. dissertation, Departamento de Arquitetura, EESC/USP. (in Portuguese)

[4] Haller, P. 1998. “Progress in Timber Joint Development and Modeling.” In Proceedings of the 5th World Conference on Timber Engineering, 337-44.

[5] Smith, I., and Foliente, G. 2002. “Load and Resistance Factor Design of Timber Joints: International Practice and Future Direction.” Journal of Structural Engineering 128 (1), 48p.

[6] Martin, P. 2006. “Etude du Comportement des Poutres Lamell´ees Clou’ees Boulonnées en Flexion.” Dr. thesis, ENGREF (Agro Paris Tech). (in French)

[7] Forest Products Laboratory. 1999. Wood Handbook: Wood as an Engineering Material. General technical report (FPL-GTR-113). Madison.

[8] EN 1995-1-1:2004. Eurocode 5: Design of Timber Structures—Part 1.1: General Rules and Rules for Buildings. European Committee for Standardization, Brussels, Belgium.

[9] Associação Brasileira de Normas Técnicas, ABNT. 1997. NBR 7190—Projeto de Estruturas de Madeira. Rio de Janeiro: ABNT. (in Portuguese)

[10] Oliveira, M. A. M. 2001. “Ligações com Pinos Metálicos em Estruturas de Madeira.” M.Sc. Dissertation, Escola de Engenharia de São Carlos, Universidade de São Paulo, São Carlos. (in Portuguese)

[11] Calil, J. R. C., Rocco Lahr, F. A., and Dias, A. A. 2003. Dimensionamento de Elementos Estruturais de Madeira. 1st ed. Barueri: Editora Manole Ltda. (in Portuguese)

[12] Moura, J. D. M., Pletz, E., and Recco, E. G. 2012. Qualidade e Processo Produtivo da Madeira para Utilização em Mobiliário. Londrina: Universidade Estadual de Londrina. (in Portuguese)

[13] American Society for Testing Materials. 2002. ASTM D 245—Standard Test Methods for Mechanically Properties of Lumber and Wood-Base Structural Material. Philadelphia: ASTM.

Design and Experimental Analysis of NLT Trussed Rafter … · Design and Experimental Analysis of...

Documents

Transcript of Design and Experimental Analysis of NLT Trussed Rafter … · Design and Experimental Analysis of...