Mechanical behaviour of wood fibre reinforced cementitious composites

BEFIB2012 – Fibre reinforced concrete Joaquim Barros et al. (Eds) © UM, Guimarães, 2012

MECHANICAL BEHAVIOUR OF WOOD FIBRE REINFORCED

CEMENTITIOUS COMPOSITES

Lupita Sierra-Beltran

*

, Erik Schlangen

*

Microlab, Faculty of Civil Engineering and Geosciences, Delft University of Technology Stevinweg 1, 2628CN, Delft, The Netherlands

e-mail: m.g.sierrabeltran@tudelft.nl, web page: www.microlab.citg.tudelft.nl

Keywords: Wood fibres, cement-based materials, deflection-hardening, ductility.

Summary: This paper shows the possibility to use bundles of wood fibres as reinforcement in cement-based materials for specific ductility requirements. The strength of the fibres, the bond between the fibres and the cement matrix and the matrix properties were measured. Then the composites were designed. Under four-point bending load the composites with 5.3% per volume of pine fibres and a low compressive strength cement matrix developed improved deflection capacity and multiple cracking prior to failure, independent from the specimen geometry. This lightweight material can be considered for applications such as low-budget housing in countries subject to seismic risk, where ductility and low weight are desirable characteristics of the building material.

1 INTRODUCTION

Developing countries face, among other challenges, a deficit of habitations. As part of this problem, there is a growing need for low cost and sustainable materials with enhanced mechanical performance for housing, especially in those countries subject to seismic hazard. During an earthquake, structural elements in houses should be able to sustain alternating tensile and compressive forces. The structure could sustain damage but it should not fail, to allow the inhabitants to evacuate, thus ductility is a desirable characteristic for the structure itself, the individual structural elements and ultimately for the building material itself.

Natural fibres are considered promising as reinforcement of cementitious materials because of their availability, reduced costs and low consumption of energy. Several studies have been done on the mechanical properties and durability of cementitious composites with naturally occurring fibres such as sisal, flax or coconut. This paper proposes the use of randomly distributed softwood fibres as reinforcement.

According to the global concern over environmental issues such as the greenhouse effect, a matrix with partial cement replacement by fly ash and silica fume is used in order to lower the environmental impact. Fly ash is a by-product of coal-fired power generation while silica fume is a by-product of the fabrication of silicon metal. Both fly ash and silica fume are pozzolanic materials. An experimental investigation is performed to understand the behaviour of the wood fibres in the cement-based matrices. First the fibre-matrix interface was studied by means of single fibre pullout tests and image analysis of the samples. Then samples of the cementitious composites reinforced with wood have been tested under four-point bending. The material exhibits deflection-hardening behaviour with multiple cracking prior to failure. CT scans and image analysis were used to analyse the fibre distribution. The test results are discussed and conclusions are drawn for further material tailoring in order to improve the mechanical properties and durability of the material.

2 WOOD FIBRES PROPERTIES

of the mechanical behaviour of the fibres. For this research softwood was chosen over hardwood. Softwood fibres are more uniform in size throughout the tree than hardwood fibres and softwood fibres have a more simple form without vessels. Additionally, in softwood the fibres are generally longer than 2 mm while hardwood fibres average about 1 mm long. Even more important is that hardwoods are more susceptible to attack by both acids and alkalis than softwood. Since cement-based materials exhibit an alkaline environment (ph>12 [1]) this was a strong reason to prefer softwood for this project.

Single softwood fibres are strong, with tensile strength values ranging from 200 to 1500 MPa [1], but their length is very limited, about 2 mm. To be able to use wood fibres as reinforcement in cement-based matrices and obtain an increased ductility, the length should be longer. Therefore naturally bonded groups of single fibres were used. The length of these bundles was determined during the design of the composite. For convenience the bundles will from now on be referred to as fibres.

Among softwoods, pine was chosen for its availability both in The Netherlands, were this study was carried out as well as in Mexico, a country with seismic hazard were the material developed in this study can be used. The fibres were cut out of commercially available veneer sheets of Oregon pine. The fibres have a rectangular cross section, the long side being the thickness of the veneer sheet of about 0.5 mm and the short side was set to be 0.25 mm. Figure 1 shows an ESEM (Environmental scanning electro microscope) image of a typical cross section.

