Young Researchers' Forum III 2016 Innovation in ... - MAFIADOC.COM

Young Researchers’ Forum III 2016 Innovation in Construction Materials Edited by:

Veronica Ferrándiz-Mas Christopher Cheeseman Mark Tyrer

Institute of Concrete Technology

Young Researchers’ Forum III Innovation in Construction Materials Imperial College London, United Kingdom 12 April 2016 On behalf of the organizing committee and Forum sponsors we would like to welcome you to Young Researchers’ Forum III on Innovation in Construction Materials. The purpose of the Forum is to provide an opportunity for young researchers on construction materials to present their work and exchange ideas in a friendly and supportive environment. It is particularly aimed at PhD students, early career Post-Doctoral researchers and new appointees in industry. Construction materials are increasingly used in unconventional and innovative applications and include a wider range of materials than ever before. In addition, the challenges related to sustainable development, low carbon and the environment are increasingly important drivers for innovation. The large number of abstracts submitted to the conference, the wide range of work covered and the number of delegates, demonstrates that this is an increasingly important and active research area. We have over 30 research contributions and we would particularly like to thank all our speakers for their efforts in preparing abstracts, extended abstracts and presentations. The Forum will provide a unique overview of ongoing UK university research in construction materials and opportunities for informal discussion, social interaction and networking. It would not have been possible without the drive and enthusiasm of key members of the Cementitious Materials Group of the Institute of Materials, Minerals and Mining (IOM3), support from the Institute for Concrete Technology and The Mineralogical Society. We would also like to acknowledge funding for the event from The National Centre for Infrastructure Materials, UK Collaboratorium for Research in Infrastructure and Cities (UKCRIC). We hope you have an interesting, stimulating and enjoyable day at the Forum. Mark Tyrer, Chairman of the Cementitious Materials Group, IOM3 Chris Cheeseman, Imperial College London Veronica Ferrándiz-Mas, Imperial College London

Young Researchers’ Forum III Innovation in Construction Materials 12th April 2016 th

08.30-09.20 09.20-09.40 09.40-09.50

Paper

09.50-10.00

2

10.00-10.10

3

10.10-10.20

4

10.20-10.30

5

10.30-10.40

6

10.40-10.50

7

10.50-11.00

8

11.00-11.30 11.30-11.40

9

11.40-11.50

10

11.50-12.00

11

12.00-12.10

12

12.10-12.20

13

12.20- 12.30

14

12.30-12.40

15

12.40-12.50

16

12.50-13.00

17

13.00-14.00 14.00-14.10

18

14.10-14.20

19

14.20-14.30

20

14.30-14.40

21

1

Tuesday 12 April 2016 Registration and Coffee Introduction and welcome: Buenfeld, Tyrer/Cheeseman Material properties of stainless steel reinforcing bar under elevated temperature Yidu Bu, Imperial College London, UK Modelling shear failure of reinforced concrete beams with non-local and crack-band damage-plasticity approaches Dimitros Xenos, Mott MacDonald Ltd Strength and porosity of fly ash and slag based geopolymer mortar cured under ambient temperature with variant Silica Fume forms Mohammed Al-Majidi, University of Brighton, UK Microhardness test of fibre-matrix bond in ultra-high performance fibre reinforced concrete Samuel English, Queen’s University Belfast, Northern Ireland, UK Pullout behaviour of hooked end steel fibres embedded in cementitious composites with various cement replacement materials Elyas Zanjani, Portsmouth University, UK Effect of steel fibre distribution and orientation on post-cracking ductility of steel fibre reinforced concrete Olubisi Ige, Portsmouth University, UK st An innovative material of construction of 21 century: cross laminated timber Cagatay Demirci, Imperial College London, UK The timber resource in G.B.- more species for new challenges David Gil-Moreno, Edinburgh Napier University Morning Coffee break/refreshments Sheep wool insulation for the absorption of volatile organic compounds Elie Mansour, Bangor University, UK Sustainable innovation: Engineered bamboo for construction Bhavna Sharma, University of Bath, UK Early age shrinkage of sodium silicate activated slag paste: an evaluation of water evaporation mechanism Hossein Bahrami, University College London, UK Experimental investigation of hygrothermal performance of recycled aggregate concrete Marina Alagusic, University of Zagreb, Croatia Effect of asphalt emulsion on the hydration of Portland cement Wei Li, University College London, UK Timber material model for earthquake-resistant structures Luis Fernando Sirumbai-Zapata, Imperial College London, UK Behaviour of glued-in BFRP rods under combined axial force and bending moment Caoimhe O’Neill, Queen’s University, Belfast Recycling of fly ash-slag geopolymer binder in mortar mixes Napoleana Chaliasou, University of Bath, UK Use of superabsorbent polymers (SAP) to reduce autogenous shrinkage in mortars with ground granulated blast-furnace slag (GGBS) Fernando Almeida, Glasgow Caledonian University Lunch Finnish clay based pozzolan and dehydrated cement paste as cement replacement materials K. Loimula, VTT Technical Research Centre of Finland Potentials of ultra-fine fly ash concrete Bruce Kandie, Masinde Muliro University of Science and Technology Ground expanded perlite as an effective additive to common cements E. Kapeluszna, AGH University of Science and Technology, Poland Nanoscale characterisation of CSH gel pores with scanning transmission electron microscopy Olivia Wenzel, Karlsruhe Institute of Technology, Germany

14.40-14.50

22

14.50-15.00

23

15.00-15.10

24

15.10-15.20

25

15.20-15.50 15.50-16.00

26

16.00-16.10

27

16.10-16.20

28

16.20-16.30

29

16.30- 16.40

30

16.50-17.00

31

17.00-17.15

3D pore structure characterisation of cement pastes using laser scanning confocal microscopy Marcus Yio, Imperial College London, UK Effect of sonicated microsilica on the dispersion of carbon nanotubes within a cementitious composite S. Alrekabi, University of Brighton, UK Use of atomistic modelling in the development of new lime based materials for the construction industry G. Pesce, University of Bath, UK Sustainable bio-inspired stabilisers to enhance the strength, durability and structural stability of adobe bricks used for the construction of houses in the outskirts of Khartoum Amal Balila, University of Reading, UK Afternoon tea and coffee/refreshments The development of a sustainable , energy efficient geopolymers cement-based road pavement material Alistair Wilkinson, Ulster University, Northern Ireland From binder to asphalt mixture; Multiscale experimental programme Mahmoud Elnasri,ARCADIS Design and Consultancy, UK; The University of Nottingham, UK Effects of activator synthesis conditions on alkali-activated binders from tungsten mining waste Gediminas Kastiukas, Brunel University, UK Development of clog resistant permeable concrete Alalea Kia, Imperial College London, UK Development of specially designed 6R construction products Ana Bariceviv, University of Zagreb, Croatia Carbide lime as a cement raw meal component W. Roszcynialski, Eurocement Consulting Group, Poland Concluding Remarks (Tyrer/Cheeseman) and End of YRF

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 1

Material properties of stainless steel reinforcing bar under elevated temperature Y. Bu and L. Gardner Department of Civil and Environmental Engineering, Imperial College London

ABSTRACT Corrosion of carbon steel reinforcing bar can lead to deterioration of concrete structures, especially in regions where road salt is heavily used or in areas close to sea water. The selective use of stainless steel reinforcing bar for high risk elements is cost-effective when the whole life costs of the structure are taken into account. Attention is given to the elevated temperature properties of stainless steel reinforcing bars, which are needed for structural fire design, but have been unexplored to date. A programme of isothermal and anisothermal tensile tests on four types of stainless steel reinforcing bar is described. Bars of diameter 12 mm and 16 mm were studied, plain round and ribbed. Reduction factors were calculated for the key strength, stiffness and ductility properties and compared to equivalent factors for stainless steel plate and strip, as well as those for carbon steel reinforcement. The test results demonstrate that the reduction factors for 0.2% proof strength, strength at 2% strain and ultimate strength derived for stainless steel plate and strip can also be applied to stainless steel reinforcing bar. Revised reduction factors for ultimate strain and fracture strain at elevated temperatures have been proposed. 1. INTRODUCTION The traditional approach to improving the durability of reinforced concrete structures is to modify the concrete specification, in terms of composition and/or cover requirements. Whilst this approach can improve the performance, it is not an inherently durable solution to the problem of chloride-induced corrosion. The use of stainless steel reinforcing bar can be a cost-effective option for structures in potentially corrosive environments which are expensive to maintain and repair because stainless steel is highly resistant to corrosion from chloride ions and does not rely on the high alkalinity of concrete for protection. As well as reduced maintenance costs, the use of stainless steel reinforcement will give the structure a longer design life (> 100 years) compared with carbon steel and enable a reduction in concrete cover and weight of deck and substructure.

Figure 1. Progreso Pier, Mexico (Courtesy of the Nickel Institute).

Unlike carbon steel, stainless steel exhibits a nonlinear stress-strain curve, with increasing strength accompanied by reducing stiffness. For materials which exhibit these stress-strain characteristics, it is conventional to use the 0.2% proof strength f 0.2p as the design strength. Beyond this point, no further strain hardening is considered in traditional design, though it is considered, and systematically harnessed, in the deformation based continuous strength method (Theofanous et al., 2016; Afshan and Gardner, 2013; Buchanan et al., 2016). The thermal properties of stainless steels, including the thermal conductivity, heat capacity and thermal expansion, differ from those of carbon steels because of the effect of the differences in microstructure and alloying content. Although there have been a number of investigations into the performance of stainless steel flat material at elevated temperatures, data on the performance of stainless steel reinforcing bar at elevated temperatures are scarce and no information is given in EN 1992-1-2, the Eurocode dealing with the fire design of concrete structures at elevated temperatures (EN 1992-1-2, 2004).

Stainless reinforcing bar was first developed in the 1930’s (The Concrete Society, 1998) and the earliest known structure with stainless steel reinforcement was the 2100 m long Progreso Pier in the Gulf of Mexico, which was built in 1940 and is still fit-for-purpose (background, Figure 1). In 1969, a neighbouring pier was built with carbon steel reinforcement which has now suffered very severe corrosion (foreground, Figure 1). When the issue of chloride ingress began to be recognised as a significant problem for reinforced concrete structures in corrosive environments, stainless steel reinforcing bar has been used around the world in a range of large and small structures. 1

Table 1. Summary of test specimens No. of Stainless Bar Diameter steadyID steel type* (mm) state grade tests 4307R12 1.4307 R 12 10 4162R12 1.4162 R 12 10 4307R16 1.4307 R 16 10 4362R16 1.4362 R 16 10 4311D12 1.4311 D 12 10 4162D12 1.4162 D 12 10 4162D16 1.4162 D 16 10 4311D16 1.4311 D 16 10 4362D16 1.4362 D 16 10 *R = Plain round; D = Ribbed

The stress-strain relationship for reinforcing steel is the same as that given in EN 1993-1-2 for flat structural steel material. The simplified design rules for carbon steel structures can be safely applied to stainless steel in combination with the specific strength and stiffness reduction factors for the grade of stainless steel. The reduction factors are the ratio of strength (or stiffness or strain) at the elevated design temperature to the strength (or stiffness or strain) at room temperature. As mentioned previously, the reduction factors for stainless steel differ quite significantly from those for carbon steel because of the different microstructure and alloying elements. These specific factors for stainless steel are given in Annex C of EN 1993-1-2, where a stress-strain relationship for stainless steel at elevated temperatures is also defined.

No. of transient -state tests 9 9 9 9 5 8 7 9 9

This paper describes the outcomes of a test programme investigating whether the same reduction factors for stainless steel flat material can be applied to stainless steel reinforcing bar. Two test methods (anisothermal and isothermal) were used to assess the mechanical behaviour at elevated temperatures of plain and ribbed bars of diameter 12 mm and 16 mm in four grades of stainless steel. Figure 2. Apparatus for elevated temperature tests.

2. Elevated temperature test programme An experimental study to determine the elevated temperature material stress-strain properties of stainless steel reinforcing bars was performed. A total of 164 elevated temperature material tests were conducted, covering both plain and ribbed reinforcement, austenitic and duplex material and two bar diameters. The test programme is summarised in Table 1 and the test apparatus is shown in Figure 2. Two testing methods were employed – steady-state (isothermal) and transient-state (anisothermal). In the steady-state tests, the coupons were heated to a target temperature that was then held constant while the coupon was subjected to an increasing axial tensile load until fracture. Typical isothermal stress-strain curves (for the 12 mm diameter grade 1.4307 specimens) are shown in Figure 3. In the transientstate tests, the coupons were loaded with a target tensile stress that was then held constant while the coupon was heated until fracture. By examining the full set of transient-state temperature-strain curves, of which there is one at each applied stress level, stress-strain curves can be derived by extracting strains corresponding to a specific temperature from each curve, and plotting these against the respective applied stresses, as demonstrated in Figure 4. Values were obtained for the following temperature-dependent material properties, where θ is temperature: Modulus of elasticity E θ , 0.2% proof strength f 0.2p,θ , strength at 2% strain f y,θ , ultimate strength f u,θ , strain at ultimate tensile stress, ε u,θ and strain at fracture,ε f,θ .

Figure 3. Isothermal stress-strain curves for grade 1.4307 austenitic stainless steel bars with a diameter of 12 mm.

Figure 4.1 Derivation of isothermal stress-strain curves from transient-state test temperature-strain data.

3. Test results and recommendations The results of the tests (i.e. the derived reduction factors) from the austenitic stainless steel (grades 1.4307 and 1.4311) and the duplex stainless steel (grades 1.4162 and 1.4362) reinforcing bars are 2

analysed. The derived reduction factors are compared with the reduction factors given in EN 1992-1-2 for cold-worked carbon steel reinforcement and the reduction factors set out by Gardner et al. (2010) for different groups of austenitic and duplex stainless steels. The three groups of austenitic stainless steels defined by Gardner et al. (2010) are: austenitic I (1.4301, 1.4318 and 1.4818), austenitic II (1:4401/4 and 1.4541) for more highly alloyed or stabilised grades and austenitic III (1.4571), this stabilised grade being treated separately due to its superior elevated temperature performance. The two austenitic stainless steel grades tested herein are most closely aligned to the austenitic I group, and comparisons are therefore made with the reduction factors proposed for this group. The two groups of duplex stainless steel are duplex I (1.4362) and, with higher nitrogen content, duplex II (1.4462 and 1.4162). The comparisons and observations are made and the results of a representative grade (1.4311) are shown as follows.

below those predicted by EN 1993-1-2, and hence revised reduction factors shown in Figures 6 are recommended.

Figure 6. Reduction factors for ultimate strain for austenitic grade 1.4311 stainless steel, compared with values given in EN 1993-1-2 Annex C and recommended by Chen and Young 2006 for grade 1.4301 stainless steel, together with proposed values for austenitic group I.

The reduction factors derived from the test results for the fracture strain (k εf = ε f,θ/ ε f ) where ε f is the fracture strain at room temperature, are plotted in Figure 7 for the austenitic grade 1.4311. In the absence of reduction factors recommended in codes or the literature, proposed values shown in Figure 7 are proposed.

The reduction factors derived from the test results for the 0.2% proof strength (k 0.2p = f 0.2p,θ /f 0.2p ) are plotted in Figure 5 for the austenitic grade 1.4311 the reduction factors of Gardner et al. (2010), and also the proportional limit reduction factors for carbon steel cold-worked reinforcement given in EN 1992-1-2 (2004). All the results from the isothermal tests lie on or above the recommended stainless steel reduction factor curves. It is proposed that the existing reduction factors derived for flat material (Gardner et al., 2010) for 0.2% proof strength, ultimate strength and strength at 2% strain may be safely applied to stainless steel reinforcement up to strength class 500.

Figure 7. Reduction factors for fracture strain for austenitic grade 1.4311 stainless steel, together with proposed values for austenitic group I.

4. Conclusions A total of 164 elevated temperature material tests (isothermal and anisothermal) were conducted to study the elevated temperature material properties of stainless steel reinforcing bars, covering both plain and ribbed reinforcement, austenitic and duplex material and two bar diameters. The obtained test results showed that the reduction factors for 0.2% proof strength, strength at 2% strain and ultimate strength derived for stainless steel plate and strip can also be applied to stainless steel reinforcing bar. Revised reduction factors for ultimate strain and fracture strain at elevated temperatures have been proposed.

Figure 5. Reduction factors for 0.2% proof strength for austenitic grade 1.4311 stainless steel, compared with recommendations of Gardner et al. (2010) for stainless steel (austenitic group I) and EN 1992-1-2 for carbon steel.

The reduction factors derived from the test results for the ultimate strain (k εu = ε u,θ/ ε u ), where ε u is the ultimate strain at room temperature, are plotted in Figure 6 for the austenitic grade 1.4311 with the values for stainless steel given in EN 1993-1-2 and the recommendations made by Chen and Young (2006). The test ultimate strains are consistently 3

References Afshan S. and Gardner L. The continuous strength method for structural stainless steel design. ThinWalled Structures. 2013; 68: 42-49. Buchanan C, Gardner L and Liew A. The continuous strength method for the design of circular hollow sections. Journal of Constructional Steel Research. 2016; 118: 207-216. Chen J, Young B. Stress-strain curves for stainless steel at elevated temperatures. Engineering Structures 2006:28 (2):229-239. EN 1992-1-2. Eurocode 2: Design of concrete structures. Part 1.2 General rules. Structural fire design. CEN; 2004. Gardner L, Insausti A, Ng KT, Ashraf M. Elevated temperature material properties of stainless steel alloys. Journal of Constructional Steel Research 2010; 66:634-647. The Concrete Society. Guidance on the use of stainless steel reinforcement, Concrete Society Technical Report No. 51. 1998. Theofanous M, Prospert T, Knobloch M and Gardner L. The continuous strength method for steel cross-section design at elevated temperatures. Thin-Walled Structures. 2016; 98: 94-102.

4

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 2

Modelling shear failure of reinforced concrete beams with nonlocal and crack-band damage-plasticity approaches D. Xenos Special Services Division, Mott MacDonald Ltd P. Grassl School of Engineering, University of Glasgow ABSTRACT In the present paper, the mesh dependence of nonlocal and crack-band constitutive laws is investigated by analysing reinforced concrete specimens. A nonlocal extension for the advanced damage-plasticity constitutive law CDPM2 is proposed that was originally formulated according to the crack-band theory. The nonlocal and the crack-band material models are applied in the analysis of a coarse, a medium and a fine mesh of a reinforced concrete beam failing in shear. The results are evaluated in the form of loaddisplacement curves and failure patterns. and nonlocal models for analysing shear failure of reinforced concrete beams. This failure mode includes both distributed cracking around the reinforcement, bending cracks in the zones of maximum moment and diagonal shear cracks in the area between the supports and the loading points (Bažant & Kazemi, 1991). The advanced damage-plasticity constitutive law CDPM2 is formulated according to the two constitutive approaches and the resulting material models are calibrated initially by analysing a one-dimensional bar subjected to tension (Grassl and Jirásek, 2002). Then, they are applied to model the reinforced concrete beam, which was experimentally investigated in Leonhardt and Walther (1962). The influence of the mesh-size and mesh alignment is evaluated by comparing both global results in the form of load-displacement curves and contour plots of the maximum tensile principal strain.

1. INTRODUCTION The challenge in the nonlinear analysis of shear failure in reinforced concrete beams relies on modelling the quasi-brittle material behaviour of concrete mesh-independently. Two types of constitutive laws commonly used to achieve this aim are the crack-band and the nonlocal models. Crack-band approaches describe concrete failure zones as nonlinear bands that deform inelastically according to a stress-crack opening law (Bažant and Oh, 1983). These models are known to provide mesh-independent strain profiles and global load-displacement curves when failure localises in a single row of elements (Jirásek and Bauer, 2012). However, for distributed crackpatterns both global load-displacement curves and local failure patterns depend on the analysed mesh. The main advantages of crack-band models is that they are easy to implement and computationally efficient. In nonlocal models, the stress at a point is evaluated by weighted averages of the variables describing the state of the material in the vicinity this point (Bažant and Jirásek, 2002). Finer meshes are required compared to the crack-band approach to ensure that there are enough integration points to provide an accurate description of the nonlocal averages. These material models result in regularised strain profiles, which are insensitive to the analysed mesh both for localised and distributed failure patterns. However, there is limited information in the literature on the mesh dependence of nonlocal models when analysing reinforced concrete members.

