Nurdan C?etin Yerlikaya?Alaattin Aktas?

Compressive failure of spruce wood rings reinforced with glass epoxy composite

Nurdan C?etin Yerlikaya1?Alaattin Aktas?2

We experimentally tested under radial compressive loads and statistically analyzed rings constructed from spruce wood and reinforced with glass fiber.We used the Weibull distribution in statistical analysis,and tested five types of rings including unreinforced and composite reinforced(CR)as wound around the ring,oriented as two layers atangles of 45°,60°,75°and 90°to the column axis. We calculated 95%reliability of load carrying capacity of the rings by Weibulldistribution.The highestload carrying capacity was obtained with CR rings at60°to the axialaxis of the ring.Load carrying capacities of rings at CR90, CR75,CR60 and CR45 were 137,192,215 and 126% greater,respectively,than unreinforced rings.For unreinforced rings,failures resulted from catastrophic breaking of wood materials.None of the reinforced rings failed catastrophically because the outer surface of the rings was reinforced with glass–epoxy composite fiber.Cracks began at the core of the materials under the composite layer for all specimens and resulted in failure of the rings.

Reinforced ring·Composite material· Composite reinforced·Spruce wood

Introduction

Haller(2007)developed and patented a procedure for manufacturing wooden profiles.Circular hollow sections perform well when subjected to axial forces so they are well suited for use as columns(Heiduschke et al.2008). Fiber-reinforced plastic(FRP)glued to the outer surface of the profile can strengthen the wood in a transverse direction and prevent the wood from splitting.A wooden core can eliminate local buckling effects and strengthen the FRP profile in an axial direction.In addition,the wooden core stiffens the compound section and prevents thin composite layers from buckling(Cabrero et al.2010a;Heiduschke et al.2008;Haller 2007).

Wood benefits from the mechanical characteristics of FRP.Wood profiles are well suited for use in light-weight structures,the classic field of FRP composites.Furthermore,the orientation of the fiber reinforcement copes with the anisotropy of wood and preserves itagainstweathering. The fiber or textile reinforcement benefits from the low price of wood,its aesthetic appearance,and its environmental friendliness(Fam et al.2010).

The formed profiles can be reinforced with technical fibers and/or textiles laminated to the outer wood surface. The purpose of such composite confinement is to strengthen the wood profile in the circumferential direction and to protect wood against environmentally induced damage(Heiduschke and Haller 2010).Cabrero et al. (2010b)concluded that the maximum failure stress for a compressive force is achieved for fiber reinforcement at±0°(reinforcement perpendicular to the longitudinal direction of the wood).They stated that the maximum strength was obtained for a fiber reinforcement of±25°. Heiduschke and Haller(2010)stated thatbrittle failure was observed for unreinforced columns,whose longitudinalsplitting was due to the expansion of the tubes in a circumferentialdirection,resulting in tension perpendicular to the grain failure.Cabrero et al.(2010a)concluded that the analyticalresults were within an error less than 10%of the available experimental results,with a mean error ratio less than 3%.Shin et al.(2002)concluded that without any triggering mechanism,the failure mode at 90°ply orientation was stable and progressive,while catastrophic failure resulted at 0oand mixed mode at 0°/90°and±45°ply orientation.Heiduschke et al.(2008)concluded that, compared to unreinforced columns,the load carrying capacity and ductility of reinforced tubes increased by factors of 1.46 and 1.22,respectively.Han etal.(2007)considered the height and thickness of a rib and the spacing between two adjacent ribs as factors affecting the buckling strength of a pipe.

Weibull distribution has the capability to model experimental data of very different character.Dodson(1994) described developments regarding the estimation approaches for Weibull distribution parameters.Barbero et al.(2000) applied thisanalysis in modeling the mechanicalproperties of composite materials and suggested Weibull distribution as a practical method for determining 90 and 95%reliability values used in composite materialmechanics.Yerlikaya and Aktas(2012)analyzed statistically the testresults by Weibull distribution to obtain a 95%reliability levelfor failure load. They concluded thatthe 95%reliability value foreach corner jointconfiguration was approximately equivalent to the 0.53 average value ofthe failure load.

The aim of this study was to obtain the buckling and failure strength of rings constructed with spruce wood,and to determine the effects of the rings reinforced by a composite layer having differentangles(45°,60°,75°and 90°). Four bearing tests were performed for each specimen configuration.Using test data,we determined a Weibull distribution to delimit the 99%reliability of each compressive failure load value.

Materials and methods

Materials

The experimental materials were spruce wood,adhesive, and glass fiber.We used spruce boards of approximately 4 cm thickness and 7 cm width and 53–60 cm length to fabricate cylinders.The oven-dry density of specimens was 0.40 g cm-3.The moisture content of boards was about 11%.The epoxy resin used in the matrix material was Bisphenol ACY-225 and the hardener was Anhydride HY-225.Cylinders were assembled with the polyvinyl acetate(PVAc)adhesive.The mass of glass-fibers was 130 g km-1.

