Experimental Analysis of the Performance of Doweled Connections Reinforced with Glass-Fiber-Reinforced Polymer (GFRP) in Wood Pinus spp.

Abstract

In line with technological advancements, the construction industry worldwide has sought more efficient building systems in relation to aspects such as increased productivity, reduced material waste and meeting the growing demand. The objective of this research was to use structural joints composed of composite dowels in laminated wood beams as an alternative to connect pieces of wood. Composite materials are composed of a matrix phase and a reinforcement phase and, in civil engineering, are generally applied as reinforcements in concrete structures. This article presents the structural performance of laminated pine timber with composite dowels made of fiber-reinforced polymeric resin (epoxy resin, hardener and glass fiber) (glass-fiber-reinforced polymer, GFRP) with a diameter of 12.5 mm, which was subjected to tensile force in the direction of the connection. For this, an experimental program was carried out that included characterization of the GFRP dowel, characterization of the woods used to make the connection and a tensile test of the connections with the dowels reinforced with fiberglass through a prototype designed specifically for the test. Subsequently, the results were compared with those observed in the literature. In the comparisons, it was possible to conclude that the joints with FRP pins exhibited better performance in terms of shear strength per section than those such as common nails, helical nails (Ardox) and composite structural pins (half-lap, 90°), which were 3.8 mm, 3.4 mm and 6 mm in diameter, respectively. This indicates that this composite material has potential for application in these types of connections. As an original contribution, it proves the feasibility of using this material in dowel joints for wooden structures.

1. Introduction

According to technological advancements, the construction industry worldwide has been looking for more efficient building systems regarding aspects such as increasing productivity, decreasing material waste and meeting increasing demand.

In accordance to [1], wood is a material that has advantages in relation to ease of manufacturing parts, as well as technological developments that have been increasingly explored. Other factors, as suggested by the research of [2,3], are the advantages obtained regarding the consumption of energy and the carbon incorporated in wood and engineered wood when compared with materials such as steel and concrete.

In this context, wood structures are a sustainable solution with great importance in a global context. However, the gradually increasing shortage of wood originally from native forests and its effect on the economy, as well as the strengthening of environmental preservation, have raised the need for the development of sustainable alternatives from cultivated forests [4].

In Brazil, the soil and the climate (light and humidity) are favorable for the growth of forests, such as pine species. However, as a result of their fast growth, the wood originating from these forests exhibits natural flaws. As a result, grading the lumber is crucial before its application in structures [5]. According to [6], visual defects influence the wood’s mechanical properties. Variations found in the modulus of the rupture (MOR) of the compressive strength parallel to the fibers can reach 25.3%, and in terms of stiffness (MOE), the difference was 19.7%.

According to [7], the use of plantation wood in structures in a roof, for instance, may lead to significant savings in terms of materials and cost reductions for the entire structure. However, as this wood has inferior mechanical properties compared with the traditionally used tropical wood species (peroba, cumaru, etc.), adding technology for its utilization is crucial. Consequently, the application of this lumber in structures adds value to this widely available raw material.

Yellow pine wood includes over 100 species. Even though Pinus spp. has been cultivated in Brazil for over a century, initially for ornamental purposes, it has been cultivated for commercial purposes only since the 1960s, mainly for pulp and paper, and secondarily for sawn timber, as well as for resin production. According to [8], the most cultivated species in Brazil are Pinus taeda and Pinus elliottii due to their fast growth and adaptation to the local climate. Brazil’s southern and southeastern regions have the most significant planted area. Regarding the mechanical properties, the research carried out by [9] showed results that suggest a modulus of rupture ranging from 53.14 MPa to 79.97 MPa in cultivated wood aged 9, 13 and 20 years, showing that the material meets the calculated needs of wooden structures.