Both constitutive approaches used in the present paper are based on a damage-plasticity constitutive law. The stress-based plasticity part is efficient in describing multi-axial stress states, as the yield surface corresponds to the strength envelope at a certain stage of hardening (Grassl et al., 2013). Moreover, the plastic strains allow for a realistic description of the permanent displacements of confined concrete under compression. However, plasticity models fail in describing the reduction of stiffness observed during unloading in tension and low confined compression, which in CDPM2 is described by the damage part. The latter is explicitly formulated and is computationally efficient. Hence, the constitutive law is capable of describing the concrete response for various stress states during the shear failure of the reinforced concrete beam.

The aim of the present study is to investigate the mesh sensitivity of damage-plasticity crack-band 1

equivalent strains and , respectively. Each damage variable is evaluated according to an exponential stress-strain law and is a function of three history variables in the form of κ ,κ ,κ . In the nonlocal approach, the equivalent strains and the history variables at a point are calculated as weighted averages of the local ones, at each point in the vicinity of . In the general case, the is calculated as nonlocal average ̅ ̅

2. Constitutive laws The damage-plasticity constitutive law CDPM2 used in the present paper is presented in more detail in Grassl et. al. (2013) and its main components are summarised in this section. The stress-strain law relies on the split of the stress tensor in a tensile part and a compressive part . The two tensors are determined by calculating the positive part and the negative part of the principal effective stress tensor . Then, and are rotated back to the original coordinate system to evaluate and . The stress strain law is defined as

1



Here, , is the weight function that describes the interactions between points and in the actual Euclidean space . The weight function is derived by scaling the function , by its integral over the domain V so that nonlocal averaging does not modify uniform fields of the local parameters (Figure 2). The only parameter of , is the interaction radius that controls the width of the failure zones during tensile failure. Nonlocal averaging was applied in both the tensile and the compressive damage parts.

Figure 1 Schematic representation of the stressstrain law for one-dimensional tension.

1

,



Here, and are the tensile and the compressive scalar damage variables, ranging from 0 (undamaged) to 1 (fully damaged). A schematic description of the stress-strain law is presented for a one-dimensional stress state in Figure 1.

Figure 2 Schematic overview of the exponential, used in the present type weight function paper. In the crack-band formulation only the tensile part of the damage algorithm is modified. Fracture is modelled as a nonlinear band that deforms inelastically. The inelastic behaviour is described according to a stress-crack opening law and the exponential stress-strain law is replaced by a stress-inelastic displacement relationship by multiplying the strains with the width h of the crackband (Figure 3).

The plasticity part is based on the effective stress tensor , which is independent of damage and is evaluated as

where is the elastic material stiffness, is the strain tensor and is the plastic strain tensor. The domain of all elastic and plastically-admissible effective stress states is defined by the yield , . The size and the shape of function , the yield function are controlled by the hardening variables and that depend on the value of the hardening variable κ .

Figure 3 Schematic overview of the crack-band width h of the crack-band approach for the case of a triangular element and Poisson's ratio equal to zero.

In the damage part, the evolution of the damage variables and is controlled by the two 2

(a) (b) Figure 4 (a) Geometry and setup (b) experimental crack patterns of the reinforced concrete beam (Leonhardt and Walther, 1962).

(a) (b) Figure 5 Load-deflection curve for the reinforced concrete beam (Leonhardt and Walther, 1962) analysed for three different meshes with the (a) nonlocal and (b) crack band approach. Deflection is measured at the lowest point at the midspan of the beam.

coarse

medium

fine nonlocal crack-band of the shear beam (Leonhardt and Walther, Figure 6 Contour plots of the maximum tensile principal strain 1962) for all mesh sizes at loading stage 3, marked in Figure 5. Light grey colour corresponds to values of whereas black colour corresponds to values of . 3

the reason why the load-displacement curves are not converging for decreasing mesh size. Moreover, the mesh arrangement does not influence the orientation of the final failure zones.

3. Shear failure of a reinforced concrete beam In this section the results of the analyses of the reinforced concrete beam no. 5 from the series of experiments by Leonhardt and Walther (1962) are presented. The geometry, the loading setup and the experimental fracture patterns are shown in Figure 4. A coarse, a medium and a fine mesh with mesh sizes h=0.02m, 0.01 and 0.005m were used to analyse half of the specimen by constraining the horizontal displacements of the nodes lying on the symmetry plane. Direct displacement control was applied using the vertical displacement at the bottom of the beam midspan. Plane strain triangular elements were used to reduce computational time. This is in accordance with experimental observations because the stress state in the beam during the failure process is neither satisfying plane strain nor plane stress conditions (Bažant, and Kazemi, 2002). The material parameters were calibrated by analysing a one-dimensional bar subjected to tension whose Young’s modulus, fracture energy and tensile strength were specified based on CEB-FIP Model Code 2010. The performance of both constitutive approaches is evaluated in the form of load-displacement curves and contour plots of the maximum tensile principal strain shown in Figure 5 and Figure 6, respectively.

4. Conclusions The present study focuses on the meshdependence of damage-plasticity constitutive laws formulated according to the crack-band and the nonlocal approach. The damage-plasticity law CDPM2 (Grassl et al., 2011) was extended according to the two approaches and the two material models were applied to model a reinforced concrete beam subjected to four-point bending (Leonhardt and Walther, 1962). Both the load-displacement curves and the crack patterns obtained with the crack-band model are meshdependent. Moreover, it was demonstrated that nonlocal models provide mesh-independent results provided that the selected mesh size is small enough to ensure that there are enough material points within the integration zone contributing to the calculation of the nonlocal history variables. References Bažant, Z.P., Oh, B.H., 1983. Crack-band theory for fracture of concrete. Materials and Structures, RILEM, 16:155-177.

The overall failure process is well reproduced by both models. The failure patterns shown in Figure 6 are in agreement with the experimental ones (Figure 4b) and observations reported in similar experimental campaigns (Bažant, and Kazemi, 2002). In Figure 5, it is illustrated that the loaddisplacement curves are close to the experimental ones. All models overestimate the initial stiffness of the specimen. This could be attributed to the plane strain assumption or to an overestimation of the experimental value of the Young’s modulus, which was calculated from the equations provided in CEB-FIP Model Code 2010. In both approaches the tensile damage part and the plasticity part are important to describe the failure mode.

Jirásek, M., Bauer B.H., 2012. Numerical aspects of the crack band approach. Computers and Structures, 110-111:60-78. Bažant, Z.P., Jirásek, M., 2002. Nonlocal Integral Formulations of Plasticity and Damage: Survey of Progress. Journal of Engineering Mechanics, ASCE, 128:1119-1149 Bažant, Z.P., Kazemi, M.T., 2002. Size effect on diagonal shear failure of beams without stirrups. ACI-Structural Journal, 88:268-276. Grassl, P., Jirásek, M., 2010. Meso-scale approach to modelling the fracture process zone of concrete subjected to uniaxial tension. International Journal of Solids and Structures, 47:957-968.

Both constitutive laws provide mesh-independent results for low load levels but closer to the maximum load their responses are different (Figure 5). The nonlocal approach leads to an almost mesh-independent response. The deviation of the results of the coarse mesh with the nonlocal model from the rest of the analyses with nonlocal models is due to the large mesh meaning that there only a few integration points located within the zone considered for nonlocal averaging leading to a less accurate description of the constitutive response is obtained. The use of the crack-band approach results in mesh dependent peak loads and post-peak responses (Figure 5). The smaller the mesh size is, the larger the number and the width of the final failure zones. This difference in the final crack patterns is

Leonhardt, F., Walther, R., 1962. Schubversuche an einfeldrigen Stahlbetonbalken mit und ohne Schubbewehrung zur Ermittlung der Schubtragfähigkeit und der oberen Schubspannungsgrenze. DAfStb Grassl, P., Xenos, D., Nyström, U., Rempling, R., Gylltoft, K., 2013.CDPM2: A damage-plasticity approach to modelling the failure of concrete. International Journal of Solids and Structures, 50(24):3805-3816. 4

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 3

Strength and porosity of fly ash and slag based geopolymer mortar cured under ambient temperature with variant Silica Fume forms Mohammed Al-Majidi, Andreas Lampropoulos, Andrew Cundy School of Environment and Technology, University of Brighton, Moulsecoomb, Brighton BN2 4GJ, UK

ABSTRACT Geopolymer cement-free concrete is an environmental friendly alternative to conventional cementitious concrete made from by-product aluminosilicate materials together with alkaline activator. Most of the published studies on Geopolymer mortar are focused on geopolymer materials hardened under heat curing conditions, which is considered to be a severe limitation for the use of geopolymer concrete in large scale structures. The main aim of the current study is to examine the influence of commercial Silica Fumes (SF), on the mechanical performance of geopolymer mortar. The addition of various SF types have been examined (densified, undensified and slurry silica fume) and Compressive strength tests and porosity measurements have been conducted to evaluate the mechanical and physical properties of the examined mixes. The experimental results indicate that the inclusion of silica fume can considerably improve the mechanical strength of geopolymer mortar, depending on the size of the silica fume particles and on their degree of agglomeration. Geopolymer mixes with undensified silica fume were found to have higher mechanical strength compared to the respective mixes with densified and slurry silica fume. Keyword: Ambient temperature, Geopolymer, Fly Ash, Slag, Silica Fume and aluminium (Al) with alkaline activators (Deb et al., 2014; Zhao et al., 2007). The source material used to produce geopolymers has to contain silicon (Si) and aluminium (Al) oxides in amorphous form. Some natural minerals could be considered as sources for sialate ions such as micas, clays, kaolinite, and spinel (Davidovits, 1999). Alternatively, industrial waste material such as fly ash, slag, rice husk ash, red mud, silica fume, etc are also suitable for producing Geopolymers. This source material might be in single form or in combination with different types of material (Barbosa, 2000; Xu, 2000). Millions of tonnes of by-product material are generated every year by the power generation industry and the iron and steel making industry. Using these by-product materials to produce geopolymer concretes has the potential both to reduce waste flows from industry, and reduce the environmental “footprint” of the construction industry by replacing the traditional OPC (Part et al., 2015). A number of authors have examined the use of fly-ash, Ground Granulated Blast Furnace Slag (GGBFS) and silica fume in particular in geopolymer concrete production. Most previous studies on fly ash based geopolymer mortar indicate however that the material has low strength development when cured under ambient conditions (Deb et al., 2014; Lee and Lee, 2013). Therefore, some researchers have attempted to enhance the reactivity of fly ash based geopolymer cured under

1. Introduction Ordinary Portland Cement (OPC) has traditionally been used as the primary binder material to produce concrete. The usage of conventional concrete is greater than any other man-made material on the planet, and concrete is the secondmost utilised substance after water (Provis, 2008). The quantity of OPC produced has increased year upon year, and according to the European cement Association (Association, 2014), the global production of cement amounted to 4.3 billion tonnes in 2014. However, the production of conventional OPC not only consumes a considerable amount of natural resources and energy but also releases a substantial quantity of carbon dioxide to the atmosphere (Gartner, 2004). Global cement production accounts for 5-7% of worldwide industrial energy consumption (Turner and Collins, 2013). The emissions of greenhouse gases through cement production range from 0.66 to 0.82 kg of CO2 emitted per kilogram of cement manufactured, or about 2.8 billion tons of greenhouse gas emissions annually (Vora and Dave, 2013). This has led to a focus on development of more sustainable concretes, particularly using so-called geopolymer technology to produce more environmental friendly construction materials. Geopolymer materials are synthesized by the reaction of inorganic materials rich in silicon (Si) 1

ambient temperature by increasing the fineness of fly ash particle size, and by adding quantities of calcium containing materials such as slag to react with fly ash particles (Chi and Huang, 2013; Islam et al., 2014; Nath and Sarker, 2014). There has been very limited study of the use of ternary binder mixtures to produce geopolymer cured under ambient temperature. The purpose of this study is to investigate the mechanical properties of fly ash and slag based geopolymer mortar cured under ambient temperature with variant particle size distribution of silica fume. A ternary geopolymer binder composition, including fly ash, slag and silica fume, mixed with potassium silicate as an alkaline activator, has been examined. The mechanical and physical properties of the geopolymer mortar were assessed by compressive strength and porosity tests.

Mixture compositions of geopolymer mortar used in the present study. Table.3:

Mixture ID  20S  20S‐10DSF  20S‐10USF  20S‐10SSF 

Fly ash conforming to BS EN 450-1 (2012) (450-1) standard was used in this study as the main binder. GGBS and variants of commercial silica fume (densified, undensified and slurry silica fume) were used as replacements for the fly ash binder. The chemical compositions of the fly ash and slag and silica sand used are given in Table 1. The physical properties of silica fume forms are presented in Table 2. Potassium silicate and silica sand were used as alkaline activator and fine aggregate respectively.

Fly ash 

Slag 

Silica Sand

59 

35 

99.73

Aluminium Oxide, Al2O3 

23 

12 

0.1

Calcium Oxide, CaO 

2.38 

40 

‐‐

Ferric Oxide, Fe2O3 

8.8 

0.2 

0.051

Sulphur Trioxide, SO3 

0.27 

‐‐ 

‐‐

Sodium Oxide, Na2O 

0.74 

‐‐ 

500 gf) at a loading rate of 5s, and the resulting SEM images were studied to check for micro-cracking or damage around the interface. The results from microhardness testing show very little weakness within the composite and at high loads the steel fibre-matrix interface does not show microcracking or damage. Reinforced Concrete (UHFRPC) has enhanced the prospect of fibre reinforcement due to the increased steel fibre-matrix bond slip resistance. This, as stated earlier, is due to the compact nature of the concrete allowing a better frictional resistance to sliding. One problem with increasing fibre slip resistance is that the failure mechanism can change from fibre pull-out to fibre rupture. This can cause the failure to be sudden however a higher failure stress is achieved. Therefore, the fibre-matrix bond behaviour can play a crucial role in determining the behaviour of the material. In light of this a test programme was developed to provide an explanation of the fibre pull-out machanism. The Vickers hardness test was chosen to provide a maicorhardness profile through the sand particles, matrix and fibres to give an indication as to where the weakest point is in the composite material. Indentation analysis has been carried out on concrete in the past, mainly nanoindentation (Constantinides, Ulm, & Vliet, 2003; Mondal, Shah, & Marks, 2009; Wei et al., 1986). It is anticipated that this study will give a micro-mechanical view on the fibre pull-out mechanism that is considered a significant design factor in composite materials such as UHPFRC.

1. INTRODUCTION Ultra-high performance concretes (UHPC) have attracted a significant amount of attention worldwide in the last few years. It is characterised by a water to binder ratio below 0.2, achieved by the addition of superplasticisers (Morin, Cohen Tenoudji, Feylessoufi, & Richard, 2001) and an increase in binder content (Resplendino, 2011). The aggregate (sand) particle size is very small when compared to conventional concrete, with the largest size being in the order of 0.5-1.0 mm. Additional components such as microsilica, which is 5-10 times smaller than cement particles and has a specific surface area around 50 times greater than cement (Habel, Viviani, Denarié, & Brühwiler, 2006; Resplendino, 2011; Spasojević, 2008; Yu, Spiesz, & Brouwers, 2014) helps to reduce the capillary porosity. The resulting concrete is extremely compact with ultra-high compressive strength (usually > 120 MPa) and excellent durability. UHPC is usualy very brittle and various fibres may be added into the mix to increase ductility along with tensile strength and fracture energy. The ultimate tensile failure of fibre reinforced concrete is controlled by the bond strength between the fibre and the concrete matrix. Studies (Aoude, Belghiti, Cook, & Mitchell, 2012; Banthia & Trottier, 1995; Banthia, 1990; Dougill & Hughes, 1975; Naaman & Najm, 1991; Wei, Mandel, & Said, 1986; K Wille & Naaman, 2010; Kay Wille & Naaman, 2012) have been carried out to analyse the mechanism of fibre pull-out and/or rupture under tensile loading. In recent years the development of Ultra High Performance Fibre

2. EXPERIMENTAL METHODS Disc specimens of 25mm diameter and 5mm thickness were cut out from tested tensile specimens. The volume fraction of fibres present in the mix was 2.5%, Table 1. The disc specimens were polished using silicon carbide paper of different grades using a technique explained in Hou et al. (2014). Epoxy resin was not used as it was 1

deemed that it would influence the microhardness results. A microhardness tester (LECO, M-40) was employed in this study.

3. TEST RESULTS The results for the effect of loading level are shown in Figure 1. For hard materials, such as the sand particles and the steel fibre, the load level has a more pronounced effect. This is evident by the large increase in microhardness at low load levels as shown in Figure 1(a) and 1(b). In the matrix itself the load level does not have a large effect on the microhardness values, with a large scatter in the results. There is however a general trend in the data that the measured microhardness value is slightly higher at lower loads. SEM images show that higher loads caused significant damage to the sand particles, indicated by the large cracks and even shattering that occurred. The higher loads also had a similar effect on the matrix, as microcracks were identified around the indent and the indent tip. All indents on the steel fibre were well formed and regular.

Table 1. Mix proportions used in this study Constituent Proportion Cement 0.209 GGBS 0.144 Microsilica slurry 0.176 Silica sand 0.386 Superplasticiser 0.048 Water 0.075 Steel fibres 0.025

To evaluate the effects of load level and loading rate on the microhardness values for each of the three materials (sand particle, matrix and steel fibre), two sets of tests were carried out. Firstly, load levels of 10 gf, 25 gf, 50 gf, 100 gf, 300 gf, 500 gf, and 1000 gf were applied to each of the three materials at a constant loading rate of 15s. Secondly, loading rates varying from 5s up to 30s at 5s intervals were applied to the three different materials at a constant load level of 100 gf. From these preliminary tests a load level of 25 gf applied at a loading rate of 30s was chosen for determining the cross sectional microhardness profiles. This test involved performing Vickers indentations along a path that ran through the matrix, sand particles and steel fibre. This gave a profile of microhardness through all three materials in the composite. A further test involved applying a load of 500 gf upwards at a loading rate of 5s on the fibre-matrix interface. This was carried out to induce crack or damage at the interface between matrix and steel fibre. Measurement of the indent’s diagonal distances were obtained from images taken from SEM. These images were calibrated using Photoshop and the measurement tool available in that software was used to measure diagonal dimensions of the indent. Six readings of each diagonal dimension were taken and the average reading was used to obtain the microhardness value. The average standard deviation and average coefficient of variance of each test are given in Table 2.

(a)

(b)

(c)

Table 2. Average standard deviation and average coefficient of variance for measurements from microhardness tests Average Average Standard Coefficient of Material Test Deviation of Variance of measurements measurements (µm) (%) Sand 0.766 5.770 particle Load Matrix 0.495 1.956 level Fibre 1.881 2.888 Sand 0.472 4.505 particle Loading Matrix 0.160 0.964 rate Fibre 1.433 3.440

Figure 1. Effect of load level on the microhardness value for (a) the sand particle (b) the steel fibre and (c) the matrix

2

The results for loading rate effect are shown in Figure 2. An increase in the loading rate causes an increase in the measured microhardness of sand particles and the matrix. It has no effect on the microhardness values for the steel fibre. The SEM imaging showed that the slower loading rates produced better formed and more regular indents in the matrix and sand particles.

concrete where there is a visible zone of weakness around the fibre (Wei et al., 1986). The area where failure will occur in the steel fibre-matrix interface at pull-out is difficult to predict from the microhardness tests, as there are no clear areas of weakness. This is however why UHPFRC has an increased pull-out resistance over normal strength concrete. Previous research (Kay Wille & Naaman, 2012) and images taken of pulled out fibres (Figure 4) has shown that there is a small amount of matrix that comes away with the steel fibres on complete pull-out, however the fibre is mainly smooth. This may indicate that failure/slip occurs within a very small band at the steel fibre-matrix interface.