Figure 1: ESEM image of typical cross section of pine fibre

The fibres were tested under direct tensile strength using a micro tension-compression testing device (developed by Kammrath & Weiss). Figure 2 shows the test setup. The fibre was glued to two steel non-rotating loading plates using a two component epoxy resin. The tensile tests were done under deformation control, at speed 2 µm/s. The load was measured with a 50N load cell. The displacement of the fibre is given by the displacement of the actuator (± 6mm). Under this circumstance it is necessary to take into account the compliance of the machine. The specimen is assumed to be connected in series with the load cell in the testing machine and this series connection gives the total stiffness of the system (ktotal). For each test the total stiffness is calculated from the test results. Since the stiffness of the load cell and the machine (kc) is known the stiffness of the specimen (ks) can be calculated as:

1

1

1

s total c

k



k



k

(1)

The ratio ks/ktotal increases as ktotal increases. For the fibres tested in this project, the values of ktotal were between 0.01 and 0.36 N/µm.

BEFIB2012:

3

Figure 2: Tensile test setup for testing strength of fibres

Table 1 presents the mechanical properties measured by means of the tensile tests. The results are the average of 50 tests. In literature, only references to the properties of lumber or single fibres can be found. The strength of the chemical bonds of cellulose, which is the basic molecule that makes up the wood fibres, have been estimated around 7000 MPa [3]. Research did with softwood fibres report values of tensile strength between 200 and 1500 MPa. On the other hand, a piece of pine lumber without defects could have a tensile strength of 70 MPa [2]. The tensile strength of bundles of fibres would be, thus, in between the values for single fibres and for lumber, which correspond to the average value here reported.

Table 1 : Dimensional and mechanical characteristic of pine fibres (Standard deviations are indicated between brackets)

Tensile strength σf (MPa) As measured Young’s modulus Ef (GPa) Young’s modulus corrected for compliance Ef (GPa) Strain to failure (%) Area (mm2) 123.9 [75] 4.5 [2.6] 4.7 [2.9] 3.2 [1.2] 0.106 [0.039]

The scattering of the results can be explained by three factors: the test conditions, the fibre characteristics and the area measurements. The test conditions were controlled and similar to all the tests. The fibre microstructure was investigated using a stereomicroscope and the ESEM. When possible, the fibre cross sections at the fracture were photographed as well as the cross section of an adjacent piece of the tested fibre. The obtained images were post-processed using the program ImageJ. The area was calculated as an integral of the whole cross-section of the fibre, including the lumen, which is the open space surrounded by cell wall in the single fibres, and the micro pores inside the cell walls. Using this precise method reduces the influence of the area measurement in the scattering of the physical properties. The physical and mechanical properties of natural fibres exhibit an inherent variability.

Some tests were performed inside the ESEM which makes it possible to observe the lateral deformation of the fibres during the tests. The lateral deformation appears to be less than 1%, thus the Poisson ratio is almost zero for the pine fibres.

3 CEMENT MATRIX AND FIBRE-MATRIX INTERFACE

A fibre-reinforced cementitious material that shows multiple microcracking prior to failure develops enhanced ductility. To promote multiple cracking, the cement matrix fracture toughness should be

limited and the fibre bridging stress should be enhanced. There should be an efficient stress transfer between the fibre and matrix; therefore good bonding between fibre and matrix, to enhance the fibre bridging stress but fibre rupture should be avoided. The matrix design is thus, intimately related to the fibre-matrix interface properties.

In this study, two different cement matrices are studied to achieve a low strength, low fracture toughness mixture that will be compatible with the wood fibre strength discussed in the previous section. In both matrices, cement is partially replaced by fly ash (FA) and silica fume (SF) to lower the environmental impact of a cement-based composite. The production of cement is an energy intensive process responsible for 5% of the global greenhouse gas emission created by human activities [4].

Table 2 presents the mix composition of the two matrices studied. ω stands for the water-to-powder ratio. Even though not all the FA reacts as a binder, due to its high content in relation to the cement, all the fly ash is included in the ω ratio. SP stands for superplasticizer. In both matrices the cement type is CEM I 42.5 N. The sand particle size range between 0.125 and 2 mm. Each mix has its own particle sand distribution.