In wooden structures, the connections are often the weakest points and therefore need special attention. The literature has pointed out [10] that there is a lack of information regarding the behavior of wooden dowel connections in terms of their analytical expressions, as well as numerical models. The mechanical behavior of the binding element and the wood depends on several factors, including the geometric and mechanical properties of the fastener components, the mechanical properties of the wood and the interaction between both elements [11]. The studies by [12,13] presented the types of connections used in wooden structures, including connections through metal pins, subjected to double or single shear stress. However, this type of connection may present disadvantages, such as pathological consequences resulting from the presence of moisture [14].

A structural composite combines the properties of two or more materials and consists of two phases: the matrix and the reinforcement. Fiber-reinforced polymeric composites have high mechanical strength and hardness in relation to the weight of the material. As for the type of fiber, They can be solid or broken fibers, presenting an alignment parallel to the longitudinal axis of the fiber or otherwise [15].

The research carried out by [16,17,18,19] described the techniques used for evaluating defects in composite materials. This allows for better use of the material. Glass fiber has advantages such as wide availability in the market at a low cost and, when associated with a polymeric matrix, it can produce a high-resistance composite.

Among the various types of applications for glass-fiber-reinforced polymers, the research developed by [20] presents studies of the behavior of dowels made with polymeric composites as an alternative for replacing lag-screw connections.

This research article presents the characterization of polymeric composite dowels reinforced with fiberglass (GFRP, glass-fiber-reinforced polymer) wood connections submitted to double-cut loading. This article presents the results from the development of a test program, in which the joints were exposed to traction in a line and in the direction of the tension force, and it was possible to analyze the resistance of the last connection, as well as the modulus of rupture, when compared with the results found in similar research. The samples were loaded under tension, and the ultimate strength and the stiffness were recorded, and then the results were compared with the literature.

2. Materials and Methods

The test program was carried out at the Structures Laboratory of the Londrina State University (UEL), and it included the stages describing the structural composite dowelled joints as well as those developed in the research of [20]. These stages were obtained from a performance test of direct tension; later, the performance of tension in a line and in the direction in which the force was applied was also tested.

2.1. Characterization of the GFRP Dowels

The manufacture of the dowel specimens and the loading test were guided by the provisions of the ASTM [21]. The standard allows the measurement of the geometrical features of the dowel in order to perform the test. The dowel’s cross-section area showed varying diameters in the central section and edges. Figure 1 shows the specimen’s dimensions.

The dowels consisted of a polymeric matrix made of low-viscosity, mature, room-temperature epoxy resin (Araldite^® Ly 1564BR epoxy) and a hardener (Aradur^® 2963). The reinforcement was Advantex E-CR GLASS fiberglass, provided by Owens Corning. In total, 11 dowels were prepared through the pultrusion process, submitted to curing at room temperature and taken out of the mold after 24 h. Figure 2 shows the specimens.

The number of specimens was according to [21]. For each specimen, the longitudinal and transversal deformation parameters, the ultimate tensile strength and the modulus of elasticity were measured, and Poisson’s ratio was calculated.

The proportion by volume used in the mix for the manufacture of the dowels was 37% resin and 63% reinforced glass fiber, as indicated by the fiberglass manufacturer Advantex E-CR GLASS (2011) in the technical manual. For a dowel with the dimensions of 195 mm in length, 6 mm in diameter at the ends and 4 mm in diameter at the center, the total volume (3.705 cm³) was filled with 1.65 g of Araldite Ly 1564BR, 0.72 g of Aradur 2963 and 4.04 g of fiberglass. The dowel’s total weight was recorded, and the dimensions were measured. Eleven specimens were produced (Specimens 1 to 11), and six of them were selected through visual grading for the tension test (Specimens 5 to 10). The test followed the provisions of [21]. A tension force with a controlled loading rate was applied in order to simultaneously measure the tangential and longitudinal deformations with the help of strain gages (brand KFG-5-120-C1-11, Kyowa, Osaka, Japan). Each specimen was lightly sanded so that the strain gauges adhered more easily. Markings were made to centralize the gauges in the specimens. Before being glued, the strain gages were tested with a Minipa multimeter ET3021 to detect possible defects, which did not occur. The precision adhesive used for bonding the strain gages was Super Bonder (Loctite brand). This adhesive is based on cyanoacrylate, and its main characteristic is its anaerobic curing in the absence of air. It hardens within a brief time from 10 to 30 s, and must be applied with uniform pressure over the whole area of the strain gage. After bonding, the strain gages were tested again with a multimeter, which confirmed the resistance of 120 ohms, according to the technical data sheet on the package.