(a)

(b)

(c)

Figure 2. Effect of loading rate on the microhardness value for (a) the sand particle (b) the steel fibre and (c) the matrix

The cross sectional microhardness profiles obtained from the tests are shown in Figure 3. The peaks shown on the graphs are from sand particles. The higher microhardness of the sand particle would suggest that failure does not occur through the sand particle. The steel fibre-matrix interface is indicated by the dashed lines on the graphs. The microhardness values obtained across the fibre are very consistent, around a value of 800 MPa. The profiles show that there are no significant areas of weakness within the matrix, most likely due to the increased compactness of the matrix. This is different in normal strength fibre reinforced

Figure 3. Cross sectional profiles through the sand particles, cement matrix and fibre

Figure 4. Image of the surface of a fibre pulled out from a UHPC matrix in tension

3

17, 610–620. doi:10.1139/l90-069 Banthia, N., & Trottier, J.-F. (1995). Concrete Reinforced with Deformed Steel Fibers Part II: Toughness Characterization. ACI Materials Jounal, 92(2), 146–154. doi:10.14359/9765 Constantinides, G., Ulm, F., & Vliet, K. Van. (2003). On the use of nanoindentation for cementitious materials. Material and Strucutres, 36(April), 191– 196. Dougill, J. W., & Hughes, B. P. (1975). Fibre bond strengths in cement and concrete. Magazine of Concrete Research, 27(92), 161–166. doi:10.1680/macr.1975.27.92.161 Habel, K., Viviani, M., Denarié, E., & Brühwiler, E. (2006). Development of the mechanical properties of an Ultra-High Performance Fiber Reinforced Concrete (UHPFRC). Cement and Concrete Research, 36(7), 1362–1370. Retrieved from http://linkinghub.elsevier.com/retrieve/pii/S000888 460600069X Hou, P., Zhang, R., & Cheng, X. (2014). Case study of the gradient features of in situ concrete. Case Studies in Construction Materials, 1, 154–163. doi:10.1016/j.cscm.2014.08.003 Mondal, P., Shah, S. P., & Marks, L. D. (2009). Nanoscale Characterization of Cementitious Materials. ACI Materials Journal, 105(2). Morin, V., Cohen Tenoudji, F., Feylessoufi, A., & Richard, P. (2001). Superplasticizer effects on setting and structuration mechanisms of ultra-high performance concrete. Cement and Concrete Research, 31(1), 63–71. Retrieved from http://linkinghub.elsevier.com/retrieve/pii/S000888 4600004282 Naaman, A. E., & Najm, H. S. (1991). Bond-Slip of Steel Fibers in Concrete. ACI Materials Journal, 88(2), 135–145. Resplendino, J. (2011). Designing and Building with UHPFRC. In F. Toutlemonde & J. Resplendino (Eds.), Introduction: What is a UHPFRC?. (pp. 3– 14). London: ISTE Ltd. Spasojević, A. (2008). Structural implications of ultrahigh performance fibre-reinforced concrete in bridge design, 4051. Retrieved from http://infoscience.epfl.ch/record/116412/files/EPFL _TH4051.pdf Wei, S., Mandel, J. A., & Said, S. (1986). Study of the Interface Strength in Steel Fiber-Reinforced Cement-based Composites. ACI Journal Proceedings, 83(4), 597–605. doi:10.14359/10453 Wille, K., & Naaman, A. (2010). Fracture energy of UHPFRC under direct tensile loading. FraMCoS-7 International Conference, Jeju, Korea. Retrieved from http://framcos.org/FraMCoS-7/01-10.pdf Wille, K., & Naaman, A. E. (2012). Pullout Behavior of High-Strength Steel Fibers Embedded in UltraHigh-Performance Concrete, (109), 479–488. Yu, R., Spiesz, P., & Brouwers, H. J. H. (2014). Mix design and properties assessment of Ultra-High Performance Fibre Reinforced Concrete (UHPFRC). Cement and Concrete Research, 56, 29–39. Retrieved from http://linkinghub.elsevier.com/retrieve/pii/S000888 4613002275

The SEM image in Figure 5 shows the microcracks caused by a Vickers indenter at a load level of 1000 gf and a load rate of 5s. The crack has formed at the steel fibre-matrix interface with a further crack propagating away from the interface at the indent tip. Other similar tests do not show many cracks forming around the interface, indicating a very strong bond between matrix and steel fibre.

Figure 5. SEM image of the steel fibre-matrix interface showing the micro-cracks caused by the indenter applied at a load level of 1000 gf

4. CONCLUSIONS The following conclusions can be drawn from this study; •

•

• •

The measured microhardness is higher at lower load levels for harder materials such as sand particles and steel fibres, but the the effect is insignificant for the matrix within the tested range; The measured microhardness is higher at higher loading rates for sand particles and the matrix but such effect is not evident for steel fibres; Cross sectional microhardness profiles show very little weak points within the UHPFRC composite; Targeting the steel fibre-matrix interface with high loads did not cause significant microcracking or damage also suggesting that the bond strength between fibre and matrix is very high.

References Aoude, H., Belghiti, M., Cook, W. D., & Mitchell, D. (2012). Response of steel fiber-reinforced concrete beams with and without stirrups. ACI Structural Journal, 109(109), 359–367. Banthia, N. (1990). A study of some factors affecting the fiber–matrix bond in steel fiber reinforced concrete. Canadian Journal of Civil Engineering,

4

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 5

PULLOUT BEHAVIOUR OF HOOKED END STEEL FIBRES EMBEDDED IN CEMENTITIOUS COMPOSITES WITH VARIOUS CEMENT REPLACEMENT MATERIALS Elyas Malaki Zanjani, Stephanie Barnett, David Begg School of Civil Engineering and Surveying, University of Portsmouth ABSTRACT In this research project, fibre-mortar bonding properties of steel fibre reinforced concrete are studied experimentally. Water/binder ratio and use of cement replacements materials (CRM) have a major effect on strength of concrete and are expected to have similar effects on bond strengths so these factors are focussed on in this project. In addition, CRM and high performance reinforcement materials will enable the amount of concrete being used in civil engineering applications to be reduced, so there is potential for such material developments to help reduce the carbon footprint associated with concrete. The cement replacements which have been used in this research included silica fume (SF), pulverised fuel ash (PFA), limestone filler (LF) and ground granulated blast-furnace slag (GGBS). In total, around 800 samples have been manufactured for experimental research on compressive strength and pullout behaviour of hooked end steel fibres from cement-based matrices. The effects of parameters such as water/binder ratio, cement replacement type and level, age of sample, hooked end shape and tensile strength of fibre on fibre-matrix pullout behaviour were determined. The results of tests and analysis indicate that hook shape and tensile strength of fibre affect the peak pullout force and CRMs can improve the pullout energy which would be useful for the optimisation of steel fibre reinforced concrete. The outcome of this research may be useful to widen the potential applications of the material across civil engineering. 1. INTRODUCTION

Therefore, pullout test idea was raised up as an effective method to fulfil the requirements of this research. As the samples could be smaller and need less concrete the drawback of the beam test was also overcome. There is still no standard or guideline to conduct pullout tests. Mono-fibre sample or double-bridging samples with fibre can be used. The trial tests showed that both methods gave the same result. As the preparation of sole sample method was easier, the volume of concrete to use was 50% less and the fibre embedded length and orientation could be controlled much more, the sole sample method was chosen for the pullout test procedure.

New types of fibres are still being developed by manufacturers as reinforcement for concrete, the most widely used man-made material, while the design codes for use of steel fibre reinforced concrete (SFRC) are also still in development. There are also some newly released fibres in the industry which seem to behave differently in concrete and are able to be used as sole reinforcement, enabling use of steel fibre reinforced concrete in applications where it hasn’t been considered before. Fibre-mortar bonding is one of the most important factors in steel fibre reinforced concrete and is even more so in these new types of steel fibres where ductility is provided by the fibres stretching as well as pulling out of the matrix. The properties of this composite material are affected by the properties of both the fibres and the matrix. In addition to the type of steel, the formation of end anchorage of steel fibres is important in terms of pullout force (and hence post-peak load carrying capacity) (Soutsos et al., 2012). The common method to study the behaviour of fibres in crack bridging is applying bending load on prismatic samples to determine the flexural properties. However, to study the effects of different variables on SFRC, the number of samples required is usually high and so a huge volume of concrete is needed for manufacturing the beams.

2. Methodology A detailed investigation of the effects of varying water binder (w/b) ratio and type and level of CRM on the fibre-matrix bond in different concrete and cementitious paste composites was carried out. Properties of steel fibres used in this research are presented in Table 1. Mix proportioning was based on reference to literature, previous work by the author and trial mixes to determine water/binder ratios, types and percentage levels of CRM (Concrete Society Working Group, 2011 and British standard institute, 2011). Aggregate content was kept constant for all concrete mixes. The slump of concrete was kept in the range of 100mm-200mm in concrete mixes by varying superplasticizer content, so this varies depending on both w/b ratio and CRM type and 1

the reason to remove aggregate from some of the mixes and re-test fibres in paste samples which give more reliable results (Figures 3). According to the results, there is no obvious improvement in maximum pullout force by using the CRM in concrete mixes. However, in the paste mixes, SF increases pullout strength for all the fibre types. Similar to compressive strength, there is slight reduction in the pullout strength of the fibres by increasing the w/b ratio; the pullout strength increases dramatically between 7 and 28 days and slightly changes after then. Generally, improving hooks and material in type IV fibres increased the pullout strength and pullout energy more than 120% compared to type I, 90% compared to type II and 20% compared to type III. There is not any simple relation between compressive strength, pullout strength and pullout energy. According to the results from concrete mixes, GGBS increases the residual pullout strength of fibre after fibre starts to move in the sample (frictional pullout stage). However, PFA increased this parameter in the paste mixes and therefore needs further investigation.

level. The mixes are named based on their water/binder ratio, the type of CRM used and the replacement percentage. Therefore, 25, 35 and 45 refer to w/c ratios of 0.25, 0.35 and 0.45 respectively. CEM, SF, PFA, LF and GS refer to the cementitious material used. For example, C25SF10 refers to 0.25 water/binder and 10% replacement of cement by silica fume. Table 1. Steel fibre properties TYPE I TYPE II TYPE III TYPE IV Tensile 1160MPa 2300MPa 1500MPa 2300MPa Material Strength properties Young’s ±210000 ±210000 ±210000 ±210000 Modulus MPa MPa MPa MPa

Fibre Series

Geometry

Hook Shape

Length 60mm 60mm 60mm 60mm Diameter 0.90mm 0.90mm 0.90mm 0.90mm 3183 3183 3183 3183 Fibre Fibres/kg Fibres/kg Fibres/kg Fibres/kg network

In total, 297 cubes were manufactured for compressive strength testing and 456 cubes for the pullout testing. The samples for compression testing were cast in 100 mm polystyrene moulds and for the pullout test in 50 mm cube steel moulds. A sole steel fibre was embedded 30mm into the pullout cubes after pouring. The samples were demoulded 24 hours later. They were cured in water tanks at normal laboratory temperature (20 °C) till their test ages (7, 28 and 56 days). Figure 1 shows the schematic pullout test arrangement.

Figure 2. Effect of SF on type IV fibres (concrete samples)

Figure 1. Fibre pullout test schematic

3. Results and conclusions Figure 3. Effects of fibre type and CRM on pullout strength and pullout energy (paste samples)

Higher w/b ratio gives lower compressive strength in concrete mixes either containing CRM or not. The effect of CRM on compressive strength is different in concrete mixes with various w/b ratios. However, all CRMs increase compressive strength in paste mixes with 0.35 w/b ratio. The mixes containing silica fume, PFA and limestone filler get higher strength after 28 days. However, GGBS reaches almost to its ultimate compressive strength at the age of 28 days. In some cases, the variability in the results of pullout tests in concrete mixes is very high (Figure 2). This can be because of effects such as an aggregate particle in fibre bonding area. This was

References Concrete Society Working Group. (2011). Technical Report No. 74, Cementitious Materials. Surrey: The Concrete Society. Soutsos, M., Le, T., & Lampropoulos, A. (2012). Flexural performance of fibre reinforced concrete made with steel and synthetic fibres. Construction and Building Materials 36, 704–710. British standard institute. (2011). BS EN 197-1: Cement, Composition, specifications and conformity criteria for common cements [Electronic version]. London: Author. 2

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 6

Effect of Steel Fibre Distribution and Orientation on PostCracking Ductility of Steel Fibre Reinforced Concrete Olubisi Ige, Stephanie Barnett, Ayman Nassif, John Williams School of Civil Engineering and Surveying, University of Portsmouth, Portsmouth, PO1 3AH, UK

ABSTRACT Steel fibres are being added to concrete matrix to improve the structural integrity and obtain a more superior resistance to cracking and crack propagation. This study therefore investigates the distribution and orientation of steel fibres in steel fibre reinforced concrete (sfrc) employing various methods of influencing the distribution and orientation with analysis of how the positioning of steel fibres affect the post-cracking strength and other properties enhancing the ductility of sfrc. Hooked-end steel fibres with 50 mm and 60 mm length, of varying diameter resulting in different aspect ratio (ratio of length to diameter of fibre) of 45, 65 and 80, and dosages of 0 kg/m³, 25 kg/m³, 40 kg/m³, 50 kg/m³ and 60 kg/m³ were used with maximum sizes of coarse aggregate of 10 mm and 20 mm. The same mix proportions of concrete were used throughout the investigation. X-ray Computed Tomography was employed for imaging cores taken from the slab specimens after testing and subsequent analysis of the positioning of fibres carried out by Insight software. The experimental results show that the strength performance of sfrc improved drastically when compared to plain concrete without fibres. Remarkable improvements were observed at higher dosages of fibres in concrete while combined effects of fibre geometry/aggregate size give improved ductility. type of steel fibre has been found to give best results in post cracking strength and ductility of sfrc, in a study where all steel and synthetic fibres commonly used in reinforcing concrete were studied, (Soutsos et al., 2012). The inclusion of fibres to a concrete matrix contributes mainly to the energy absorption capacity, controls the cracking of structural elements and enhances the load bearing capacity of the material (Abrishambaf et al., 2013). This is in aggrement with most researchers that have carried out investigations on sfrc.

1. INTRODUCTION Concrete is the most used man-made material in the world and its consumption continues to increase on daily basis around the globe because of many reasons which include availability of its constituents everywhere, relatively cheap as bulk material, wide range of where it can be used and simple way of manufacture. This popularity notwithstanding, concrete is a brittle material and its failure mode of sudden collapse is catastrophic in nature. This is because while concrete can withstand a compressive load to an extent, it is very weak when it comes to tensile load. Added to the challenge of brittleness or sudden failure is the subjection of concrete structures to more loadings as a result of increase in the world population, advancement in technology, increase in natural disaster and terrorist attack. These challenges are being met with the evolution of steel fibre reinforced concrete (sfrc) which contains thin steel fibres spread throughout the mass in order to obtain multi-dimensional reinforcement with main objective of modifying the properties of concrete.

Notwithstanding many studies on sfrc with results showing relationship between post-cracking strength and fibre volume, type, geometry, and with varying specimen geometry and binder formulation, the experimental results on monitor of distribution and orientation of steel fibres in sfrc and how the positioning affects the maximum output of the resulting material are considered few. In order to fully exploit the full potential of steel fibre reinforced concrete in practice, especially for those in construction industries, the distribution and orientation of fibres within the concrete matrix must be fully investigated and evaluated so as to maximise the contribution of fibres to the postcracking ability of sfrc.

The inclusion of fibres as reinforcement for brittle matrix materials is not a recent phenomenon; it has been since time immemorial such as horsehair in mortar and straw in bricks since Biblical times as recorded in Exodus, over 3500 years ago (Brandt, 2008) while in more recent times, steel, polypropylene and glass fibres have been used for the past 40 years, (Hannant, 2003). Hooked end

This research work studies the effects of fibre geometry, fibre volume and maximum aggregate sizes on the workings of fibres within the concrete matrix and the influence on distribution and 1

orientation, hence, the effects on flexural properties of steel fibre reinforced concrete. 2. Experimental Works

deduced that the higher the dosage, the better the strength and hence, the effectiveness of steel fibres in post-cracking ductility of sfrc. At higher dosage, there is possibility of better interaction between the fibre and the concrete matrix. The importance of aspect ratio and length in good interaction and post cracking ductility is observed in the flexural results as well.

Variables selected for this study are those considered to influence how steel fibres and concrete matrix associate together during mixing Hooked-end steel fibres with 50 mm and 60 mm length, of varying diameter resulting in different aspect ratio (ratio of length to diameter of fibre) of 45, 65 and 80, and dosages of 0 kg/m³, 25 kg/m³, 40 kg/m³, 50 kg/m³ and 60 kg/m³ were used with maximum sizes of coarse aggregate of 10 mm and 20 mm. Slump test on workability of the fresh mix was carried out while compression and three-point bending tests were carried out at a testing age of 28 days on various specimens (100 mm cubes, 150 mm by 150 mm by 600 mm beams and 600 mm by 600 mm by 100 mm slab) employed in the study. X-ray Computed Tomography was employed for imaging core samples of 100 mm diameter taken from the slab specimen after testing and subsequent analysis of the positioning of fibres carried out by Insight software.

Figure 2. Flexural strength of sfrc and normal concrete at 50 3 kg/m dosage.

The square panels cast at 50 kg/m3 were tested according to (British Standards Institute, 2006) after which core samples of 100 mm diameter were extracted from the tested slab and subjected to X-ray computed tomography (CT) imaging for the investigation of internal positioning of steel fibres in hardened concrete matrix. The volumetric images from X-ray CT are then processed using Open source image processing C++ libraries, The Insight Toolkit (ITK).

3. Results and Discussion The compressive strength results conducted on 100 mm cubes as given in figure 1 show the influence of steel fibres. The results generally show a slight improvement in the compressive strength of the mixes by the addition of steel fibres to concrete as compared to compressive strength of plain concrete with 20 mm aggregate size. A fibre type is denoted by its aspect ratio/length, for example, 45/50 means a concrete mixture containing fibre of aspect ratio 45 with length 50. The mean strength of sfrc varies between 58.10 to 68.05 MPa while plain concrete had 61 MPa. The optimum compressive strength was however achieved at 50 Kg/m³ dosages of 80/60 steel fibre after which it declined.

4. Conclusion Inclusion of steel fibre in concrete without doubt has improved the properties of concrete significantly. The final result from image analysis will be used to validate the effects of the chosen variables on fibre distribution and orientation and the effects on mechanical performance of sfrc. References Abrishambaf, A., Barros, J.A., Cunha, V.M., 2013. Relation between fibre distribution and postcracking behaviour in steel fibre reinforced self-compacting concrete panels. Cement and Concrete Research, 51: 57-66. Brandt, A.M., 2008. Fibre reinforced cement-based (FRC) composites after over 40 years of development in building and civil engineering. Composite structures, 86(1): 3-9. British Standards Institute, 2006. BS EN 144885:Testing sprayed concrete. Determination of energy absorption capacity of fibre reinforced slab specimens, Part 5. BSI, London. Hannant, D.J., 2003. Fibre-reinforced concrete. In: Newman, J.S.C., B (Ed.), Advanced Concrete Technology, Burlington, USA.

Figure 1. Relationship between compressive strength and dosages of different fibres in 20 mm maximum aggregate mix.

The results of flexural test of notched beams show a remarkable improvement influenced by steel fibres in concrete up to 83% increase in maximum strength reached which are distinctively noticed at higher dosages as shown in figure 2 It can de 2

Soutsos, M., Le, T., Lampropoulos, A., 2012. Flexural performance of fibre reinforced concrete made with steel and synthetic fibres. Construction and Building Materials, 36: 704710.