Methods

Spruce boards were machined in a planer.Thus,the boards whose thickness were 30 mm and width were 25 mm were obtained.Then,as shown in Fig.1,these boards were cut in width of 22.6 mm with 5°angle and in length of 530–600 mm using a diamond saw blade in a circular saw. The cleaned angular surfaces of 36 boards were glued by hand with PVAc adhesive.The glued boards were assembled into a cylinder by inserting them into a plastic mold (Fig.2)in which they were left to dry for two days. Cylinder outside diameters measured 26 cm and lengths were 50,and 58 cm.The outer and inner surfaces of the resulting cylinders were sanded.

The outer surface of composite-reinforced(CR)cylinders was glued with a mixture of epoxy adhesive and hardener.Glass-fiber yarn was then wound around the ring in two layers,each 1 mm in thickness,at angles of±45° (CR45),±60°(CR60),±75°(CR75)and±90°(CR90)to the column axis.These specimens were left to dry for 3 days.We prepared five cylinders for testing,one unreinforced(UR)and the other four composite-reinforced (Fig.3).Rings for testing were cut to lengths of 80 mm from the 500 mm cylinders.Four replicate ring samples were prepared for each of five test groups.

Before testing,all specimens were conditioned to approximately 12%moisture content in an environment chamber at(20±2)°C and 65±5%relative humidity until weights were constant.

Tests were carried out under radialcompression loading at room temperature of 20°C with a 10 kN loading capacity universal testing machine at a speed of 1.5 mm min-1(Fig.4).The load was applied on the axial center of the specimen.Load was applied to each specimen until a significant decrease in strength was observed.The load and displacement graphs were computer-plotted at±0.0001 N sensitivity for all tests.

Fig.1 Specimen geometry

Fig.2 Preparing cylinder

Fig.3 Examples of wooden cylinders reinforced by fabrics:unreinforced(a),90°surrounding fiber(b),75°surrounding fiber(c),45° surrounding fiber(d),60°surrounding fiber(e)

Fig.4 Loading type

Weibull distribution

We used a two-parameter Weibull distribution,which is appropriate for bearing strength studies.The distribution function used in this case was that of Kim and Heffernan (2008):

F(x;b,c),represents the probability that the bearing strength is less than or equalto x.Using the equality F(x;b, c)+R(x;b,c)=1,the reliability R(x;b,c),that is,the probability that the bearing strength is at least x,was defined by Chellis(1961)as:

The parameters b and c ofthe distribution function F(x;b, c)are estimated from observations.Linear regression was used for parameter estimation using MicroSoft ExelTM(Chellis 1961;Ibrahim etal.2000;Guden etal.2007;Aktas 2007).This method is based on transforming Eq.1 and calculating double logarithms forboth sides.Hence,a linear regression model in the form Y=mX+r is obtained:

F(x;b,c)is an unknown in Eq.(4)and therefore it is estimated from observed values:order n observations from smallest to largest,and let x(i)denote the i th smallest observation(i=1 corresponds to the smallest and i=n corresponds to the largest).Then a good estimator of F(x(i);b,c)is the median rank of x(i):

Results and discussion

Load carrying capacity

Mean load-carrying capacities and 95%reliability obtained by Weibull distribution are shown in Fig.5.Loadcarrying capacity was greatest at CR60,in experimental tests and statisticalanalyses.Lowestload-carrying capacity was recorded for unreinforced rings in experimental and statistical analyses.In experimental tests load-carrying capacity declined in rank order as CR60＞CR75＞CR90＞CR45.In statistical analyses load-carrying capacity declined in rank order as CR60＞CR75＞CR45＞CR90.

The average load-carrying capacities of rings CR90, CR75,CR60 and CR45 were 137,192,215 and 126% greater than for unreinforced rings.Mean load-carrying capacities were obtained at 53,52,52,52 and 53%of reliability(for unreinforced,CR90,CR75,CR60,and CR45,respectively).

Weibull distribution

The results of the experiments are given in Table 1.Values b and c were calculated by firstranking them from smallest to largest and then computing(Xi,Y)values.We then applied linear regression to the computed(X,Y)values to produce linear regression models(Fig.6).The firstpointin Fig.6 does not appear to fit the line well.This is an expected situation when using linear regression:among consecutive(Y(i),Y(i+1))pairs,(Y(1),Y(2))has the largest absolute difference from the mean.The slope of the regression line was 6.01(for CR90),which is the value of the shape parameter c.

Fig.5 Load carrying capacity

When c＜1.0,the material displayed a decreasing failure rate,c=0 indicates constant failure,and c＞1.0 indicates an increasing failure rate.The value b was computed as b=1934 using the Y axis intercept(=-45.498)in b=e(-Y/c).Therefore,when c=0.368,there was a higher probability that the material would fracture with every unit of decrease in applied compression.The scale parameter b measures the spread in the distribution of data.As a theoretical property R(b;b,c)=0.368. Therefore,R(1934;1934,6.01)=exp(-(x/b)c)=0.368, that is,36.8%of the tested specimens had a load carrying capacity of at least 1934 N.