The test was performed with a hydraulic universal press with a 30,000 kN load capacity (brand EMIC DL 30000) equipped with a pair of wedge grips, as presented in Figure 3. The data acquisition system used was Lynx AqDados 7.2.

2.2. Wood Characterization

To test the specimen, a grading protocol of the wood pieces was performed. As indicated by the literature [13] at least six specimens were prepared for each connection type. Twenty Pinus spp. wood boards with dimensions measuring 20 mm thick, 300 mm wide and 3000 mm long, and showing a 12% average moisture content (MC) were purchased in the market of Londrina, PR, Brazil.

After measurement of the dimensions and weighing, the wood pieces were visually graded according to [13]. Afterwards, they were submitted to mechanical grading to determine the effective modulus of elasticity (MOE) as indicated by the same standard. The four-point bending test procedure consisted of applying a 500 N load to the thirds of the boards (20 mm in thickness, 140 mm in width and 3000 mm in length), according to Figure 4.

The values of MOE were submitted to a statistical analysis with a confidence interval of 95%, with the aim of including the whole group of specimens with a similar average MOE and standard deviation.

The setup was designed so that the board with the lowest modulus of elasticity was part of the same specimen as the board with the highest modulus of elasticity, and so on. For the manufacture of the specimens, the method of [22] was adapted, in which the average dimensions of the specimens were 60 mm thick, 140 mm wide and 1500 mm long. Each specimen was composed of five pieces of 20 mm in thickness, 140 mm in width and various lengths, with the external layers of one piece being 650 mm and another being 850 mm long, and a continuous middle (internal) layer 1500 mm long, configured across a 330 mm region of the splice. In this cutting process, the regions containing defects and knots were discarded, according to Figure 5.

2.3. Tested Connection Specimen

The connection used to perform the test was based on the research of [22]. Nine dowels were inserted into pre-drilled holes, with no fastening material between the dowels and the timber pieces. The process of analysis consisted of checking the transmission of effort in the splicing area as well as the rupture mechanism, submitting the connection to shear stress with a double shear section. The dowels’ positioning and the minimum space between the connective elements were according to the directions of [13], with the splice length set to 330 mm. Figure 6 shows the elements’ dimensions and the dowels’ positions. In the figure, it is possible to see that the splicing length of 330 mm consisted of three layers: a continuous inner layer and discontinuous outer layers. Nine GFRP 12 mm diameter dowels were inserted in the splicing area as shown in Figure 6.

To perform the tension test, an apparatus simulating the connection’s performance when submitted to tension was conceived, as shown in Figure 7.

2.4. Calculation of the Connection’s Stiffness

After determining the theoretical resistance of each connection, the instantaneous slip module found (k_ser) (kN/mm) for the connection was displayed according to [23] through a linear regression analysis. Equations (1) and (2) present the formulae for calculating the characteristics of resistance and slip. Equations (1) and (2) present the formulae for calculating the characteristics of the strength and slip displacement.

$${\mathrm{K}}_{\mathrm{ser}}=0.4 {\mathrm{F}}_{\mathrm{est}}/{\mathsf{\upsilon }}_{\mathrm{i},\mathrm{m}\mathrm{o}\mathrm{d}}$$

(1)

where F_est is the maximum estimated load, ${\mathsf{\upsilon }}_{\mathrm{i},\mathrm{m}\mathrm{o}\mathrm{d}}$ is the initial slip.

$${\mathsf{\upsilon }}_{\mathrm{i},\mathrm{m}\mathrm{o}\mathrm{d}}=4/3({\mathsf{\upsilon }}_{\mathrm{0,4}}-{\mathsf{\upsilon }}_{\mathrm{0,1}})$$

(2)

where ${\mathsf{\upsilon }}_{\mathrm{0,4}}$ is the slip at 0.4 F_est and ${\mathsf{\upsilon }}_{\mathrm{0,1}}$ is the slip at 0.1 F_est.