3

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 7

An Innovative material of construction of 21st century: Cross Laminated Timber C. Demirci, C. Malaga-Chuquitáype and L. Macorini Department of Civil and Environmental Engineering, Imperial College London

ABSTRACT The population of the world has increased from 1 billion to 6.9 billion in the last two hundred years and it is expected to reach approximately 9 billion in four decades down the line. Therefore, the way be build cities will shape the future of our planet. An alternative and efficient way to achieve this with the help of American invention: the tall building. Megacities have been traditionally built in concrete and steel which are great materials for construction except for the very greenhouse gas emission produced in their fabrication. On the contrary, wood is a distinguished material that stores carbon amazingly, instead of increasing carbon in the atmosphere. Wood is currently being used for mid-rise construction of 4 to 6 storeys and also in high rise construction of 8 to 10-storey buildings. While it is possible to build mid-rise timber buildings with light-frame timber systems, taller buildings with wood require an alternative material of construction. To this end, the field of timber engineering has made a valuable progress with new massive engineered timber products such as Cross Laminated Timber (CLT) which has the potential to make wooden skyscraper dream come true. The use of CLT has been very common in high-rise construction for approximately two decades around the globe. CLT panels behave as solid wood pieces in load bearing which makes multi-storey construction with CLT very efficient in resisting gravity loads. On the other hand, seismic response of high-rise CLT buildings has not been fully covered for particularly large lateral seismic inputs. Therefore, the aim of the present study is to evaluate seismic behaviour of multi-storey CLT buildings.

require an advance, alternative material of construction to build with timber. To this end, the field of timber engineering has made a valuable progress with new massive engineered timber products such as Cross Laminated Timber (CLT).

1 INTRODUCTION The global population is increasing rapidly and people are moving to cities which are dense and innovative-urban lands. The way we build cities will shape the future of our planet. A leading way to accomplish this is an American invention: the tall building (Sarkisian, 2012). Most of the current tall buildings are built in steel and concrete which are materials associated with very high energy and very high greenhouse gas emissions in their fabrication. Concrete causes more than 5% of greenhouse gas emissions and steel causes around 3% so that gives 8% of man’s contribution to greenhouse gases is provided from only steel and concrete (Green, 2012). In addition, two-thirds of the global carbon emissions are created by buildings and transportation systems.

CLT is a new engineered timber product that is produced using industrially dried, fast growing spruce planks, assembled crosswise on top of each other and glued together to form large-sized solid wood components. CLT panels are generally made of 3, 5, 7 and 8 layers with the standard panel thicknesses in the range of 57-to-320 mm (KLH, 2011) but can be created up to 500 mm if required. The maximum board length can be up to 16.50 m and the maximum width is 2.95 m.

On the other hand, there is such a material that stores carbon instead of releasing it into the atmosphere, which is wood. Only one cubic meter wood can store around one tonne of carbon dioxide. Therefore, while reducing man’s contributions, wood also stores a great amount of carbon dioxide (Green, 2012). It is possible to construct mid-rise timber buildings around 4 to 6 storeys with a light-frame construction. However, high-rise construction of 8 to 20-storey buildings

Figure 1. Cross Laminated Timber

1

CLT panels have been developed in Europe, where the trend of tall timber construction was originated, for approximately two decades ago, and spreading Australia in the past 7 years. A few significant highrise CLT buildings which have been completed recently around the world were listed in Table 1. Similarly, CLT utilization and manufacturing has also been gaining recognition in Canada (Pei et al., 2014).

at the hold-down connectors at lower levels due to high overturning demands causing considerable uplift on tension connectors. However, the buildings remained stable without any residual plastic deformation. On the other hand, high floor accelerations were experienced at the top levels of a multi-storey CLT building. Therefore, authors recommended introducing more ductility to the structure.

Table 1. Recently completed tall CLT building projects Project Location Storey

3 SEISMIC BEHAVIOUR OF CLT SYSTEMS

Statdhaus

London, UK

9

Bridport House

London, UK

8

Forte

Melbourne, Australia

10

Holz8

Bad Aibling, Germany

8

Limnologen Project

Vaxjo, Sweden

8

Cenni di Cambiamento

Milan, Italy

9

In order to investigate the seismic behaviour of tall CLT buildings, adequate 2D numerical models are developed using the Finite Element Framework OpenSees (1999). CLT panels are modelled with linear 4 node quad elements whereas 2 node nonlinear link and zero length elements are utilized to simulate hold-downs, angle brackets, and screws. A nearly rigid panel stiffness is considered for CLT wall panels. In order to verify the modelling approach, a comparative study of cyclic behaviour of a single CLT wall panel, followed by a numericalexperimental comparison of 7-storey SOFIE building are prenseted in this study.

High-rise construction with CLT is very efficient in resisting gravity loads because CLT panels act as solid wood load-bearing pieces. On the contrary, the seismic response of multi-storey CLT buildings has not yet been fully covered by codes, especially for large lateral seismic demands (Ceccotti et al., 2013). Experimental shake table tests of mid-rise multi-storey CLT buildings have proved a good dynamic performance. However, in the presence of a seismic event, significantly higher uplift forces are expected in tall CLT buildings. Moreover, experimental tests have shown that multi-storey CLT buildings might be prone to high accelerations at top floor levels (Ceccotti, 2013, Pei, 2014). The problems mentioned herein can be handled by implementing ductile dislocations at selected floor levels along the building height by means of integrated isolation devices or single and multiple rocking mechanisms (Pei, 2014). The aim of the present study is to investigate the seismic behaviour of high-rise CLT buildings.

3.1 Cyclic behaviour of a single CLT wall panel A single CLT wall panel tested by Gavric et al. (2012) was modelled and subjected to a cyclic loading for verification purposes. The panel was connected to the foundation with four angle (shear) brackets and two hold-down (uplift) connectors at both ends. BMF 90x48x3x116 mm angle brackets with 11 Φ 4x60 mm nails were connected with Φ 12 mm bolts and HTT22 hold-downs with Φ 4x60 mm nails were connected with Φ 16 mm bolts to the foundation. The CLT panel was formed by 5 17-mm thick layers. The wall panel dimensions were 2.95x2.95 m. Figure 2 illustrates the numerical model of the CLT wall. The loading protocol was the one prescribed by BS EN 12512 (2001) and shown in Figure 3. An additional 18.5 kN/m vertical load was applied on top of the panel.

2 BACKGROUND 15 CLT wall panels with different connection configuration and for different vertical load levels were tested by Dujic et al. (2004) under monotonic and cyclic loading to investigate lateral resistance of CLT walls at the University of Ljubljana, Slovenia. The main findings of the study was that the load-bearing capacity of CLT wall components was limited by the connection stiffness and local failure of the wood was experienced at the connections. In addition to CLT wall panel tests, full-scale 3- and 7-storey CLT buildings were tested on a shake table as part of the SOFIE project (Ceccotti, 2008). A number of ground motion records were applied to buildings to explore the seismic response of multistorey CLT buildings. The major findings of the experimental study on multi-storey CLT buildings were that damage was reported to be concentrated

Figure 2: OpenSees model of the wall panel

All connectors were calibrated on an experimental study conducted by Gavric et al. (2011). Table 2 2

presents calibrated values of angle brackets and hold-down connectors. A numerical model of tested CLT wall panel was developed in Abaqus by Gavric et al. (2012) and OpenSees with the modelling approach which will be employed for taller buildings herein this research.

3.2 7-storey SOFIE building SOFIE (Ceccotti, 2008) project was a research effort on full-scale multi-storey CLT timber buildings on a shaking table. As part of this research program, a series of seismic tests were carried out on a 7-storey CLT building at (NEID) E-Defence laboratory of Japanese National Research Institute for Earth Science and Disaster Prevention in 2007, remain today as the only pseudo-dynamic testing of a tall timber structure under strong earthquakes. The response to the Kobe JMA 1995 record was analysed herein for verification purposes. Figure 5 shows North-East view of the 7-storey SOFIE building (Dujic et al., 2010).

Table 2: Calibration of mean values obtained from experimental results on CLT connections Parameter Angle Angle HoldHolddescription bracket bracket down down shear tension tension shear Elastic stiff. kel [kN/mm]

1.78

2.76

4.82

0.99

1st inel. stiff. kpl [kN/mm]

0.22

0.41

0.69

0.12

Yield force Fy [kN]

23.77

19.23

40.3

9.76

Max. force Fmax [kN]

27.71

23.51

48.33

13.88

The hold-down connectors used in the SOFIE project are HTT22 with maximum 32 nails and angle brackets are BMF 90x48x3x116 with maximum 18 nails. CLT panels are made of 5layered panels with three different thickness values for different storey levels. First two storeys are 142 mm, middle two storeys are 125 mm and upper two storeys are 85 mm thick CLT panels. The thickness of CLT wall panels is reduced through the building height in direct relationship with the decreasing lateral seismic forces applied to each floor level (Dujic et al., 2010).

Figure 3: Cyclic testing protocol employed

The comparison of cyclic behaviour of numerical model and experimental results are given in Figure 4 as lateral force against deflection at the top of the wall. It can be seen in Figure 4 that modelling approach in OpenSees (1999) presents a good approximation of similarity to the model developed by Gavric et al. (2012) and experimental results in terms of cyclic response of CLT wall panel.

Figure 5: N-E view of 7-storey SOFIE building

A numerical model was developed in OpenSees (1999) to investigate the seismic behaviour of the 7-storey building performing non-linear response history analysis and compare results of the modelling assumptions employed in this research against numerical-experimental results reported in Rinaldin and Fragiacomo (2016). Figure 4: Comparison of numerical and experimental cyclic behaviour of the CLT wall panel

The measured period corresponding to first mode on a shake table is 0.43 s and the period corresponding to second mode is 0.301 s (Rinaldin and Fragiacomo, 2016). The modelling approach utilized to model the SOFIE building herein gives a 3

fundamental period of 0.46 s and the period corresponding to second mode of 0.313 s. Therefore, a good agreement was found between the shake table test results and numerical modelling approach employed herein. In addition, fundamental period was determined using kpanel stiffness expression provided by Blass and Fellmoser (2004) for which a period of 0.73 s was obtained. Since this determined fundamental period using proposed stiffness equation (Blass and Fellmoser, 2004) did not correspond the measured fundamental period of structure it was recommended to use a nearly rigid body panel stiffness in numerical models. Table 3: Fundamental period of SOFIE building Fundamenta Shake table OpenSees l period measuremen model t 0.427 0.463 T1 [s] 0.301 0.313 T2 [s]

seismic regions: The XLAM case. Structural Engineering International, 156-165. Ceccotti, A., Sandhaas, C., Okabe, M., Yasumura, M., Minowa, C., and Kawai, N. (2013). SOFIE project-3D shaking table test on a sevenstorey full-scale cross-laminated timber building. Earthquake Engineering Structural Dynamics, 2003-2021. CEN. (2004). European Committee for Standardization (CEN), Eurocode 8: Design of structures for earthquake resistance-Part 1: General rules, seismic actions and rules for buildings. Brussels, Belgium: EN 1998-1. Dujic, B. Puceli, J. and Zarnic, R. (2004). Testing of racking behaviour of massive wooden wall panels. 37th CIB-W18 Meeting, International Council for Building Research and Innovation. Rotterdam, Netherlands. Dujic, B., Strus K., Zarnic, R. and Ceccotti, A. (2010). Prediction of dynamic response of a 7storey massive XLam wooden building tested on a shaking table. World Conference on Timber Engineering (pp. 20-24). Roma, Italia: Italian National Research Council. OpenSees. (1999). Open System for Earthquake Engineering. Berkeley, California, USA. Gavric, I. Fragiacomo, M., and Ceccotti, A. (2012). Strength and deformation characteristics of typical XLam connections. World Conference on Timber Engineering. Auckland, New Zeland. Gavric, I. Rinaldin, G., Amadio, C., Fragiacomo, M., and Ceccotti, A. (2012). Experimentalnumerical analyses of the seismic behaviour of cross-laminated wall systems. 15th World Conference of Earthqauke Engineering. Lisboa. Green, M. (2012). Wood Works. Retrieved from www.woodworks.org/download.php? KLH Massivholz GmbH. (2011). http://www.klhuk.com/ Pei, S. Van de Lindt, J., Popovski, M., Berman, J., sause, R., Blomgren, H., and Rammer, D. (2014). Progress on the development of seismic resilient tall CLT buildings in the Pacific Northwest. World Conference on Timber Engineering. Quebec City, Canada. Popovski, M. Schneider, J., and Schweinsteiger, M. (2010). Lateral load resistance of crosslaminated wood panels. World Conference on Timber Engineering (pp. 20-24). Roma, Italia: Italian National Research Council. Rinaldin G., Fragiacomo, M. (2016). Non-linear simulation of shaking-table tests on 3- and 7storey X-Lam timber buildings. Engineering Structures, 113-148. Sarkisian, M. (2012). Designing Tall Buildings Structure as Architecture. New York and London: Routledge, Taylor and Francis Group.

Error % 8.43 3.89

In order to verify the modelling approach, a nonlinear response history analysis was also performed under Kobe 1995 ground-motion record. The peak acceleration at the top floor was observed as 2.61 g in the experiment and 2.23 g in the numerical model. Similarly, the first floor peak displacement was 38.6 mm in the experiment and 23.12 mm in the numerical model. It is worth to mention here that experimental tests were performed in a 3 dimensional building and ground motion record of the Kobe earthquake used for non-linear response history analysis herein may not be the one employed in the experiment. However, the results were found in good agreement and verified the modelling approach. 4 CONCLUSIONS In this paper, a cyclic behaviour of a single CLT wall panel and seismic response of 7-storey SOFIE building were examined. Following the modelling approach considered herein this study provided a good agreement with experimental test results with regard to the cyclic response of the CLT wall panel, fundamental period, peak storey displacements and peak floor accelerations of the 7-storey SOFIE building. Therefore, the modelling approach employed herein this research was verified and can be used to model taller buildings of 8, 12, 16 and 20 storeys. 5 REFERENCES EN 12512. (2001). EN 12512 Timber structures, Test methods, Cyclic testing of joints made with mechanical fasteners. Brussels: B. En. Blass, H., Fellmoser, J. (2004). Design of solid wood panels with cross layers. 8th World Conference of Timber Engineering. Lahti, Finland. Ceccotti, A. (2008). New techologies for construction of medium-rise buildings in 4

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 8

The timber resource in G.B. - more species for new challenges 1,2

1

David Gil-Moreno , Dan Ridley-Ellis Centre for Wood Science and Technology, Edinburgh Napier University

1

2

Paul McLean Forest Research, Northern Research Station, Scotland.

2

ABSTRACT Timber in Great Britain is mostly produced from Sitka spruce (Picea sitchensis). Biotic threats have led tree growers to consider the planting of a wider range of species to offer a more resilient forest. There is little information on how this will impact on domestic timber production, and the future supply of graded structural timber. Samples from some of the candidate species were tested according to standard procedures and assessed against current timber grades. Results showed that it is possible to produce timber suitable for use in construction from these species.

after adjustments for statistical confidence and other considerations. Typically, one of these three properties will limit the timber to a strength class.

1. INTRODUCTION In order to design structures, engineers and architects require to know the forces and stresses experienced by a structure, and, correspondingly the mechanical and physical properties of the material they work with.

Contrary to the popular opinion that wood density is that limiting property in G.B.; previous research has shown that it is generally stiffness that limits the grading of British timber to a particular strength class.

All construction materials are inherently variable, and timber, being a natural material, is particularly so. The characterisation of construction timber is achieved by grades. In Europe, a construction timber grade, with values for use in design, is known as a strength class. Those values correspond to the population of timber assigned to that grade, and not the properties of individual pieces. Table 1 shows the first five general strength classes for softwood in Europe as described in EN338 (CEN, 2016b).

Table 1. Characteristic values for strength classes C14 to C22 Wood property Characteristic values C14 C16 C18 C20 C22 Bending strength 14 16 18 20 22 (N mm-2) Modulus of Elasticity* 7 8 9 9.5 10 (kN mm-2) -3 Density (kg m ) 290 310 320 330 340 * the characteristic value for MoE must equal or exceed 95% of the value given in table 1.

The majority of wood produced in G.B., particularly for construction, comes from Sitka spruce, a species that has been grown for this purpose since around 1920. Many years of research and industrial experience have characterised the wood properties of this species. However, recent outbreaks of pests and diseases in the UK have highlighted the potential threats to a species monoculture; diversifying the species being grown is one way to manage the risk and ensure that there is a wood supply for future generations.

In order to examine the potential grade of a population of timber, a representative sample of timber (from trees grown under similar conditions to the timber that will be graded) must be assessed. The three properties used to determine grades are: bending strength (or modulus of rupture, MoR), bending stiffness (or modulus of elasticity, MoE) and density. Grading for these three properties is established by testing, and measurements are standardised. All other properties of the strength class are estimated from these three properties using established equations in EN384 (CEN, 2016a).

There are differences in wood properties between species and in order to provide guidance on which species are likely to be useful for timber, we need to collect some evidence. The environment plays a big role in shaping the timber that a tree produces therefore we need to examine trees in the regions in which are likely to be grown.

The important values for MoR and density are the lower 5th percentiles, while values for MoE refer to the mean. The strength classes define minimum requirements for these (referred to as characteristic values) for the population graded, 1

This study investigates the wood properties of four conifer tree species that currently form a minor component of the wood resource in G.B. but may play a more important role in the future.

Timber specimens of 50x100 mm nominal size were cut following radial transects (Figure 3), with annual ring numbers being identified in the battens to ensure material was comparable with trees of a similar age.

2. Material and methods Material obtained for studying came from three even-aged pure species plantations, representing growing regions at three different latitudes (Figure 1).

Figure 3. Sawmilling of a log.

Battens were kiln dried and stored in a controlled environment prior testing in a four point bending test (Figure 4) according to EN 408:2010 (CEN., 2012). Values of MoR and both global and local stiffness (MoE G and MoE L ) were obtained from the tests. MoE G is determined over a longer span than MoE L , which is based on the deflection of the central section tested, and therefore considered shear-free. An empirical relationship between MoE G and MoE L was examined to obtain “pure bending” MoE (MoE PB ). Density was measured afterwards from samples cut from each batten, and oven dried to obtain the moisture content, so that wood properties measured can be adjusted to a standard 12% moisture content as indicated in the standard EN384 (CEN, 2016a). Figure 1. Location of sampled sites.

The four species investigated in this paper are: Norway spruce (Picea abies), noble fir (Abies procera), western red cedar (Thuja plicata) and western hemlock (Tsuga heterophylla). A total of 109 trees (28 for western hemlock and 27 for the rest of species) were felled.

Figure 4. Bending test of a batten.

A total of 126 structural sized specimens of noble fir were tested and analysed, 128 of Norway spruce, 115 of western red cedar and 138 of western hemlock. Material was included up to 45 years old, which is the typical rotation length of a Sitka spruce plantation in G.B. Figure 2. Collection of sampled logs in the forest.

2

3. Results

4. Conclusions

Table 2 summarises the wood properties for the whole sampled population studied at an age of 45 years (Gil-Moreno et al., In press),

Our research demonstrates that the four species investigated are potentially capable of producing high yields of C16 (the basic customary strength class used in the United Kingdom) for a rotation length of 45 years. Norway spruce and western hemlock can also produce high yields of C18, though this grade is not in a large demand in G.B.

MoR

Density

MoE

Table 2. Characteristic values (in bold), mean values, grade and standard deviations (in parentheses) by species restricting the material to 45 years (Gil-Moreno et al., In press). Wood Norway Western Noble Western Properties spruce hemlock fir* red cedar E 0 , mean (kNmm−2) ρ mean (kg m−3) ρk (kg m−3) f mean (N mm−2) f m,k (N mm−2)

Strength class**

8.6 (1.7) C18

8.3 (2.0) C16

7.7 (2.3) C16

7.4 (1.7) C14

378 (36.6) 345 C22 31.1 (9.0)

444 (38.8) 385 C30 34.5 (10.7)

358 (36.7) 324 C18 31.1 (13.1)

365 (30.3) 318 C16 30.1 (8.0)

19.1 C18

18.2 C18

14.8 C14

16.3 C16

C18

C16

C14

C14

Further research could explore the manipulation of desirable wood properties in these species through forest management practices or tree breeding. References CEN, 2016a. Structural timber-Determination characteristic values of mechanical properties and density. 384:2016. Brussels, European Committee for Standardization. Brussels. CEN, 2016b. Structural timber—Strength classes. EN338:2016. European Committee for Standardisation, Brussels, Belgium. CEN., 2012. Timber structures—structural timber and glued laminated timber—Determination of some physical and mechanical properties EN408:2010+A1:2012. European Committee for Standardization, Brussels, p 38. Gil-Moreno, D., Ridley-Ellis, D., Mclean, P., In press. Timber properties of noble fir, Norway spruce, western red cedar and western hemlock grown in Great Britain. Forestry Commission. Gil-Moreno, D., Ridley-Ellis, D.J., 2015. Comparing Usefulness of Acoustic Measurements on Standing Trees for Segregation by Timber Stiffness. In: General Technical Report FPLGTR-239. Madison, W.U.S.D.o.A., Forest Service, Forest Products Laboratory. (Ed.), 19th International Nondestructive Testing and Evaluation of Wood Symposium, Rio de Janeiro, Brazil, pp. 378-385.