The plot of R(x;b,c)is shown in Fig.7.The reliability curve in Fig.7 shows that load-carrying capacities less than or equal to 450,700,700,750 and 1,000 N(for unreinforced,CR90,CR75,CR60,and CR45,respectively) would provide high reliability.For a more certain assessment,consider 0.95 a reliability level.When these values are put as R(x;b,c)in Eq.3,and the equation is solved for x,the load carrying capacity values 613,1,180,1,375, 1,410 and 1,343 N(for unreinforced,CR90,CR75,CR60, and CR45,respectively)are obtained.In other words,this material will fail with 0.95 probability under loads of 613, 1,180,1,375,1,410 and 1,343 N(for unreinforced,CR90, CR75,CR60,and CR45,respectively)or more.

Failure mode

Figure 8 shows photographs of failed rings.For unreinforced rings,failures resulted from breaking of wood materials,not from separation of glued surfaces.In other words,specimens failed catastrophically.In addition,for all composite reinforced rings,specimens were not completely broken because of the outer surface of composite reinforcement.Wood was only cracked under thecomposite material.The edges of the rings were squeezed and compressed fibers were moved outward.

Fig.6 Regression line for CR90.a the load-carrying capacity values (N);b median rank

Table 1 Load-carrying capacity values(N)

Fig.7 Weibull reliability distribution for failure load

Fig.8 Photography of failed rings

Conclusion

We quantified the load-bearing capacities of four types of rings constructed of spruce wood and reinforced by glass fiber under radial compressive loads both experimentally and statistically.Load-carrying capacity was highest at CR60 for both experimental and statistical analyses.The lowest value was recorded for unreinforced rings in both experimental and statistical analyses.During ring failure, cracks began in the core materials for all specimens. Failures forunreinforced rings were formed by catastrophic breaking of wood materials.No composite-reinforced rings failed catastrophically.

Aktas A(2007)Statistical analysis of bearing strength of glass-fiber composite materials.J Reinf Plast Compos 26:555–564

Barbero E,Fernandez-Saez J,Navarro C(2000)Statisticalanalysis of the mechanical properties of composite materials.Compos B 31:375–381

Cabrero JM,Heiduschke A,Haller P(2010a)Analytical assessment of the load carrying capacity of axially loaded wooden reinforced tubes.Compos Struct 92:2955–2965

Cabrero JM,Heiduschke A,Haller P(2010)Parametric analysis of composite reinforced wood tubes under axial compression.In: World conference on timber engineering

Chellis RD(1961)Deterioration and preservation of piles.Pile foundations.McGraw-Hill,New York,pp 339–372

Dodson B(1994)Weibull analysis.American Society for Quality, Milwaukee

Fam A,Kim YJ,Son J(2010)A numerical investigation into the response of free end tubular composite poles subjected to axial and lateral loads.Thin-Wall Struct 48:650–659

Guden M,Yuksel S,Tasdemirci A,Tanoglu M(2007)Effect of aluminum closed-cell foam filling on the quasi-static axial crush performance of glass fiber reinforced polyester composite and aluminum/composite hybrid tubes.Compos Struct 81:480–490

Haller P(2007)Concepts for textile reinforcements for timber structures.Mater Struct 40:107–118

Han TH,Han SY,Kang YJ(2007)Elastic buckling behavior characteristics of GFRP pipe with reinforced ribs,composites and polycon.American Composites Manufacturers Association, 17–19 Oct,Tampa

Heiduschke A,Haller P(2010)Fiber-reinforced plastic-confined wood profiles under axial compression.Int Assoc Bridge Struct Eng 3:246–253

Heiduschke A,Cabrero JM,Manthey C,Haller P,Gu¨nther E(2008) Mechanical behaviour and life cycle assessment of fibrereinforced timber profiles.In:Braganca L,Koukkari H,Blok H,Cervasio R,Velkovic M,Plewako RUV,Landolfo Z,Silva L, Haller P(eds)Cost C25 sustainability of constructions-integrated approach to lifetime engineering.COST C-25.European Commission,Dresden,pp 3.38–3.46

Ibrahim S,Polyzois D,Hassan SK(2000)Developmentof glass fiber reinforced plastic poles for transmission and distribution lines. Can J Civil Eng 27:850–858

Kim YJ,Heffernan PJ(2008)Fatigue behavior of externally strengthened concrete beams with fiber-reinforced polymers: state of the art.J Compos Constr 12(3):246–256

Shin KC,Lee JJ,Kim KH,Song MC,Huh JS(2002)Axial crush and bending collapse of an aluminum/GFRP hybrid square tube and its energy absorption capability.Compos Struct 57:279–287

Yerlikaya NC,Aktas A(2012)Enhancement of load-carrying capacity of corner joints in case-type furniture.Mater Des 37:393–401

23 November 2013/Accepted:7 February 2014/Published online:28 April 2015

?Northeast Forestry University and Springer-Verlag Berlin Heidelberg 2015

The online version is available at http://www.springerlink.com

Corresponding editor:Yu Lei

?Nurdan C?etin Yerlikaya ncyerlikaya@gmail.com

1Department of Industrial Design,Faculty of Artand Design, Yalova University,77100 Yalova,Turkey

2Department of Mechanical Engineering,Faculty of Engineering,Istanbul University,Avc?lar,34320 Istanbul, Turkey