Table 2 : Mix compositions and mechanical properties Mix ID CEM I (Kg/m3) FA (Kg/m3) SF (Kg/m3) Sand (Kg/m3) ω SP / powder fcm (MPa) fctm (MPa) C1 281 562 14 843a 0.39 0.011 26.5 1.9 C2 178 891 9 539b 0.32 0.013 20.7 1.8 a

, b Are sand particle size distributions

Cubes with dimensions 40x40x40 mm3 were cast for compressive tests and cured for 28 days. Three parallel compressive measurements were done for each mixture. For the direct tensile test, small cubes with a cross section of 10x10 mm2 and a length between 10 and 15 mm were cut out of bigger samples. A notch was made in the sample prior to testing. For this test the micro tension-compression testing device mentioned above was used.

In mix C2, a lower content of cement, due to a higher content of FA, leads to a lower compressive strength fcm as presented in Table 2. The average direct tensile strength fctm, on the other hand, is very close in value for both mixes.

A common way to quantify the bond between fibre and matrix is with pullout tests, during which the pullout load vs slip relation is recorded. Figure 3 shows two typical stress-displacement curves recorded for pine embedded in C1 and C2.

0 0.05 0.1 0.15 0.2 0.25 0 0.2 0.4 0.6 0.8 1

Normalised Pullout Displacement

In te rf a c ia l S h e a r R e s is ta n c e ( M P a ) C1 C2

BEFIB2012:

5

The pullout behaviour depends on a number of factors, namely: the physical and chemical properties of the fibres, the matrix composition and the curing time and conditions where the nature and shape of the fibres is the most important factor. For the pullout tests, a pine fibre embedded in a small block of cement matrix is pullout out. The tests were conducted under deformation control at a speed of 0.002 mm/s. The same setup as for the tensile tests of fibres was used. The cement block and the free end of the fibre were glued to metal plates. A free fibre length of 1 mm was left. At least 6 samples were tested for each cement matrix composition.

The average interface shear stress τmax at 28 days for fibres in C1 matrices is 0.16 MPa (standard deviation 0.06) and for fibres in C2 matrices, 0.25 MPa (standard deviation 0.06). Even though the compressive strength is lower for matrix C2 the interface shear stress is higher. This is due to several factors. Matrix C2 has lower water-to-cement and an improved particle sand distribution, which lead to a more dense matrix and thus to an improved bonding with the pine fibre.

Beside the pullout stress values, the performed tests also give information about the behaviour in the interface when the fibre is pulled out. From the curves of the pullout tests it is clear that the behaviour of pine fibres is governed by friction.

4 BEHAVIOUR OF THE COMPOSITE

A material that shows a strain-hardening behaviour with multiple cracking prior to failure develops enhanced ductility. In fibre reinforced materials, two criteria must be satisfied in order to achieve strain-hardening behaviour: a strength-based criterion and an energy-based criterion [5, 6]. The strength-based criterion, also referred to as the first-crack stress criterion; states that the first crack strength σfc should not exceed the maximum fibre bridging strength σ0. This criterion determines the initiation of cracks. The energy-based criterion states that the complementary energy Jb´ should be

equal or greater than the matrix crack tip toughness Jtip. This criterion assures the presence of multiple cracking. The complementary energy is estimated from the bridging stress versus crack opening relation.

Composites were prepared with the same mix compositions as presented in Table 2 and pine fibres. The water-to–cement ratio was adjusted, depending on the amount of fibres, to achieve proper viscosity and workability. Matrix C1 was prepared with 4 and 5% per volume of fibres and matrix C2 with 5.3%. Composites were casted into samples with two different sizes: thin samples and beam samples. The thin samples have size 120x30x11 mm3 and the beam samples were 160x40x40 mm3. After casting the samples were cured for one day in moulds covered by plastic. Afterwards, the samples were demoulded and cured under sealed conditions at room temperature of 20°C until they were tested.

(a) (b)

Samples were tested under four-point-bending at age 7, 21 and 45 days. The tests results shown in the paper are the average of at least 8 samples tested for each mixture at each age. The samples are from 2 different batches, 4 samples from each batch. Figure 4 shows the setup for tests with thin and beam samples. As can be seen in this figure, two LVDTs fixed on both sides of the test set-up measured the flexural deflection of the sample during the test. The tests were conducted under deformation control at a speed of 0.01 mm/s.