* Noble fir material is comparatively younger due to difficulties in finding more mature plantations. ** for 100% yield

For allocation of timber to a certain strength class, the three properties must achieve the required values. Table 2 shows that stiffness is the main limiting property of the four species studied: the MoR and density satisfy values of higher strength classes. The given characteristic values in Table 2 define the minimum strength class achieved for the whole population (100% grading yield). Grading allows to sequentially remove the worst material so that the remaining population achieves a higher strength class for more mechanically demanding construction. So, at the expense of reducing the yield (percentage of the population graded to a certain strength class), higher strength classes can be achieved. Table 3 shows how the yields decrease for the material studied as the requirements increase. Table 3. Yields achieved for a rotation length of 45 years, with a perfect grading machine (Gil-Moreno et al., In press). C14 C16 C18 C20 C22 Norway 100% 100% 100% 84% 65% spruce Western 100% 100% 94% 80% 65% hemlock Noble fir

100%

96%

77%

62%

49%

Western red cedar

100%

96%

58%

41%

27%

As part of the wider study we are investigating how non-destructive testing can help us segregate material (Gil-Moreno and Ridley-Ellis, 2015) early in the timber production process to make the best possible use of the resource. 3

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 9

Sheep wool insulation for the absorption of volatile organic compounds E. Mansour, R. Marriott, G. Ormondroyd BioComposites Centre, Bangor University

ABSTRACT This paper summarises the quantitative analysis of the absorption of volatile organic compounds (VOCs), namely gaseous formaldehyde, toluene, limonene and dodecane, by different wool types. VOCs are of increasing concern due to their role as accumulating indoor air contaminants. The potential of sheep wool insulation as a sustainable and natural solution to this problem is demonstrated. Different wool types were tested via two different methods to assess their absorption ability with regard to these VOCs. For formaldehyde analysis, samples were subjected to cycles of exposure to excess formaldehyde gas whilst the weight gain was measured. For the other VOCs, a modified micro-chamber was used to flow the gases through samples, followed by trapping and quantification of the non-absorbed VOCs using Tenax TA tubes followed by their thermal desorption and gas chromatography. The data presented indicates that wool is able to absorb a range of VOCs in differing amounts and that the amounts and types of VOCs absorbed were dependent on the sheep breed and the processing of the wool. The data also shows that absorption occurs in a distinctive manner, more or less as a function of percentage of the amount wool fibres are exposed to rather in an indiscriminate absorptive manner; rather it buffers indoor VOC levels to smooth down spikes in concentrations over time. The use of wool in building design may therefore have an important role not only in imparting thermal efficiency but also in the improvement of indoor air quality. (European Respiratory Society, 2013). VOCs are frequently linked to what is termed “sick building syndrome” (SBS), which refers to a range of symptoms that include eye irritation, nasal congestion, dry skin, headache, fatigue, and difficulty in concentrating; for example, an estimated cost of $1 million was incurred due to SBS at the Environmental Protection Agency’s (EPA) U.S. headquarters due to decreased productivity (Wallace, 2001). In response, the industry introduced a wide range of ‘air cleaning/treating’ products to the market, and the removal of both chemical and biological indoor contaminants remains a subject of interest (Carslaw et al., 2013). However, such devices can be energy intensive, contribute to some other form of contamination, and have a short operational life span compared to the building’s life. It is possible that a passive solution can overcome such limitations. Sheep wool fibre has emerged as a niche construction product, specifically as an insulation material. In addition to its desired thermal and hygric properties that contribute to indoor comfort, this keratinous construction material is known to absorb volatile and very volatile organic compounds (Seo et al., 2009). Curling et al. showed via a quick and simple method that gaseous formaldehyde is absorbed by sheep wool (Curling et al., 2012).

1. INTRODUCTION Indoor air quality (IAQ) has intrigued scientists since the mid-1800s (von Pettenkofer, 1858). Historical events such as the London smog of 1952 instigated substantial air pollution investigations, and differences in the health of people working indoors and outdoors were explored (Fairbairn and Reid, 1958). There has been a mild interest in the capacity of construction materials to contribute to a better atmospheric environment (Braun and Wilson, 1970); but the main studies investigating volatile organic compounds (VOCs) in buildings didn't start until relatively recently, with at least 50 studies conducted between 1978 and 1990 (Brown et al., 1994). Unfortunately, indoor air pollution remains a recognised socio-economic problem (EEA, 2013; Franchi et al., 2006), potentially causing the loss of a projected $10 billion to $20 billion annually of projected savings and productivity gains in the US alone (Fisk and Rosenfeld, 1997). Based on further scientific findings, the World Health Organisation (WHO) compiled a set of statements emphasising the right to breathe healthy indoor air and the obligations of responsible authorities (WHO, 2000). According to the European Respiratory Society, pollutants “may have an important biological impact even at low concentrations over long exposure periods” 1

The aim of this study is to differentiate between the quantities of a representative range of VOCs absorbed by different wool types (Mansour et al., 2016, 2015). This will assess the ability of wool in the built environment to passively enhance indoor air quality in addition to its usual uses.

nitrogen into a 9.25% solution of formaldehyde and water. This flow can be adjusted to give differing partial pressures of the formaldehyde in the test chamber e.g. the amount of exposure to gaseous formaldehyde increases with increasing partial pressure. A micro-balance was used to detect any uptake of moisture and formaldehyde by the fibre. The sample was subjected to the following cycles to calculate the weight of formaldehyde that the wool was able to chemically bind with: a. Sample was left to equilibrate at 0% RH; i.e. it was not exposed to moisture or formaldehyde. This sets its baseline weight. Equilibration at all steps was based on a weight change of less than 0.002% over 10 minutes. b. Sample was left to equilibrate at 90% RH; i.e. it was exposed to high levels of moisture and formaldehyde where it sorbed both and gained weight. c. Sample was again equilibrated at 0% RH; at this point it lost all the water it had sorbed. Any weight gain relative to the sample’s state at step 1 was therefore attributable to sorbed formaldehyde. d. Steps a to c were repeated several times to determine the total sorption capacity.

2. WOOL SAMPLES AND VOC SELECTION To represent the widest range possible, 4 v/VOCs were taken into account in this study covering boiling points from -19 to 216.2°C, a large polarity range, and different chemical conformations: formaldehyde (simple and polar molecule), toluene (aromatic), limonene (cyclic and non-polar), and dodecane (straight chain and non-polar). Wools from a number of differing wool breeds were selected for use in the formaldehyde absorption study, with further comparison made to other species, as detailed below. All samples were obtained from commercially available sources. • Sheeps (Ovis aries) wool o Swaledale: A hardy United Kingdom (UK) hill sheep with a coarse durable wool predominately used for home furnishings and insulation. o Welsh Mountain: A hardy UK hill breed and the wool has been commonly used for home furnishings. o Light Herdwick: A hardy UK mountain breed with the wool commonly used for home furnishing and insulation. o Drysdale: A New Zealand breed noted for its coarse wool that is used in home furnishings. o Blackface: A UK mountain breed with the wool used mostly for home furnishing and tweed cloth • Comparison species o Alpaca. (Vicugna pacos) A llama like camelid, with the hair in this case obtained from domestic animals raised in the UK. In addition, the Swaledale and Light Herdwick samples were obtained and initially tested in their scoured and unscoured state. Scouring of wool is an alkali- and detergent- based washing process used to remove contamination material, grease and lanolin which could affect the properties of the wool. Based on the results of the formaldehyde study further selections were made to reduce the number of wool types tested to Swaledale, Blackface and Light Herdwick for the other 3 VOCs.

4. LIMONENE, TOLUENE AND DODECANE ANALYSIS Emissions generated from products can be tested using micro-chambers but to study the absorption potential however, certain modifications were applied to introduce gaseous toluene, limonene and dodecane. Due to differences in partial pressures, it was not possible to elute them from a highly compressed state. Therefore, sources were prepared containing the VOCs in their liquid form in steel tubes, sealed on one side and stoppered with a flexible plastic cover on the other side; this allowed the slow release of the VOCs in their gaseous form under controlled temperature and pressure. A flow of nitrogen gas, cleaned of any VOCs already present, was passed through the ≈100ml chamber containing the VOC sources. Thereafter the flow was controlled at 2.5±0.1ml/min, with an additional 2.5±0.1ml/min flow of clean nitrogen added. The controlled flow was fed into a vertical sample holder where the wool sample was held for 2.5 hours per run. This ensures that the VOCs pass through the whole sample and do not just interact with some of the surface of the sample. Any VOCs that were not absorbed by the sample were absorbed by 200mg Tenax TA contained in 89x6.4mm inert coated stainless steel Tenax TA tubes (Markes Int.), which was in turn analysed using a thermal desorber coupled to gas chromatography coupled to a flame ionisation detector (GC-FID, Perkin Elmer). The GC-FID was previously calibrated using Tenax TA tubes injected with known amounts of the three VOCs (7 differing amounts for each VOC ran in triplicates covering a

3. FORMALDEHYDE ANALYSIS Formaldehyde sorption analyses were performed using DVS system (Surface Measurement Systems, London, UK). Wool’s ability to absorb formaldehyde was thus assessed by the use of dynamic vapour sorption (Curling et al., 2012); a method that shows good repeatability. A flow of formaldehyde gas was produced by bubbling 2

range of 3.9ng to 312ng was used to obtain a linear calibration fit forced through an intercept of 0 and having adjusted R2 values of 0.9978 for toluene, 0.9989 for limonene and 0.9996 for dodecane). The injection of known amounts into the Tenax TA tubes was accomplished by injecting 0.5μl of differing concentrations of the VOCs dissolved in methanol whilst a clean nitrogen flow of 100ml/min was introduced for 15 minutes. 5. RESULTS AND DISCUSSION In comparing the uptake of the VOCs it should be remembered that two different methods were used. The formaldehyde sorption by DVS utilised an excess of formaldehyde and measured the cumulative uptake throughout multiple cycles of absorption/desorption, i.e. the wools’ total capacity to absorb gaseous formaldehyde; however, the micro-chamber approach utilised a constant flow of relatively low concentrations of limonene, toluene and dodecane (in the ng of VOC per g of wool range) and does not represent the wools’ total capacity to absorb these three VOCs but simulates real life situations. Figure 1 shows the amount of formaldehyde absorbed per kg of different wool types. It is evident that both wool type and condition (scoured or unscoured) have an effect on wool’s ability to absorb formaldehyde. It was also noticed that there is a general trend that the more darkly pigmented the wool fibre is, the higher its sorption capacity. Further research will determine if there is a correlation between fibre pigmentation of the same wool source.

Figure 2. Mass of toluene, limonene and dodecane absorbed by different wool types.

When the absorption of the studied VOCs is compared across the range of different wool types a trend can be observed based on the polarity of the VOC. For example, in the case of formaldehyde absorption, Swaledale wool absorb significantly less than Blackface. However, looking at the increasingly non-polar VOCs, Swaledale is seen to absorb more limonene and significantly more dodecane than Blackface does. This indicates that the polarity of the surface may be different between the wool types, leading to different levels of interactions between different VOCs depending on their polarity. Wool product producers may be able to take advantage of this phenomenon and create tailored products that absorb a specific range of those VOCs which the air in a certain type of building is known to contain high levels of. Unscoured wool was observed to absorb more of the tested VOCs than its scoured counterpart. This could be due to the presence of lanolin or contaminants that are removed during the scouring process, but it could also be due a modification of the wool surface due to the scouring process. Further tests were carried out to see the effect of the amount of VOCs wool fibre is exposed to in relation to its absorption. Using Light Herdwick wool, it was noted that increases in the amount of VOCs, most notably limonene and dodecane, the sample was exposed to at the same conditions were accompanied with increase in absorption, even at a much higher level, as seen in Figure 3. When we examine the percentage of the VOCs absorbed as a function of the total amount the wool was exposed to, we see that, within significance, the percentage is the same for each VOC as illustrated in Figure 4.

Figure 1. Mass of formaldehyde chemically bound by different wool types per kg of wool.

Figure 2 shows the amounts of toluene, limonene and dodecane absorbed per g of some wool types. As in the case of formaldehyde, it was evident that both wool type and condition (scoured or unscoured) had an effect on absorption. It is noteworthy that in the case of unscoured wools types, the total amount of limonene and dodecane that it was subjected to was completely absorbed, unlike the case of scoured wools which show residual peaks of unabsorbed material in their chromatograms.

3

References Braun, R.C., Wilson, M.J.G., 1970. The removal of atmospheric sulphur by building stones. Atmospheric Environ. 1967 4, 371–378. doi:10.1016/0004-6981(70)90082-X Brown, S.K., Sim, M.R., Abramson, M.J., Gray, C.N., 1994. Concentrations of Volatile Organic Compounds in Indoor Air – A Review. Indoor Air 4, 123–134. doi:10.1111/j.1600-0668.1994.t01-200007.x Carslaw, N., Hathway, A., Fletcher, L., Hamilton, J., Ingham, T., Noakes, C., 2013. Chemical versus biological contamination indoors: trade-offs versus win–win opportunities for improving indoor air quality. Indoor Air 23, 173–174. doi:10.1111/ina.12030 Curling, S.F., Loxton, C., Ormondroyd, G.A., 2012. A rapid method for investigating the absorption of formaldehyde from air by wool. J. Mater. Sci. 47, 3248–3251. doi:10.1007/s10853-011-6163-7 EEA, 2013. Air quality in Europe — 2013 report (No. 9/2013). European Environment Agency, Denmark. European Respiratory Society, 2013. European Lung White book [WWW Document]. ERS Respir. Health Dis. Eur. URL http://www.erswhitebook.org/ (accessed 9.6.13). Fairbairn, A.S., Reid, D.D., 1958. Air pollution and other local factors in respiratory disease. Br. J. Prev. Soc. Med. 12, 94–103. Fisk, W.J., Rosenfeld, A.H., 1997. Estimates of Improved Productivity and Health from Better Indoor Environments. Indoor Air 7, 158–172. doi:10.1111/j.1600-0668.1997.t01-1-00002.x Franchi, M., Carrer, P., Kotzias, D., Rameckers, E.M. a. L., Seppänen, O., Van Bronswijk, J.E.M.H., Viegi, G., Gilder, J.A., Valovirta, E., 2006. Working towards healthy air in dwellings in Europe. Allergy 61, 864–868. doi:10.1111/j.13989995.2006.01106.x Mansour, E., Curling, S., Ormondroyd, G., 2015. Absorption of formaldehyde by different wool types. Presented at the International Panel Products Symposium, Llandudno, UK. Mansour, E., Curling, S., Stéphan, A., Ormondroyd, G., 2016. Absorption of Volatile Organic Compounds by Different Wool Types. Green Mater. 4. doi:10.1680/jgrma.15.00031 Seo, J., Kato, S., Ataka, Y., Chino, S., 2009. Performance test for evaluating the reduction of VOCs in rooms and evaluating the lifetime of sorptive building materials. Build. Environ. 44, 207–215. doi:10.1016/j.buildenv.2008.02.013 von Pettenkofer, M., 1858. Besprechung allgemeiner auf die Ventilation bezüglicher Fragen. Cotta, Munich. Wallace, L.A., 2001. Human Exposure to Volatile Organic Pollutants: Implications for Indoor Air Studies. Annu. Rev. Energy Environ. 26, 269–301. doi:10.1146/annurev.energy.26.1.269 WHO, 2000. The right to healthy indoor air (Report on WHO Meeting No. Bilthoven, Netherlands). WHO.

Figure 3. Mass of toluene, limonene and dodecane absorbed by Light Herdwick wool at different levels of exposure.

Figure 4. Percentage of toluene, limonene and dodecane absorbed by Light Herdwick wool as a function of total amount of exposure.

5. CONCLUSIONS The data presented in this paper shows that wool, a natural and sustainable material, is able to absorb a range of potentially harmful chemicals from the indoor environment. This has important considerations for the prevention or reduction of sick building syndrome at a time when this issue is becoming more prevalent. The data indicates that the breed of sheep that provided the wool may have an important effect on the absorption of VOCs. Further observations is that the absorption by scoured wool is discriminate, equating to a percentage of the amount of VOC the fibres are exposed to; this indicated that sheep wool insulation does not simply absorb till a maximum limit and stops, but rather it buffers indoor VOC levels to smooth down spikes in concentrations over time. The use of wool in building design may therefore have an important role not only in imparting thermal efficiency but also in the improvement of indoor air quality. 5. ACKNOWLEDGMENTS The research leading to these results has received funding from the European Union's Seventh Framework Programme (FP7/2007-2013) for research, technological development and demonstration under grant agreement no 609234.

4

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 10

Sustainable Innovation: Engineered Bamboo for Construction B. Sharma Department of Architecture and Civil Engineering, University of Bath ABSTRACT A sustainable built environment requires innovation in a variety of construction materials. Conventional materials, such as steel and concrete, continue to be developed and be used globally. Interest in natural st materials is also growing rapidly. However to be fully utilised in 21 century construction, these materials require extensive research, along with the creation of standards and codes to be accepted in design and engineering practice. Through collaboration between academia and industry, we can work towards demonstrating the applications of innovative natural materials, such as engineered bamboo, in construction. Development of engineered bamboo for structural applications aims to increase use of the raw material in a laminated composite. With comparable structural properties, the material can be developed analogous to timber to increase adoption. The present work explores the potential for innovation, and briefly looks towards standardisation. Engineered bamboo is emerging as a novel material in construction and further research will establish the foundation needed to utilise this sustainable material in the built environment.

species varying by location. The bamboo culm is a natural composite, with fibres aligned within a lignin matrix, in the vertical direction of the culm. As a functionally graded material, the fibre distribution increases from the inner culm wall, with the greatest density found on the external culm wall. The material grows quickly, with full height reached within a year, after which the culm continues to gain strength, with optimal structural properties obtained within 3–5 years, depending on species. The rapid growth and renewability of bamboo are ideal characteristics for use in construction, yet the material is only marginally used. Challenges include geometric and material variability, degradation, and perception as a non-engineered material. Engineered bamboo aims to address these limitations to increase adoption in structural applications.

1. INTRODUCTION Natural materials are increasingly explored for use in construction as sustainable alternatives to conventional materials. To increase the sustainability of the built environment, natural materials offer low carbon alternatives to conventional materials. Expanding the concept of sustainability beyond the environment, there are potential economic and social benefits to using natural materials (John et al., 2005). For example, natural materials can be developed into prefabricated modular units and components that are shipped to site and assembled. The reduction in on-site construction time is a cost-benefit and has additional savings in terms of energy and generated waste. The opportunities exist to innovate in the way we build and that requires a systemic shift in thinking beyond low carbon materials.

The present work explores the manufacturing of engineered bamboo. The material and mechanical properties will be discussed and compared to other timber materials. The opportunities for innovation will be briefly explored, as well as the challenges in adoption of the material in construction.

The construction industry is conservative and does not easily adopt new materials or methods of construction. Furthermore, the perception of natural materials is often a barrier to adoption, as they are often viewed by stakeholders (designers, engineers, developers and the public), as low performing in comparison to ‘modern’ materials. This view leads to increased perceived risk when developing projects, which can in turn make the use of innovative materials non-competitive with conventional building products.