3. Results

Figure 8 shows the load versus the longitudinal and transversal deformation observed during the test.

Figure 8a,b above shows the linear behavior with constantly declining lines for both longitudinal and transversal deformation with the stress values $\mathsf{\sigma }=140$ MPa and $\mathsf{\tau }=140$ MPa. From the test results, it was possible to determine the features of the dowels made according to a composite formulation (resin + fibers), which showed isotropic behavior.

Table 1. Behavior of the structural composite dowel.

Properties	Structural Features of the Dowel
Ultimate normal strength (σ)	320 MPa
Longitudinal deformation (ε1)	0.00228
Cross-sectional deformation (ε2)	0.00044
Poisson ratio (ν)	0.195
Longitudinal modulus of elasticity (E)	61,403.5 MPa
Transverse modulus of elasticity (G)	25,691.8 MPa
Shear stress ($\mathsf{\tau }$)	14.4 MPa

shows the features of a structural composite dowel.

The NDT (non-destructive tests) indicated that the boards from which the connection components were sawn were of Class C20, and it should be noted that the greater the strength grade of the wood, the greater the probability of a rupture occurring in the dowels (ultimate strength, 20 MPa; MOE, 5.000 MPa).

Table 2. Modulus of elasticity (MOE) of the specimens (Pinus spp.).

Modulus of Elasticity of the Connection Pieces
IC 95%; Ec0 = 5492.19 MPa–Ec0 = 6695.15 MPa
Piece Reference	Ec0 (MPa)	Piece Reference	Ec0 (MPa)	Average Ec0 (MPa)
L08	4040.44	L17	8878.15	6359.29
L02	4418.14	L23	8557.19	6487.66
L14	4473.72	L06	8208.25	6340.99
L01	4692.05	L18	8050.00	6371.03
L04	4787.96	L24	7705.64	6246.80
L10	4801.14	L19	7446.18	6123.66
L07	4858.28	L11	7086.68	5972.48
L13	5005.07	L12	6743.05	5874.06
L09	5195.28	L15	6415.38	5805.33
L21	5349.68	L22	6342.58	5846.13
L05	5438.28	L16	6182.16	5810.22

shows the homogeneity pairs and their corresponding modulus of elasticity—MOE—observed to the connection components.

Figure 9 shows the curve of a typical connection’s behavior under tension. The resistance of the connection was obtained by tracing a parallel straight line 0.2% away from the secant line, according to [13].

Table 3. Resistance and displacement.

Specimen	Ultimate Load of the Connection (kN)	Displacement (mm)	EN 26891 (1991) Kser (kN/mm)	Linear Regression Kser (kN/mm)
SP4	73.70	2.23	38.50	41.67
SP7	74.50	1.48	35.93	58.91
SP10	51.00	3.03	29.59	16.62
SP11	28.80	1.30	32.58	25.99
SP12	34.80	2.68	13.19	14.34
SP14	75.40	1.67	27.70	38.54
Average	56.37	2.07	29.58	32.68
Standard deviation (STD)	21.19	0.70	16.06	16.98
Coefficient of variation (COV)	37.60%	33.69%	47.05%	51.97%

shows the complete test results. Only SP11 presented a low resistance value. All the other recorded values ranged from 28.80 to 75.40 kN, indicating the excellent representation of the tests. Table 3 also presents the instantaneous slip modulus found (k_ser) (kN/mm) for the connection according to [23] through linear regression analysis.