4.1 Composites with matrix C1

Thin samples were prepared with matrix C1 and fibre volume Vf 4%. The fibres had a length of 10 mm. Under bending load, these samples do not exhibit multiple-cracking behaviour. Samples with fibres have an average bending strength of 4.37 MPa (standard deviation 0.28) and samples without fibres and average bending strength of 4.63 MPa (standard deviation 0.08). Figure 5 shows typical flexural load-deflection curves for these tests as well as for tests done with samples without fibres. Even though the samples with fibres do not behave as brittle as the samples without fibres, it is evident that the fibre reinforcement is not enough for the composites to develop multiple cracking, either because the fibre volume is too low or the fibre length is too short. When the first crack strength is reached, around 4.5 MPa, the fibres are not able to sustain higher load and the material fails.

The density of the composites with 4% is 1820 kg/m3.

0.0 1.0 2.0 3.0 4.0 5.0 0 0.1 0.2 0.3 0.4 0.5 0.6 Deflection (mm) B e n d in g S tr e s s ( M P a ) 4% Pine (10 m m ) no fibres

Figure 5: Flexural load-deflection curves for thin samples with Vf 4% 10 mm long fibres

In order to achieve a multiple-cracking behaviour the fibre volume was increased from 4 to 5%. The fibre length was increased from 10 to 25 mm. Figure 6 shows representative load-deflection curves for samples tested after 7 and 21 days curing. When tested at 7 days the fibre bridging strength was higher than the first crack strength, thus after the matrix cracked for the first time the bending stress increased due to the fibres bridging action. New cracks developed until fibre failure at a localized crack. At 21 days the composite behaviour was different. At that age the matrix strength has increased as well as the fibre-matrix interface, but the later to a lower extent. Therefore, after the first crack, the fibres cannot sustain the load. With increasing deflection, few more cracks develop, and as the interface fails the composite load capacity decreases rapidly.

The average bending strength at 7 days was 2.25 MPa (standard deviation 0.23) and 2.68 MPa (standard deviation 0.16) at 21 days.

BEFIB2012:

7

Figure 6: Flexural load-deflection curves for thin samples with Vf 5% 25 mm long fibres

4.2 Composites with matrix C2

Thin and beam samples were prepared with volume fraction 5.3% and matrix C2. The fibre length was 25 mm. All samples exhibit multiple cracking prior to failure when tested at age 7, 21 and 45 days. Figure 7 presents a sample with multiple cracking.

Figure 7: Multiple cracking of samples under four point bending load

0.0 0.5 1.0 1.5 2.0 2.5 0 0.5 1 1.5 2 Deflection (mm) B e n d in g s tr e s s ( M P a ) 21 days 45 days

All thin sample exhibit deflection- hardening behaviour. Typical flexural load-deflection curves for thin sheet samples at different ages can be seen in figure 8. With age, the bending strength increased while the maximum deflection capacity decreased. This tendency has been observed in other fibre-reinforced materials [7, 8]. At 21 days, the bending strength was, in average, 1.6 MPa and the deflection capacity 1.3 mm. At 45 days, the bending strength was 2 MPa and the deflection capacity 1 mm.

The density of the composites with 5.3% is 1615 kg/m3.

The average bending strength of composites with C2 at 21 days, 1.6 MPa (standard deviation 0.2) is lower than for composites with C1, 2.7 MPa (standard deviation 0.04). The ultimate deflection capacity, on the other hand, is lower for composites C1, 1 mm, than for composites C2, 1.3 mm. These results can be explained by the increased amount of fly ash in this matrix. The presence of FA reduces the fibre/matrix interface bond and additionally lowers the matrix toughness [9]. These two trends improve the strain-hardening potential of the composites, and thus explain the increased deflection capacity and multiple cracking of composites C2.

Beams samples also exhibit multi-cracks prior to failure under bending stress. As can be seen in figure 9 the deflection capacity was lower for beam samples than for thin samples. The bending strength of beam samples was higher than for thin samples. The average bending strength for beams was 2.34 MPa (standard deviation 0.12) and for thin sheets was 1.6 MPa (standard deviation 0.2).