2. Manufacturing The general process used to manufacture engineered bamboo today is shown in Figure 1. The raw bamboo culm is cut into rectangular strips of varying thicknesses. The strips are then processed to achieve a colour preference through bleaching or thermal treatment also known as caramelisation. After a multi-day drying process, the strips are then laminated into boards. To create a structural section, boards can be laminated into the desired section size. This is the method that is

Bamboo is an example of such a material. In natural form, bamboo is a cylindrical pole, or culm, and is part of the grass family. There are over 1200 species of bamboo worldwide, with structural 1

typically used, however other products have been developed that utilise thinner strips, thereby increasing the use of the raw material (Xiao, 2016). While this is an example of innovation in the industry, further work is needed in developing optimised processing methods.

testing and standardisation for the material to be fully adopted. Table 1. Mean properties of engineered bamboo and products (adapted from Sharma et al., 2015). 𝑓𝑐 ∥ 𝑓𝑡 ∥ 𝜏∥ ρ 𝑓𝑏 kg/m3 MPa MPa MPa MPa Bleached 644 55 124 14 79 Bamboo Caramelised 686 77 90 16 83 Bamboo Raw Bamboo Phyllostachys 666 53 153 16 135 pubescens Sitka Spruce 383 36 59 9 67 Douglas-fir 520 57 49 11 68 LVL

Figure 1. General manufacturing process in China. Material efficiency estimates based on van der Lugt (2008).

timber 𝐸𝑏 GPa 10.2

10.8 9 8 13

5. Summary Engineered bamboo has strength properties that are comparable to or exceed that of timber. Development of the material for structural use will require additional testing and standardisation for the material to be fully adopted. Innovation in manufacturing methods will allow the material to be optimised and increase efficiency and reduce costs. Overall, engineered bamboo is novel material with structural properties that can be utilised as a sustainable alternative to conventional construction materials.

Automated methods would allow for the integration of efficient manufacturing and increased quality control. Furthermore, development of new equipment that combines processing steps would decrease energy and associated costs. 3. Material and Mechanical Properties The properties of engineered bamboo are comparable to or exceed that of timber, as shown in Table 1. The novel material has strength properties that vary depending on the processing method. Two materials, bleached and caramelised, engineered bamboo were investigated and the results indicated that the types of processing may be useful in creation of a hybrid composite to take advantage of the differences in strength between the thermal treatments (Sharma et al., 2015). The bending properties of engineered bamboo are comparable to that of timber and other construction materials, as shown in Figure 2. The material, however, demonstrates greater flexibility, with deflections well beyond serviceability limits. While a challenge for use in traditional frame construction, the bending properties are ideal for other types of structures, such as gridshells. These innovative structures can be employed in a variety of shelters, such as for exhibitions and shading.

References John, G., Clements-Croome, D., Jeronimidis, G., 2005. Sustainable building solutions: a review of lessons from the natural world. Building and Environment, 40: 319-328. Sharma, B., Gatóo, A., Ramage, M., 2015. Effect of processing methods on the mechanical properties of engineered bamboo. Construction and Building Materials, 83: 95-101. van der Lugt, P., 2008. Design Interventions for Stimulating Bamboo Commercialization: Dutch Design meets Bamboo as a Replicable Model. VSSD, Delft, Netherlands. Xiao, Y., 2016. Engineered Bamboo. In Nonconventional and Vernacular Construction Materials (Harries and Sharma eds.), Woodhead Publishing, pp. 433-448.

Figure 2. Comparison of bending modulus and strength for engineered bamboo and other construction materials.

Experimental testing has shown the properties of engineered bamboo are suitable for structural use, the development of the material requires additional 2

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 11

Early age shrinkage of sodium silicate activated slag paste: An evaluation of water evaporation mechanism Hossein Bahrami, Yi Pik Cheng and Yun Bai [email protected] Advanced & Innovative Materials (AIM) Group, Civil, Environmental & Geomatic Engineering, University College London, London, UK Abstract In recent years, alkali-activated slag (AAS) as an alternative low-carbon emission and high strength cementitious material has received increased attention. However, high shrinkage and crack tendency of AAS limit its wider industrial applications. The factors that influence cracking are known qualitatively, but it is not clear how to predict the initiation and propagation of cracks. This study presents a quantitative numerical method to model evaporation rate of water from the surface as the major influencing parameter to early age shrinkage. In this paper, various mechanisms suggested to cause early age shrinkage are presented. The mechanisms are evaluated from the point of view of their applicability to quantitative modelling of water evaporation. The capillary tension approach is easily applicable in a numerical model when dealing with a continuously changing microstructure. In order to test the numerical model, water evaporation, plastic viscosity and surface tension of sodium silicate activated slag and a portland cement pastes were measured during the first 24 hours. The isothermal heat evolution was also recorded to monitor the progress of hydration. Degree of saturation, viscosity, surface tension, relative humidity and temperature were used as input data for the calculation of water evaporation based on the capillary tension approach. Simulated moisture loss curves were in good agreement with the experiment for AAS pastes, while they underestimated the measured evaporation in PC paste after hydration peak.

1. Introduction

Alkali activated slag (AAS), using industrial byproducts containing silicate materials e.g. ground granulated blast furnace slag (GGBS), is an alternative to Portland cement (PC). Even though GGBS can also react directly with water, it requires an alkali activator in order to accelerate the reaction and produce good strength. Activation with liquid sodium silicate (waterglass) has been widely known to give rise to rapid hardening and high compressive strengths (Wang et al., 1994). The use of AAS has several superior advantages in terms of cost and performance. However, it has a higher shrinkage and cracking tendency compared to the PC which has prohibited its wider industrial applications. This study presents a quantitative numerical method to model evaporation rate of water from the surface of sodium silicate activated slag mixtures and compares it with that of the PC pastes.

Fresh concrete has high levels of volume changes through moisture movement caused by hydration, loss of moisture to the substrate, or evaporation to the atmosphere (Neville, 2011). Under rapid drying and hot weather conditions, early-age shrinkage cracks may appear in an early period ranging from placement up to 24 hours (Hammer, 2002). Such types of cracks may not endanger the present serviceability of the concrete structure, but can significantly affect its durability (Qi, 2003). The mechanisms leading to early-age shrinkage are poorly understood. While there is general agreement about the existence of a relationship between shrinkage and moisture loss of the cement paste, the actual mechanisms are unknown. Changes in the surface tension of the solid gel particles, viscosity, disjoining pressure and tension in capillary water are the principal factors that have been debated. For each of these parameters, knowledge of the development of the pore volume and pore size distribution, of the state of water in the capillary pores (free or adsorbed) and of the stiffness of the solid skeleton as hydration proceeds is needed.

2. Early-age shrinkage driving forces Hydration of cement is accompanied by a chemical shrinkage. The chemical shrinkage amounts typically to 6–7 ml/100 g of cement reacted (Powers and Brownyard, 1946). Not only hydration of the main clinker minerals but also the secondary 1

reactions, including formation of ettringite, result in chemical shrinkage (Mejlhede Jensen, 1993).

1. This test method can vary the relative humidity (RH) at any range of ambient temperatures. The specimen consists of two elements: 1) a plastic 3 mould with an inner size of 70 × 70 × 70 mm ; and 2) an envelope, formed by two PVC plates (called reflecting plates) attached to a plastic sheet, to assist measuring horizontal shrinkage. Pastes of PC and AAS were cast into this envelope and positioned inside the mould. The friction between the mould and the envelope was minimized by using oil. In this way, when the specimen shrinks, the reflecting plates can be dragged along by the cement paste. The shrinkage of the paste was continuously monitored every 1 min by two laser sensors located at 45mm from the bottom. To simultaneously monitor the evaporation, the entire assembly was positioned on a digital scale inside a glass desiccator as shown in Figure 1. Using this digital scale, the weight of the sample was continuously monitored throughout the whole drying cycle. By employing a vacuum pump and a pressure regulator, the air pressure inside the desiccator was lowered to around 1700 Pa (0.5 in Hg) and controlled at this level throughout the test.

As long as the cement paste is fluid, the autogenous shrinkage is totally converted into an external volume change. In other words, the fluid paste is not able to sustain the internal voids created by chemical shrinkage and contracts. When the hydrates percolate and the first solid paths are formed in the hardening paste, the stiffness increases and gas bubbles start to nucleate and grow in the bigger pores (Figure 2) (Hagymassy et al., 1969). This gives rise to the formation of water–air menisci and the RH drops (Kelvin’s law). The presence of menisci causes tensile stresses in the pore fluid (Laplace’s law). The RH drop results in a change in the thickness of the water layer adsorbed on the solid surfaces (Hagymassy et al., 1969). This is accompanied by changes in the surface tension of the solids, plastic viscosity and the disjoining pressure of adsorbed water between solid surfaces. At this early stage of hydration, the stiffness of the paste is so low and the viscous behaviour so pronounced that the slightest stress acting on the system results in a large deformation.

Figure 1. Schematic representation of the microstructure of cement paste during drying, (a) after setting, (b) after initial setting time, (c) formation of meniscus

Figure 2. Schematic of drying test setup

3.3 Isothermal conduction calorimetry

3. Experimental Programme

The heat evolution inside the samples may cause different evaporation rates. Therefore, the measurement of heat evolution is essential to better understand the various drying stages. Isothermal conduction calorimetry was performed on paste mixtures using a TAM Air isothermal calorimeter to determine the heat dissipated during the hydration reaction under a constant temperature condition (20°c). The sample container with the cement (i.e. slag or PC) and the syringe with the solution (i.e. waterglass or water) were maintained at 20°c prior to mixing. When thermal equilibrium was achieved, the cement and the solution were mixed by injecting the solution into the cement.

3.1 Materials and mix proportions Paste samples manufactured with two types of cementitious materials, namely, type II (CEM II) Portland cement and waterglass activated ground granulated blast furnace slag, were investigated. Two PC and AAS mixtures were formulated with a w/b ratio of 0.4. The modulus of sodium silicate was adjusted to 1.5 by adding solid sodium hydroxide. The hot solution was cooled down to room temperature and then stored for one day. The Na 2 O concentration in the AAS paste was formulated at a 4% Na 2 O equivalent by the mass of the slag.

4.

Results and discussions

3.2 Experimental Method

4.1 Heat evolution

The drying setup for the plastic shrinkage test under a low pressure condition is shown in Figure

Figure 3 shows the profile of the heat evolution for AAS and PC pastes during the first 96 hours and 2

24 hours respectively. After an initial peak due to the heat of wetting and initial chemical reactions, the PC reactions’ sequence begins from 1hour and reached the peak at about 10 hours. On the other hand, two major heat evolution peaks were observed in AAS: the first at 2–3 hours, and the second at around 65 hours. The first peak corresponds roughly to the time of setting; and the second peak with the onset of the development of strength (Brough and Atkinson, 2002). The second reaction sequence of AAS starts from 48 hours; which is not in the plastic shrinkage period.

If we take into account the RH drop both due to menisci formation and due to hydration process, the total RH can be approximated according to the following formula (Jensen and Hansen, 2001): 𝐻 = 𝐻𝑠 𝐻𝑘 = 𝑆 exp(−

2𝛾𝛾

𝜌𝜌𝜌𝜌

)

(4)

Where; S is saturation fraction which can be calculated as the ratio between the evaporable water content in the sealed hardening paste (Vev) and the total pore volume of the paste (Vp), which are both functions of the w/c ratio and of the degree of hydration α (Lura et al., 2003): 𝑆=

𝑉𝑒𝑒 (𝛼)

(5)

𝑉𝑝 (𝛼)

If we consider a Portland cement paste and apply Powers’ model to calculate the non-evaporable water and the total pore volume, Eq. (5) becomes (Jensen and Hansen, 2001):

𝑆=

a) b) Figure 3. Heat evolution a) PC paste b) AAS paste

p=

As a consequence of hydration, the cement paste forms a solid skeleton. From the formation of the solid skeleton onwards, the autogenous shrinkage is no longer totally transformed into external volume change. If the water supply is restricted, empty pores are formed inside the paste and air– water menisci occur. The air bubbles in the cement paste are formed in the bigger pores, which consequently empty first. Simultaneously, a drop in the RH occurs. The RH due to the meniscus formation in a circular cylindrical pore can be calculated according to Kelvin’s equation: 2𝛾𝛾 cos 𝜃 𝜌𝜌𝜌𝜌

)

2𝛾𝛾

ln(𝐻𝑘 )𝜌𝜌𝜌

2𝛾 r

𝑤 ( ) 𝑐 𝑤 𝜌 � �+( 𝑤 ) 𝑐 𝜌𝑐

𝛼=

(6)

(7)

𝑄

𝑄𝑝𝑝𝑝

(8)

In the calculation, special attention has to be given to the value of the potential heat of hydration, Qpot, which is the heat developed at complete hydration. The cement in this study has a potential heat of hydration of 645 kJ/kg.

(1)

4.3 Modified Powers’ volumetric model for AAS paste The different viscosity of sodium silicate solution may change the saturation factor in AAS. Slag modifies substantially the properties of the paste, especially the RH change and the autogenous deformation (Jensen and Hansen, 1996). To take the differences into account, Powers’ model modified for AAS was applied:

(2)

When the Kelvin radius is known, the tensile stress in the pore fluid can be calculated with the Laplace law for circular cylindrical pores: 𝜎𝑐𝑐𝑐 = −

𝑝−0.7(1−𝑝)𝛼

𝑝−0.5 (1−𝑝)𝛼

Where; w and c refer to masses of water and cement, respectively. The development of the degree of hydration α in time can be estimated by isothermal calorimetry, according to the following formula (Eq. (5)):

Where; 𝛾 is surface tension, M is molar weight of cement, 𝜃 is contact angle between water and solids, 𝜌 is density of water, r is radius of the -1 -1 meniscus, R is ideal gas constant (8.3 J mol K ) and T is absolute temperature. Assuming perfect wetting, the contact angle between water and solids is zero. Thus, cos 𝜃 = 1. With this assumption, the radius of the largest capillary pore filled with water can be calculated by: 𝑟=−

=

Where; 𝑉𝑐𝑐 (𝛼) is capillary water, 𝑉𝑖𝑖 (𝛼) interlayer water and 𝑉𝑐𝑐𝑐 (𝛼) chemical shrinkage water.

4.2 Autogenouse RH change and tensile stress

𝐻𝐾 = exp(−

𝑉𝑐𝑐 (𝛼)+𝑉𝑖𝑖 (𝛼)

𝑉𝑐𝑐 (𝛼)+𝑉𝑖𝑤 (𝛼)+𝑉𝑐𝑐𝑐 (𝛼)

(3)

3

𝑆=

𝑉𝑐𝑐 (𝛼)+𝑉𝑖𝑖 (𝛼)

𝑉𝑐𝑐 (𝛼)+𝑉𝑖𝑖 (𝛼)+𝑉𝑐𝑐𝑐 (𝛼)

p=

=

𝑝−0.7𝑘(1−𝑝)𝛼 𝜈 𝑝−𝑘(0.5−0.35( 𝑠 ))(1−𝑝)𝛼

𝑤 ( ) 𝑐 𝑤 𝜌𝑤 𝜈𝑠 � �+( )( ) 𝑐 𝜌𝑐 𝜈𝑤

k=

1 𝜈 1+1.43( 𝑠 )

casting. These quantities were used as input for the calculation of evaporation based on the capillary tension approach. Because the RH change is due to viscosity as well as to meniscus formation, a method was developed to separate these two effects. This allowed the calculation of the actual stress in the pore fluid that is associated with menisci formation. The calorimetric data were used to calculate the development of the degree of hydration. With Powers’ volumetric model for cement hydration and Powers’ volumetric model modified for AAS, the saturation fraction was estimated. The evaporation rate was calculated on the basis of the capillary tension, the saturation fraction. The Simulated moisture loss curves were in good agreement with the experiment for AAS pastes, while they underestimated the measured evaporation in PC paste after hydration peak.

(9)

𝜈𝑤

(10) (11)

𝜈𝑤

Where 𝜈𝑠 is viscosity of solution.

4.4 Calculation of water evaporation The water evaporation of the cement paste was calculated according to Eq. (5), where H was calculated with Equ. (4), S with Equ. (5). In Figure 4, the evaporation rate is plotted against the time for initial 24 hours. The calculated values agree reasonably well with the experimental ones up to 15 hours. Then, the measured evaporation rate decreases sharply while the calculated one reduces gradually, which is due both to the flattening out of the hydration.

References BROUGH, A. R. & ATKINSON, A. 2002. Sodium silicate-based, alkali-activated slag mortars: Part I. Strength, hydration and microstructure. Cement and Concrete Research, 32, 865-879. HAGYMASSY, J., BRUNAUER, S. & MIKHAIL, R. S. 1969. Pore structure analysis by water vapor adsorption: I. t-Curves for water vapor. Journal of Colloid and Interface Science, 29, 485-491. HAMMER, T. A. The relationship between settlement and plastic shrinkage of high strength concrete. Proceedings of the RILEM International Conference on Early Age Cracking of Cementitious Systems. National Building Research Institute/Technion, Haifa, 2002. 89102. JENSEN, M. & HANSEN, P. F. 1996. Autog enous deformation and change of the relative humidity in silica fume-modified cement paste. ACI Materials Journal, 93. JENSEN, O. M. & HANSEN, P. F. 2001. Waterentrained cement-based materials: I. Principles and theoretical background. Cement and concrete research, 31, 647-654. LURA, P., JENSEN, O. M. & VAN BREUGEL, K. 2003. Autogenous shrinkage in highperformance cement paste: An evaluation of basic mechanisms. Cement and Concrete Research, 33, 223-232. MEJLHEDE JENSEN, O. 1993. Autogenous deformation and RH-change-selfdesiccation and selfdesiccation shrinkage. TR, 284, 93. NEVILLE, A. M. 2011. Properties of concrete, Harlow, England ; New York, Pearson. POWERS, T. C. & BROWNYARD, T. L. Studies of the physical properties of hardened Portland cement paste. ACI Journal Proceedings, 1946. ACI. QI, C. 2003. Quantitative assessment of plastic shrinkage cracking and its impact on the corrosion of steel reinforcement. Purdue University.

It must be noticed that Eq. (3) supposes an elastic behaviour of the cement paste. In fact, in the first stage of drying, the slope of the measured shrinkage decreases, corresponding to an increase of the elastic modulus of the cement paste. Afterwards, the evaporation drop slows and the deformation per percent RH increases. This fact could be due to the onset of the latent hydraulic property of slag.

Figure 1. Measured and calculated evaporation rate of the AAS and PC pastes as a function of time

5. Conclusions This study presented a quantitative numerical method to model evaporation rate of water from the surface as the major influencing parameter to early age shrinkage. Evaporation rate of water was calculated with a combined experimental and numerical approach. Isothermal heat evolution, elastic modulus in compression and ambient conditions were measured in the first 24 hours after 4

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 12

Experimental investigation on hygrothermal performance of recycled aggregate concrete Marina Alagušić, Ivana Banjad Pečur Department of Materials, Faculty of Civil Engineering, University of Zagreb, Croatia

ABSTRACT Increasing demand on energy efficiency and natural resources conservation is challenging the construction sector. Construction and demolition waste can gain added value through its reuse as aggregate in concrete, thus it could contribute to the natural resources conservation. Prefabricated ventilated concrete wall panel, produced by using recycled aggregate, is a benchmark construction product that unites reduction of energy consumption and sustainable use of construction and demolition waste. How building envelope built with this kind of panels perform when exposed to the real weather conditions and real conditions of use, i.e. people living inside the building, defines its hygrothermal performance (combined heat, air and moisture transfer). Hygrothermal performance of a building envelope is reflected in temperature and moisture content variations within the building assemblies, thus directly affecting the energy consumption, durability of building envelope and comfort of inhabitants. Since there is no experimental confirmation on hygrothermal performance of prefabricated ventilated concrete wall panels, produced using the recycled aggregate, experimental set-up for hygrothermal monitoring was designed in order to provide a deeper insight in heat, air and moisture transfer through this kind of panels. Member States to take any necessary measures to achieve a minimum target of 70% (by weight) of CDW by 2020 for preparation for reuse, recycling and other material recovery, including backfilling operations using non-hazardous CDW to substitute other materials (European Parliament, 2008). CDW could gain added value through its use for the production of recycled aggregate concrete.