When analyzing the results, we noted that SP 11 showed a low value (28.8 kN); the reason for this result can be explained by the resistance of the board used for making the connection, where parts were cut close to the tree’s pith. During the evaluation of the mode of failure, according to Figure 10, it was observed that all of them broke in the timber, not due to the shear of the dowels, thus increasing the coefficient of variation of the results. According to Table 3, the average value of the maximum displacement was 2.07 mm. In an analysis of the average values of the resistances, not considering the performance of SP 11, the values of the other specimen showed quite high results.

The analysis of the instantaneous slip modulus (k_ser) demonstrated the variability of the results as well as the linear regression obtained in the test.

It was interesting to observe that the calculated values, in general, underestimated those obtained in the tests regarding stiffness (linear regression); therefore, the parameters of the standard [23] can be considered to be quite conservative.

The observed failure indicated evidence of the great stiffness of the connection, which caused the failure of the wood. Figure 10 shows the mode of rupture.

In all the specimens, the rupture occurred in the wood around the splice region. Ruptures due to the longitudinal shear of the wood at the cutting line of the dowels happened in four specimens (SP4, SP11, SP12 and SP14). In two others (SP7 and SP10), the failure occurred in the transversal direction.

Table 4. Comparison of the results.

Designation	Branco (2003) Smooth Common Nail (90°)	Almeida (2019) Half-Lap Composite Dowel	Barcarolo (2019) Composite Dowel	Recco (2015) Coiled Shank (Ardox) Nail
Doweled connection, F (kN)	3.51	13.18	56.37	33.84
Displacement max. (mm)	14.03	14.73	2.07	--
Number of connectors	2	4	9	9
Diameter of connectors (mm)	3.8	6	12	3.4
Length of connectors (mm)	100	100	60	72
Resistance per connector (kN)	0.88	1.65	3.13	1.88

presents the individual resistance capacity of the dowels per cross-section according to the literature [24]. Compared with the published data, it is possible to observe that the dowel connections developed in this research showed higher performance than the smooth shank nail, half-laps and coiled shank (Ardox) nail connections. In their work, the authors of [22] reported that the nails were nailed perpendicularly to the components’ surface and were manufactured with the same specimen type and the same splice connection arrangement (nine nails distributed in three rows and three columns) as [24], and were subjected to shear stress in a tensile test of the splices. Others [25] studied different angles of nailing subjected to shear stress, as in the research of [24]. The authors of [26] analyzed two types of composite dowel connections (half-lap, 90°; edge, 45°) in CLT panels, assembled with the same adhesive as used by [20] (Araldite^® Ly 1564BR epoxy adhesive, Aradur^® 2963 hardener and Advantex E-CR GLASS fiberglass) in CLT panels subjected to shear stress.

In Table 4, it is possible to see that the resistance of the connection studied here showed 66% higher resistance per connector pin than that obtained by the same connection made with Ardox coil shanked nails.

4. Conclusions

Based on the experimental program carried out, it is possible to draw conclusions regarding the characterization of the dowels as well as the resistance of the connection with GFRP dowels.

Pultrusion was a suitable dowel production process, as it allowed complete soaking of the fibers, and the visual aspects of the dowel proved to be related to its performance.

The characterization tests of the dowels allowed the calculation of the tension and deformation parameters.

The stiffness value obtained in the test (linear regression) was 11% higher than that calculated according to literature reports.

The theoretical values of resistance calculated according to standard provisions, in general, underestimated those observed in the tests and suggest the need for more studies on dowel connections made from structural composite GFRP. Dowels made of a polymer reinforced with fiberglass can be a good alternative to wooden connections, but it is important to carry out further research on various dimensions, classes of timber strength and the positioning of the dowels in the connection.

When we compared the results with similar research in the literature, the dowel connections showed better performance in terms of resistance per cutting section, as it was 3.13 kN higher than those found for smooth shanked nails 3.8 mm in diameter (0.88 kN), Ardox nails 3.4 mm in diameter (1.88 kN) and prefabricated structural composite dowel connections 6 mm in diameter (1.65 kN).

Regarding the mode of rupture, no dowels failed; in all specimens, ruptures occurred due to longitudinal or transverse shear in the timber pieces in the line of the dowels.