0.0 0.5 1.0 1.5 2.0 2.5 3.0 0 0.5 1 1.5 2 Deflection (mm) B e n d in g s tr e s s ( M P a ) Thin Beam

Figure 9: Flexural load-deflection curves for different sample geometry at 21 days

The fibre distribution of tested and un-tested samples was analyzed by means of Computer Tomography (CT) scans and later, image analysis. The scans showed that the fibres are evenly distributed in depth in all the samples, thin and beams. Even though the pine fibres are lighter, in weight, than any of the other materials in cement-based composites, they do not float because of a proper viscosity of the fresh mix. The fibres in thin samples showed a preferential orientation along the longer dimension of the samples. The beam samples, on the other hand, showed a 3-D fibre distribution, but with a tendency of the fibres to align in planes perpendicular to the casting direction. Figure 10 presents a tomography image of a beam sample and the same image after processing showing only the fibres. The casting direction is also indicated. As can be seen in the figure, there are no fibres perfectly vertical, following the casting direction. Some fibres are inclined and only a cross section can be seen. The preferential orientations of the fibres, in the thin samples as well as in the beam samples lead to an improved deflection capacity.

BEFIB2012:

9

(a)

(a)

Figure 10: CT Scan of beam sample of matrix C1 with Vf 5% 25 mm long fibres (a), only the fibres (b)

5 CONCLUSIONS

This paper shows the possibility to use bundles of wood fibres as reinforcement in cementitious materials for specific ductility requirements. The strength of the fibres, the bond between the fibres and the cement matrix and the matrix properties were measured. Under four-point bending load the composites with 5.3% per volume of pine fibres (25 mm long) and cement matrix C2 (characterized by low compressive strength) developed improved deflection capacity and multiple cracking prior to failure, independent from the specimen geometry.

Proper viscosity of the fresh mix allows the fibres to evenly distribute in 1D and 3D arrangements, depending on the geometry of the samples. Fibres do not float even though they are lighter than any other material in the composite.

A lightweight cement-based composite reinforced with pine fibres was designed to exhibit deflection hardening behaviour. This material can be cast-in place as well as pre-cast. This material can be considered for applications such as low-budget housing in countries subject to seismic risk, where ductility and low weight are desirable characteristics of the building material.

Further research can focus in different aspects of the material properties, such as direct tensile strength, as well as the durability assessments to improve and optimize the use of wood bundles in cement-based materials.

ACKNOWLEDGEMENTS

The financial support for this research from the National Council of Science and Technology (CONACYT) from Mexico and from SGS-Intron is gratefully acknowledged. Furthermore the authors thank Prof. K. van Breugel for valuable discussions on the subject.

REFERENCES

[1] M.G. Sierra Beltran, Ductile Cement-Based Composites with Wood Fibres, PhD thesis, Delft University of Technology, The Netherlands (2011).

[2] Forest Products Laboratory, Forest Service, US Department of Agriculture, Wood handbook, Agriculture Handbook 72 (1974)

[3] P. Soroushian and S. Marikunte, “Reinforcement of cement-based materials with cellulose fibres”,

Thin section fiber reinforced concrete and ferrocement, SP-124, American Concrete Institute,

Detroit: 99-124 (1990).

[4] Batelle, Climate Change, Toward a Sustainable Cement Industry, World Business Council on Sustainable Development (WBCSD), 2002. Web site:

http://www.wbcsd.org/web/publications/batelle-full.pdf.

[5] V.C. Li, “Engineered Cementitious Composites – Tailored composites through micromechanical modelling”, Fiber Reinforced Concrete: Present and the Future, Can. Soc. of Civ. Engng. (1998). [6] C. Wu, Micromechanical tailoring of PVA ECC for structural applications, PhD thesis, University

of Michigan, Ann Arbor, USA (2001).

[7] E.H. Yang, Y. Yang and V.C. Li, “Use of High Volumes of Fly Ash to Improve ECC Mechanical Properties and Material Greenness”, ACI Mat. J., 104 (6), 303-311 (2007).

[8] M.D. Lepech and V.C. Li, “Long Term Durability Performance of Engineered Cementitious Composites”, Int. J. for Restoration of Buildings and Monuments, 12 (2), 119-132 (2006).

[9] S. Wang and V.C. Li, “Engineered Cementitious Composites with High Volume Fly Ash”, ACI Mat.