1. INTRODUCTION Construction sector is facing two main challenges of the 21st century: energy poverty and depletion of natural resources. Efficient use of energy and natural resource conservation, reflected in 6th and 7th Basic requirement for construction work (European Parliament, 2011), are the prerequisite for achieving sustainability and innovation in construction sector.

With the aim to develop benchmark construction product that unites reduction of energy consumption and sustainable use of CDW, extensive research has been carried out by the research group at Department of Materials, Faculty of Civil Engineering, University of Zagreb. Conducted research resulted with the innovative prefabricated ventilated ECO-SANDWICH® (EC®) wall panels.

Since the most buildings are ‘’sub-standard’’ in terms of energy consumption, EU strategy Europe 2020 (European Commission, 2010), Energy Performance of Buildings Directive and its Recast (European Parliament, 2010) have established legal framework for energy efficiency in buildings through strictly defined energy requirements for different type of buildings. The first step towards energy efficient buildings is design and construction of energy efficient building envelope.

For detail understanding on the behavior of building envelope, its hygrothermal performance have to be known. In case of prefabricated ventilated wall panels, i.e. for the same type of sandwich wall panels as EC® panels, dynamic hygrothermal performance is only assumed, but not experimentally confirmed.

The second challenge that construction sector, but also the whole Europe, is facing to is natural resource conservation. European Commission (EC) identified construction and demolition waste (CDW) as a priority waste stream since it represents approximately 34% - 49% (Eurostat, 2014; Sáez et al, 2011) of the total waste generation in the EU, which makes it dominated EU waste. EC established the legislative framework for waste management through The Waste Framework Directive (WFD). WFD requires

2. ECO-SANDWICH® panel EC® panel is innovative prefabricated ventilated wall panel with integrated formaldehyde-free core insulation allowing very low energy design and retrofit of buildings. It consist of two recycled 1

aggregate concrete layers interconnected through stainless lattice girders, Figure 1.

the envelope, which occurs due to differences between the external environment and the interior of the building (internal environment). Primarly it refers to the different temperature conditions and different conditions of relative humidity, i.e. different partial pressures of water vapour in air, of the external and internal environment. These conditions represent the boundary conditions of the building envelope. The outer boundary conditions of the envelope are determined by the climate that shows the repeatability for the particular building location. The influence of weather on the building is therefore taken into account by choosing representative meteorological data for a particular building location. The inner boundary conditions are determined by the use type of the building and the user habits. In the average household 10 liters of water is generated as water vapour per day. It is necessary to control the amount of moisture inside the buildings in such way that the excess moisture is brought outside the building (natural ventilation or mechanical ventilation), in order to ensure comfort and hygienic conditions and to prevent the occurrence of construction damage.

a) b) ® Figure 1. EC wall system: a) Model; b) Cross section

The inner (self-load bearing) layer is made of recycled concrete aggregates while the outer façade layer is made of recycled brick aggregates. Replacement ratio of natural aggregate in both concrete layers is 50%. To prevent possibility of water vapour condensation, layer of ventilated air is placed between layer of thermal insulation and outer façade layer. Therefore, EC® combines both functions, the one of ventilated façade and the other one of external wall. EC® represents possible technological solution for fast construction of very low energy or passive house standard (A+) buildings on a large scale. It tackles three major environmental problems: reducing greenhouse gas emission by enhancing energy efficiency of buildings, increased resource efficiency through the use of CDW in panel manufacturing and minimising the use of regulated chemicals like phenol and formaldehyde from the insulation material production process.

Building envelope is a '’skin’' of the building and it is exposed to the external loads from the environment, both the mechanical and hygrothermal ones. Through the ‘'skin'’, i.e. its envelope, building resists loads and strive to establish a balance in terms of mechanical stability and resistance and in terms of dynamic hygrothermal balance. In addition to the external and inner boundary conditions, significant role in hygrothermal performance have the properties of characteristic layers of building envelope. The characteristic layers with its physical, mechanical and hygrothermal properties considerably affect the dynamics of the heat, air and moisture transfer, thereby defining hygrothermal performance of the building as a whole (Künzel and Holm, 2009).

In order to optimize concrete mixtures and determine the most favourable replacement ratio of natural aggregates with CDW, comprehensive research on mechanical, durability and thermal properties was performed. In the mixtures, proportions of the recycled aggregate of 40%, 50% and 60% were used, as they were deemed to hold the best balance between environmental, structural, durability and thermal requirements. Mixture with 50% of replacement of coarse aggregate was chosen as optimal for the EC® wall panels (Alagusic et al, 2016; Banjad Pečur et al, 2015).

So, the appropriate moisture control is a prerequisite for the design of energy efficient buildings and buildings without damage. Investigation on hygrothermal performance of ventilated facades is an active field of actual research. Review of the literature revealed that all the studies regarding hygrothermal performance of ventilated facades have been conducted on the wooden ventilated facades which is common type of building in North America and Northern Europe (Salonvarra et al, 2007; Künzel et al, 2008; Grau and Rode, 2006; Tichy and Murray, 2007) or on ventilated facades with brick walls (Van Belleghem et al, 2015; Karagiozis and Künzel, 2009; Hens and Janssens, 2007). Al-Neshawy (Al-Neshawy, 2013) conducted a study on hygrothermal performance of prefabricated concrete sandwich panels with a core of thermal insulation, however, this kind of sandwich system has no ventilated layer and cannot be compared with EC® panels.

3. Hygrothermal performance When considering the building envelope, 75 - 90% of all construction damage is caused by moisture (Milovanović and Mikulić, 2011). Moisture transfer (water vapour contained in the air and/or liquid moisture) through the building envelope cannot be fully prevented and it takes place simultaneously with the heat transfer. Moisture and heat transfer reflect by changes in temperature and moisture content within the characteristic layers of the building envelope. Thereby, hygrothermal performance of the building envelope implies a combined heat, air and moisture transfer through 2

Currently there is no available research dealing with hygrothermal performance of previously described prefabricated ventilated wall panels made of recycled aggregate concrete. Künzel et al. (Künzel et al, 2008) emphasize the necessity of knowing the hygrothermal properties of concrete since the moisture is a key factor influencing the pathology of building elements made of concrete. Also, moisture is a key factor influencing the behaviour of concrete from the aspects of energy efficiency during its service life.

Figure 3. a) Set of T-MC-RH sensors; b) Installation of sets of T-MC-RH sensors on girders; c) Arrangement of T-MC-RH ® sensors within EC layers

Since the building envelope consists of a large number of EC® panels, measurement locations had to be selected and reduced. Three panels in ground floor, with different sun orientations, were selected as measurement locations and marked as M1-M3, Error! Reference source not found. 4. As it was previously mentioned, at every measurement location, sensors needed for hygrothermal monitoring were positioned within the all characteristic layers of EC®.

Currently, there is a lack of knowledge on how recycled aggregate concrete and how ventilated prefabricated sandwich wall panels behave when exposed to the weather and real use conditions, i.e. there is a lack of knowledge on their hygrothermal performance. To gain a deeper insight into hygrothermal performance of EC® panels, a program of experimental monitoring on hygrothermal performance of a building envelope constructed with EC® panels is designed. 4. Experimental set-up At the moment, a 3-storey family house (A+ energy class) is being constructed with EC® panels within socially - supported housing program in Koprivnica, Republic of Croatia (Figure 2). House is planned for sale, which provided the unique opportunity to install a monitoring system for hygrothermal performance of a building envelope in real conditions of use, i.e. people living inside the house.

Figure 4. Measurement locations

b) a) ® Figure 2. EC house: a) Position plan for southern façade; b) Construction site in March 2016

Presented monitoring system is a combination of wire - based measurement system and wireless system. Every set of sensors (S1-S7) at every measurement location (M1-M3) will be connected to the central measurement unit, where sensors readings will be recorded. Data will be stored in online database, so that the information could be accessed via a web access.

The objective of this experimental research is to gain a deeper insight in heat and moisture transfer through the building envelope built with EC® panels, i.e. to obtain temperature and moisture content distribution and variation within all characteristic layers of prefabricated ventilated wall systems in a long-term period.

5. Conclusion It is expected that experimental monitoring of temperature and moisture variations in all characteristic layers of EC® panels will provide a dynamic real-time hygrothermal performance of building envelope constructed with the presented panels under the real terms of use, i.e. people living inside the building, and exposure to the real outdoor conditions.

Monitoring system for hygrothermal performance consists of a set of temperature (T), moisture content (MC) and relative humidity (RH) sensors (Figure 3a). The set of T-MC-RH sensors is fixed on a girder and positioned through the all characteristic layers of EC® panel. Positions of set of T-MC-RH sensors are marked as S1-S7, Figure 3c.

By controlling its hygrothermal performance, building envelope together with HVAC systems, through the smart building automatization system, can be actively involved in a regulation of consumption of energy needed for heating and cooling.

3

Hens, H., Janssens, A., Depraetere, W., Carmeliet, J., Lecompte, J., 2007. ‘’Brick Cavity Walls – A Performance Analysis Based on Measurements and Simulations’’, Journal of Building Physics 31 (2007) 2, pp. 95-124 Karagiozis, N. A., Künzel, H. M., 2009. ‘’The Effect of Air Cavity Convection on the Wetting and Drying Behavior of Wood-Frame Walls Using a Multi-Physics Approach’’, Journal of ASTM International 6 (2009) 10 Künzel, H. M., Holm, H. A., Krus, M., 2008. ‘’Hygrothermal Properties and Behaviour of Concrete’’, WTA-Almanach, pp. 161-181, 2008. Künzel, H. M., Karagiozis, A. N., Kehrer, M., 2008. ‘’Assessing the benefits of cavity ventilation by hygrothermal simulation’’, Proceedings Building Physics Symposium in honour of Prof. Hugo Hens, Leuven, pp. 17-20 Künzel, M. H., Holm, H. A., 2009. ‘’Moisture Control and Problem Analysis of Heritage Constructions’’, PATORREB, Porto, pp. 85-102 Milovanović, B., Mikulić, D., 2011. ‘’Assessment Method for Combined Heat, Air and Moisture Transfer in Building Component’’, Energy Management in Cultural Heritage, Dubrovnik Salonvarra, M., Karagiozis, A .N., Pazera, M., Miller W. A., 2007. ‘’Air Cavities Behind Claddings-What Have We Learned?’’, Thermal Performance of the Exterior Envelopes of Buildings X, proceedings of ASHRAE THERM X, Clearwater, FL Sáez P.V., del Río Merino M., Amores C. P.and de San Antonio G.A., 2011. ‘’European Legislation and Implementation Measures in the Management of Construction and Demolition Waste´, The Open Construction and Building Technology Journal, (5), (Suppl 2-M6), pp. 156161 Tichy, R., Murray, C., 2007. ‘’Developing Innovative Wall Systems that Improve Hygrothermal Performance of Residential Buildings Progress Report’’, Final Report, Washington State University Van Belleghem, M., Steeman, M., Janssens, A., De Paepe, M., 2015. ‘’Heat, air and moisture transport modelling in ventilated cavity walls’’, Journal of Building Physics 38 (2015) 4, pp. 317349

In addition, hygrothermal monitoring can provide monitoring of a material degradation, which is significant since 75 - 90 % of all damage in building envelope are caused by the moisture. Knowing the degradation process of a material, it could ensure that timely and suitable rehabilitation of a building is performed, i.e. the long term performance and thus the service life of a monitored building could be extended.

References Alagusic, Marina; Milovanovic, Bojan; Banjad Pecur, Ivana, 2016. ‘’Recycled aggregate concrete –sustainable use of construction and demolition waste and reduction of energy consumption’’, Advances in Cement and Concrete Technology in Africa Proceedings 2nd International Conference, Dar el Salaam, pp. 253-262 Al-Neshawy, F., 2013. ‘’Computerised prediction of the deterioration of concrete building facade caused by moisture and changes in temperature’’, Doctoral dissertation 96/2013, Aalto University, School of Engineering, Department of Civil and Structural Engineering, Finland Banjad Pečur, I., Štirmer, N., Milovanović, B., 2015. ‘’Recycled aggregate concrete for nearly zero-energy buildings’’, Magazine of Concrete Research, 2015 67(11), pp. 575-584 European Parliament, 2008. Directive 2008/98/EC of the European Parliament and of the Council of 19 November 2008 on waste and repealing certain Directives, OJEU L312/3 European Commission, 2010. Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the regions Energy 2020 A strategy for competitive, sustainable and secure energy, COM(2010)639 final, Brussels European Parliament, 2010. Council of the EU: Directive 2010/31/EU of the European Parliament and of the Council of 19 May 2010 on the energy performance of buildings (recast), OJEU L153/13 European Parliament, Council of the European Union, 2011. Regulation (EU) No 305/2011 of the European Parliament and of the Council of 9 March 2011 laying down harmonised conditions for the marketing of construction products and repealing Council Directive 89/106/EEC. OJEU, L 88/5 EUROSTAT, 2014. Generation of waste, acc. 07/07/2014http://epp.eurostat.ec.europa.eu/portal /page/portal/eurostat/home Grau, K., Rode, C., 2006. ‘’A model for air flow in ventilated cavities implemented in a tool for whole-building hygrothermal analysis’’, Annual report year: 2006, Danish Building Research Institute 4

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 13

Effect of Anionic Asphalt Emulsion on the Hydration of Portland Cement 1,2,3

1,2,3

3

3

2

1,3

Wei Li , Jiaping Liu , Jinxiang Hong , Xiaobin Zhu , Yun Bai , Changwen Miao 1.School of Materials Science and Engineering, Southeast University, Nanjing, China. 2.Advanced & Innovative Materials (AIM) Group,University College London,London, UK. 3.State Key Laboratory of High Performance Civil Engineering Materials, Nanjing, China. ABSTRACT Cement-asphalt (CA) mortar, a grouting material consisting of Portland cement (PC), asphalt emulsion, sand, water and other related admixtures, has widely been used as a cushion layer material in the construction of high-speed railways (HSR) in the past decade in China due to its excellent damping effect. The fresh and hardened properties of type II CA mortar (which is basically an asphalt emulsion modified PC mortar) are closely related to the PC hydration in the presence of asphalt emulsion. However, the effect of asphalt emulsion on the hydration process of PC has not been fully understood. In this paper, the hydration process of cement-asphalt mixtures was investigated at a fixed water to cement ratio of 0.41 with different asphalt to PC (A/C) ratios, namely 0, 0.16, 0.24 and 0.32, by isothermal conduction calorimeter. Portland cement complying with Chinese standard GB8076-2008 and anionic asphalt emulsion were employed as raw materials in this research. Setting time and electrical resistivity were also conducted to study the setting behavior and microstructure evolution during the early stage of hydration, respectively. The results showed that: (1) the initial and final setting time of the PC were prolonged with the increase of A/C; (2) the induction period of the PC was prolonged with the increase of A/C which has also been corroborated by the setting time results; (3) in the presence of asphalt emulsion, the electrical resistivity of the PC pastes was initially increased before the final setting whilst became lower afterwards, as compared to the pure PC control mix. Therefore, the experimental results clearly indicated that the cement hydration process was retarded by the asphalt emulsion and the retardation mechanism could be attributed to the following two aspects: (i) the active sites of aluminate phase and some silicate phases could have been occupied by asphalt emulsion through anionic emulsifier via electrostatic interaction. (ii) the surfaces of cement particles might have been covered by asphalt membrane due to the demulsification of asphalt emulsion, which could have caused the retardation of the cement hydration process. investigate the effect of asphalt emulsion on cement hydration. 2. EXPERIMENTS AND METHODS 2.1 Materials Portland cement complying with the Chinese standard GB8076-2008 produced by Qufu China United Cement Co., LTD was used in this study. Anionic asphalt emulsion(AE-601) was supplied by Jiangsu Bote New Materials Co., Ltd. 2.2 Sample preparation The heat of hydration of the cement-asphalt mixtures were investigated at a fixed water to cement ratio of 0.41 with different asphalt to PC (A/C) ratios, namely 0, 0.16, 0.24 and 0.32. The amount of water in the asphalt emulsion was also considered in the calculation of water to cement ratio. The CA pastes for the electrical resistivity test were prepared with W/C of 0.41, and A/C between 0 and 0.24, which were consistent with the hydration heat test. 2.3 Test methods 2.3.1 Setting time measurement Setting time was measured according to Chinese standard GB 1346:2011. Both the initial and final

1. INTRODUCTION As a kind of inorganic-organic composite, cement-asphalt (hereafter CA) mortar is composed of Portland cement, asphalt emulsion, water and other related admixtures, which combines the both merits of the high compressive strength of cement and the superb flexibility of asphalt (Jian and Yiqiu, 2015). Therefore, CA mortar has been mainly used as a kind of grouting material for the cushion layer in the slab track system of high-speed railway (HSR) due to its excellent damping effect (Esveld, 2003). The fresh and hardened properties of type II CA mortar (which is basically an asphalt emulsion modified PC mortar) are closely related to the hydration of PC in the presence of asphalt emulsion. However, the effect of asphalt emulsion on the hydration process of PC has not been fully understood. Therefore, this paper aims at identifying the action mechanism of asphalt emulsion on the PC hydration process. Setting time, heat of hydration and electrical resistivity tests were employed to 1

setting time of the pastes with different dosages of asphalt emulsion AE-601 were measured, and the test temperature was kept at 20℃. 2.3.2 Hydration heat measurement The fresh cement paste was put into a plastic bottle within 10 mins after being prepared according to the above formulations, and the heat of hydration was then measured by a self-regulated semi-adiabatic calorimeter. The temperature equilibrium should be reached at 20℃ between the specimen and the instrument before conducting the measurement, and the testing period lasted for about 72hrs. 2.3.3 Electrical resistivity measurement Electrical resistivity can reflect the cement hydration process and the evolution of microstructure well. Therefore, the Non-Contact Impedance Measurement (NCIM) CCR-Π (Fig1.) was employed to monitor the effect of asphalt emulsion on cement hydration. Prior to any test, the instrument was first adjusted to more than 30000Ω and the fresh cement paste was then poured into the ring-shaped plastic mould which was then allowed by a series of automatic measurements.

420 400 380

Setting time/min

360

Initial setting Final setting

340 320 300 280 260 240 220 200 180 0.00

0.08

0.16

0.24

0.32

The mass ratio of asphalt emulsion to cement

Figure 2. Effect of asphalt emulsion on setting time

In order to understand the action mechanism of asphalt emulsion on cement hydration, the measurements of cement hydration heat and electrical resistivity were employed and the results are reported below. 3.2 Effect of asphalt emulsion on the hydration heat It is well established that cement hydration is an exothermal process which is always accompanied by the release of hydration heat. Hence, by following the evolution of the heat generated during the cement hydration, it is possible to identify the effect of asphalt emulsion on cement hydration process. Figure3 presents the heat flow during the first 70 hours of cement hydration at different A/C, whilst Figure 3(b) depicts the details of heat evolution during the first 10 hours. From Fig.3(b), it can be seen that the induction period prolonged with the increase of A/C. This could be attributed to the following reasons: during the early cement hydration process, the ion dissolution rate of C 3 A was much quicker than that of the silicate phases, due to the higher reactivity of aluminate phase, which led to the concentration of calcium quickly reached saturation. Therefore, the AFt was formed during the early stage of cement hydration around the cement grains, which hindered the further dissolution of C 3 A. During 2+ the induction period, the concentration of Ca increased slowly over time, and at the acceleration period of cement hydration, large amount of C-S-H 2+ was formed until the concentration of Ca reached saturation. Therefore, the duration of induction period mainly depends on the concentration of 2+ quicker the increase of the Ca , the 2+ concentration of Ca , the shorter the duration of the induction period of cement hydration. It is generally agreed that the surface of aluminate phase shows positive charge after mixing with water, which is mainly due to the discrepancy of the migration rate of different kinds of ions from cement particles into solution according to the analysis in our previous work (Wei, Xiaobin and Jinxiang, 2015). Hence, the anionic asphalt emulsion can adsorb onto the positively charged sites of cement particles through electrostatic interaction. Consequently, the surface of cement particles were covered by the asphalt membrane which would hinder the further dissolution process. Further details about this process are discussed 2+ below. As a result, the concentration of Ca increased slowly with the increase of A/C during the

Figure 1.The Non-Contact Impedance Measurement employed in the investigation

3. RESULTS AND DISCUSSIONS 3.1 Effect of asphalt emulsion on setting time The essence of setting is a percolation process of cement particles which connect with each other through the formation of hydration products, and the setting process is usually used to depict the connective lever of cement particles (J.Bensted, P.Barnes, 2008). According to the state of cement paste, the setting time includes the initial setting time and final setting time. The initial setting time depicts the stage when cement paste converts from the flowable state to the beginning of the hardening state, which is closely related to the rheological property of fresh paste, while the final setting time indicates the formation of hardened cement paste. Therefore, setting time has been considered as an important parameter to reflect the hydration process of cement at the early stage. Figure2 shows that both the initial and final setting times were prolonged with the increase of A/C from 0 to 0.32, which implies that cement hydration was retarded by asphalt emulsion.

2

coating formed on the surface of cement grains, leading to a reduction of the dissolution of cement and a lower ion concentration. This could explain the higher resistivity of CA paste in Figure 4.

induction period, which resulted in the induction period being prolonged. In other words, the cement hydration process was retarded in the presence of anionic asphalt emulsion. 14

4.5 12

4.0

Heat flow-C-0.41(mw) Heat flow-601(0.16)-0.41 mw/g Heat flow-601(0.24)-0.41 mw/g Heat flow-601(0.32)-0.41 mw/g

8 6

a

4

Resty2-W-C Resty2-EA601-A/C-0.24

3.5

Resistivity

Hydration heat/mw

10

3.0 2.5 2.0 1.5

2

1.0

0 0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

0.5

Hydration time/h

6

0

100 200 300 400 500 600 700 800 900 1000 1100 1200 1300

Hydration time/min

Hydration heat/mw

Figure 4. Effect of asphalt emulsion on electrical resistivity

From Figure 4, it can also be seen that during the induction period, the electrical resistivity of both CA paste and pure PC paste kept reasonably stable. However, the induction period of CA paste lasted longer which could again be attributed to the reduced dissolution rate of CA paste due to the asphalt coating formed on the surface of cement grains. On the contrary, during the accelerated hydration period, the electrical resistivity of the CA paste was lower than that of the pure PC paste which would indicate the microstructure formed in CA paste is more porous than that of the PC paste. This, again, could be due to the barrier formed by the asphalt on the surface of cement grains which has adversely affected the dissolution process of cement and, hence, the formation of microstructure of CA paste.

Heat flow-C-0.41(mw) Heat flow-601(0.16)-0.41 mw/g Heat flow-601(0.24)-0.41 mw/g Heat flow-601(0.32)-0.41 mw/g

4

b

2

0 0

2

4

6

8

10

Hydration time/h

Figure 3. Effect of asphalt emulsion on hydration heat

3.3 Effect of asphalt emulsion on the electrical resistivity It is well known that cement paste is a typical porous material filled with different kinds of ions, and the electrical resistivity is mainly controlled by the ion concentration and the ion migration path which in turn, is closely related to the microstructure of cement paste (Lianzhen, 2007). Therefore, the electrical resistivity can be used to follow the cement hydration process from the aspect of the evolution of microstructure over time. Based on the theoretical analysis and some references (G.E.Archie, 1942), the electrical resistivity of cement paste confirms to the Archie’s law as below: -m(t) ρ(t)= ρ 0 (t)*φ(t) Where, ρ(t) and ρ 0 (t), are the electrical resistivity of cement paste and the pore solution respectively. φ stands for the porosity of cement paste. t, the hydration time and m, a constant. Figure 4 shows the evolution of electrical resistivity of the cement paste over time. During the early hydration period, the electrical resistivity of both of CA paste and the pure PC paste all decreased to a minimum value after 100min and 50min, respectively, and then started to increase at different rate afterwards. Moreover, the electrical resistivity of the CA paste was higher than that of the pure PC paste. It is known that, during the early hydration stage, the electrical conductivity (and hence the resistivity) of the fresh cement paste is mainly determined by the ion concentration in fresh paste which, in turn, is controlled by the dissolution process. However, as highlighted before, the ion dissolution process of CA paste could have been hindered by the asphalt

3.4 The retardation mechanism of asphalt emulsion on cement hydration From the setting time, heat of hydration and electrical resistivity data presented before, it can be concluded that all these properties are affected by the interaction between asphalt emulsion and cement particles at the early hydration stage. Therefore, the possible interaction mechanism between asphalt emulsion and cement particles are proposed and discussed below. Immediately upon in contact with water, different mineral phases or hydration products would generate different charges. For example, the aluminate phase, C 3 A, usually shows a positive charge with a zeta potential of +4.15mv, while the silicate phases, C 2 S and C 3 S, show negative charge. However, partial areas of silicate phases may give positive charge due to the complex 2+ 2existing between calcium ions Ca and H 2 SiO 4 around the surface of silicate phases via electrostatic attraction (Johann, Christian, 2007; H. Viallis, A. Nonat, 2001). The asphalt emulsion (AE-601) used in this study was negatively charged because of the anionic emulsifier (EA-601) employed to emulsify the asphalt. Therefore, asphalt emulsion AE-601 can be selectively adsorbed onto those positively charged sites on the surface of cement grains, such as those sites occupied by aluminate phase 3

and partial silicate phases, through emulsifier via electrostatic adsorption. This layer of adsorbed asphalt emulsion would then hinder the dissolution and subsequent deposition reactions on the surface of cement particles. Figure 5 further illustrates this adsorption behavior of emulsifier EA601 onto the cement particles.

and the electrical resistivity of hardened CA paste was lower than that of the pure PC paste. (4)The retardation mechanism of asphalt emulsion caused to the hydration of cement could be attributed to the following two aspects: (i) the active sites of aluminate phase and some silicate phases could have been occupied by asphalt emulsion through anionic emulsifier via electrostatic interaction. (ii) the surface of the cement particles might have been covered by a layer of asphalt membrane due to the demulsification of asphalt emulsion, which could have hindered the hydration of cement. Acknowledgements The authors are greatly appreciated for the supports from the National Natural Science Foundation of China (NO.51278231) and Jiangsu Province Natural Science Foundation of China (NO.BK2012867). And also thanks to the Advanced & Innovative Materials (AIM) Group, University College London in UK and the State Key Laboratory of High Performance Civil Engineering Materials in China for supporting the project.

Fig5. Adsorption behavior of EA-601 onto cement particles

In addition, asphalt emulsion belongs to a kind of thermodynamically unstable system, which only maintains its dispersion through electrostatic repulsion between the double electrode layers. 2+ However, once the counter ions, such as Ca , are dissolved from cement particles into the double electrode layers of asphalt emulsion or with the consumption of water during cement hydration process, the asphalt emulsion would demulsify, resulting in the formation of aggregation of asphalt particles and the formation of asphalt membrane (Wang, Liu and Hu, 2013). And thus, the ions releasing and deposition reactions were hindered due to the surfaces of cement particles being covered by asphalt membrane (as shown in Figure 6).

References Jian Ouyang, Yiqiu Tan, Yunliang Li and Jiuye Zhao., 2015. Demulsification process of asphalt emulsion in fresh cement-asphalt emulsion paste. Materials and Structures, 2015(48): 3875-3883. Esveld C., 2003. Recent developments in slab track. Eur Railway Rev 2: 81-85. J.Bensted, P.Barnes., 2008. Structure and Performance of Cements. The 2nd Edition, Published by Taylor & Francis. Wei Li, Xiaobin Zhu, Jinxiang Hong., 2015. Effect of anionic emulsifier on cement hydration and its interaction mechanism. Construction and Building Materials, 93:1003-1011. Lianzhen Xiao., 2007. Interpretation of hydration process of concrete based on electrical resistivity measurement. Doctoral thesis for The Hong Kong University of Science and Technology. G.E.Archie., 1942. The electrical resistivity log as an aid in determining some reservoir characteristics. Trans Am Inst Min Metall Pet Eng, 146:54-62. Johann Plank, Christian Hirsch., 2007. Impact of zeta potential of early hydration phases on superplasticizer adsorption. Cement Concrete Research, 37: 537-542. H.Viallis-Terrisse, A.Nonat, J. -C. Petit., 2001. Zetapotential study of calcium silicate hydrates interacting with alkaline cations. J. Colloid Interface Sci 244: 58-65. Wang Fazhou, Liu Yunpeng, Hu Shuguang., 2013. Effect of early cement hydration on the chemical stability of asphalt emulsion. Construction and Building Materials, 42:146-151.

Fig.6 The interaction modeling between cement and asphalt emulsion

4. CONCLUSIONS This study clearly demonstrated that the cement hydration process was retarded by an anionic asphalt emulsion, and the main conclusions of this investigation can be drawn as follows: (1)The setting time of PC was prolonged with the increase of A/C. (2)The heat released from the cement hydration was delayed with the increase of A/C, which was also corroborated by the setting time results. (3) During the early hydration period, the electrical resistivity is mainly controlled by the ion concentration in the fresh CA paste, and the electrical resistivity of fresh CA paste was higher than that of the pure PC paste; on the other hand, at the accelerated hydration period, the electrical resistivity is mainly affected by the microstructure, such as the porosity and connectivity of the pores, 4

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 14

Timber material model for earthquake-resistant structures L.F. Sirumbal-Zapata, C. Málaga-Chuquitaype and A.Y. Elghazouli Department of Civil and Environmental Engineering, Imperial College London

ABSTRACT A damage constitutive model for timber is proposed in this article that builds upon previous models developed for other materials like concrete. Due to the anisotropic characteristics of timber, Hill’s expression was used as equivalent stress damage criterion for both, compressive and shear-tensile failure. The damage evolution process starts once the failure criterion is reached, and is handled by two different monotonically increasing functions of the equivalent stresses for tension (exponential softening) and compression (perfect plasticity). The damage variables associated to this functions gradually reduce the timber stiffness material parameters (elastic modulus and shear modulus). The main advantage of incorporating continuous damage model for timber is to capture the post-elastic stiffness degradation and thus enable an accurate prediction of experimental results. The proposed constitutive material model represents the first step towards a faithful estimation of the earthquake-induced collapse capacity of timber buildings. timber material under compression, tension and shear stresses is discussed. Then, the theoretical basis and constitutive equations of the elasticity and damage parts of the model are developed. The plasticity part and the coupled plasticitydamage model algorithm are not included in this article.

1. Introduction The structural performance of timber structures subjected to earthquake loading is governed by the non-linear response in the connection zones, where high deformation levels and stress concentrations are developed around the fasteners like nails, dowels or bolts. For this reason, the performance-based assessment of earthquakeresistant timber structures requires the implementation of a material constitutive model in a FE software which allows a faithful simulation of the nonlinear behaviour of these type of connections under large deformation levels until failure.

2. Non-linear behaviour of timber material Timber is an anisotropic material with different types of failure modes for shear, tension and compression stresses. Besides, timber failure modes depend on the loading direction relative to its grain alignment. This means that the mechanical properties (elastic modulus, shear modulus and Poisson’s ratio) and the strength of timber material vary with the directions and sign (similarly to concrete, compressive and tensile strengths are different) of the principal stresses. Based on many experimental tests carried out in the past, the three most important and frequent timber failure modes can be identified and classified as (Xu et al., 2014):

Timber failure modes can broadly be defined as ductile failure due to compression stresses and brittle failure due to the combination of shear and tension stresses. Based on recent developments available in the relevant literature, this research work is focused on the formulation and implementation of a simplified, yet consistent and comprehensive, material constitutive model for timber which can deal with both types of failure modes. This constitutive model is based on continuum damage theory and, although not suitable for explicit crack modelling, it allows to monitor the evolution of damage and the identification of potential rupture zones through a smeared continuous approach. To this end, a 3D coupled plasticity-damage constitutive model is being implemented in the FE software DIANA. The model will reproduce the non-linear behaviour of dowelled timber-steel connections subject to cyclic loading.

• • •

Ductile failure due to compression parallel to the grain. Ductile failure due to compression perpendicular to the grain. Brittle failure due to shear parallel to the grain and tension perpendicular to the grain.

Fig. 1 shows the compression stress-strain curves of Scandinavian spruce with a mean density of 430 3 kg/m experimentally obtained by Karagiannis et al. (2016). In the direction parallel to the grain (Fig. 1a) the behaviour is approximately linear elastic until the compressive strength is reached, around 40 MPa. After this point, a minor stress drop is produced first followed by perfect plasticity,

In the next sections the experimental non-linear behaviour and failure modes characteristic of 1

generating permanent deformations. On the other hand, the compressive stress-strain relationship in the direction perpendicular to the grain (Fig. 1b) shows plastic behaviour with slight hardening after a strength of around 2.5 MPa is reached. It is important to notice that the compressive strength in the direction perpendicular to the grain is less than 10% of the strength in the direction parallel to the grain.

Figure 2. Scandinavian spruce tension stress-strain curve. a) Parallel to the grain. b) Perpendicular to the grain. (Karagiannis, 2016).

3. Orthotropic linear-elastic behaviour of timber Before yielding (compression) or failure (tension) is produced, the strain-stress relationship is expressed as: Figure 1. Scandinavian spruce compression stress-strain curve. a) Parallel to the grain. b) Perpendicular to the grain (Karagiannis et al., 2016)

𝜺 = 𝑪𝒆 : 𝝈

(1)

where 𝑪𝒆 is the fourth-order orthotropic linearelastic compliance tensor, defined as a 6 by 6 matrix in Voigt’s notation such that:

The tension stress-strain curves of the same Scandinavian spruce specimen are shown in Fig 2. An initial linear elastic followed by a brittle failure is observed in both directions, parallel (Fig. 2a) and perpendicular (Fig. 2b) to the grain. The postelastic behaviour in tension is completely different than in compression due to absence of plastic deformation and the total loss of strength capacity. After the maximum tensile strength is attained, a sudden failure occurs and the capacity of the material to sustain additional loading vanishes. Similarly than for compression, the tensile strength parallel to the grain (around 30 MPa) is considerably higher than the tensile strength perpendicular to the grain, around 0.5 MPa.

1

⎡ 𝐸𝑋 ⎢− 𝜈𝑋𝑋 ⎢ 𝐸𝑋 ⎢− 𝜈𝑋𝑋 𝐸𝑋 𝒆 𝑪 =⎢ ⎢ 0 ⎢ ⎢ 0 ⎢ ⎣ 0

−

𝜈𝑌𝑌 𝐸𝑌 1

𝐸𝑌 𝜈𝑌𝑌

−

𝐸𝑌

0

0

0

−

−

𝜈𝑍𝑍 𝐸𝑍 𝜈𝑍𝑍

𝐸𝑍 1

𝐸𝑍

0

0

0

0

0 1

0

𝐺𝑋𝑋

0

0

0

0

0 0 1

𝐺𝑌𝑌

0

0 ⎤ 0 ⎥ ⎥ 0 ⎥ ⎥ 0 ⎥ ⎥ 0 ⎥ 1 ⎥ 𝐺𝑍𝑋 ⎦

(2)

The orthotropic directions X, Y and Z correspond to the longitudinal, transverse and radial timber local axes, respectively. It is assumed that the longitudinal axis direction is parallel to the grain of the timber material, while the transverse and radial axes lay in the cross-section plan and act in the direction perpendicular to the grain. Notice that due to timber anisotropy the values of the elastic modulus, the shear modulus and the Poisson’s

In light of the above, timber failure modes can be summarized as elasto-plastic ductile failure with large deformation for compressive stresses, and elastic brittle failure for the interaction of shear and tensile stresses. Based on this, a physically consistent constitutive model is proposed to study the non-linear response of 3D timber structures subject to cyclic earthquake loading. 2

order tensor 𝑴, which is a function of the fourth order damage tensor 𝑫:

ratio are different for each of the three orthogonal axes or planes. However, due to the symmetry property of 𝑪𝒆 , the conditions expressed in Eq. (3) hold for the Poisson’s ratios and the elastic modulus. 𝜈𝑋𝑋 𝐸𝑋

=

𝜈𝑌𝑌 𝐸𝑌

;

𝜈𝑌𝑌 𝐸𝑌

=

𝜈𝑍𝑍 𝐸𝑍

;

𝜈𝑋𝑋 𝐸𝑋

=

𝜈𝑍𝑍 𝐸𝑍

𝝈 = 𝑴(𝑫): 𝝈 �

Anisotropic damage is considered by assigning different values to the damage variables components of 𝑫. However, in spite of being an anisotropic material, there are two main disadvantages of including anisotropic damage in the timber material model (one physical and the other numerical). First, the damage variables evolution laws for the stresses terms in each direction are not known and difficult to obtain through experimental tests. Second, the strain equivalence hypothesis is not valid for anisotropic damage, and therefore, it is not possible to obtain a mechanically consistent anisotropic damage tensor without losing the symmetry of the Cauchy or the effective stress tensors. Given that, by definition, the Cauchy stress tensor is symmetric, the only possible option would be to work with a non-symmetric effective stress tensor (Valliappan et al., 1999). The reasons for keeping the effective stress tensor symmetric will be explained later.

(3)

As a consequence, in order to define the linearelastic orthotropic behaviour of timber a total of nine mechanical material parameters are required: three elastic modulus, three shear modulus and three Poisson’s ratios. Nevertheless, transverse isotropy is generally assumed for timber, reducing the number of parameters to six, by means of the following additional set of relations: 𝐸𝑌 = 𝐸𝑍 ;

𝐺𝑋𝑋 = 𝐺𝑍𝑍 ;

𝜈𝑋𝑋 = 𝜈𝑋𝑋

(4)

In summary, Eqs. (1) to (4) are used to model the linear elastic behaviour of timber before ductile compression yielding or brittle tension failure are produced. 4. Strain-based isotropic damage model for timber

Therefore, isotropic damage is selected as the best option for the development of a mechanically consistent damage model. One of the most common and successful techniques for modelling isotropic damage consists in replacing both tensors 𝑴 and 𝑫 by scalar terms. Thus, Eq. (5) for the particular case of isotropic damage with a scalar damage variable 𝜔 is expressed as:

Brittle failure due to tension and shear stresses generates voids and micro cracks in the timber matrix which not only lead to a sudden reduction of the material strength, but also to a gradual degradation of its mechanical properties, and thus stiffness. When further loading is applied, the micro-cracks grow and their coalescence produce macro-cracks zones and irreversible damage (Khelifa et al., 2015).

𝝈 = (1 − 𝜔)𝝈 �

(6)

𝝈 � + = ∑3𝑖=1〈𝜎�𝑖 〉𝒑𝒊 ⨂𝒑𝒊

(7)

The damage process starts with the fulfilment of the damage criteria condition. Once this damage criteria is met, the value of the damage variable increases gradually and monotonically from 0 (undamaged state) to 1 (total damage state). A timber material constitutive model for cyclic loading should be capable to reproduce the different inelastic behaviour observed experimentally for tension and compression. This does not only include the different strengths and failure modes, but also the different post-elastic strength and stiffness degradation. The latter is particularly important to capture correctly the unloading stiffness degradation and the stiffness recovery after load reversal, when passing form tension to compression stress states and viceversa (Faria et al., 2004). For this reason, a split of the effective stress tensor into a tensile (𝝈 �+ ) and a compressive − (𝝈 � ) components is performed such that:

Continuum damage mechanics (CDM), based on the thermodynamics of irreversible processes theory, has been widely used for modelling the non-linear behaviour of different brittle materials, like concrete, rock (Simo and Ju, 1987), and more recently, timber (Sandhass et al., 2012). Strainbased damage models rely on the concept of effective stress and the hypothesis of strain equivalence. The former is defined as the stress acting in the reduced undamaged net surface area of the material, without considering the portion of area taken by the micro-cracks and voids. Considering that the total force acting in the material body is constant, the magnitude of the effective stress acting in the reduced undamaged area is higher than the Cauchy stress acting over the total nominal surface area. On the other hand, the hypothesis of strain equivalence states that the strain associated with the Cauchy stress in the damaged state is equivalent to the strain associated with the effective stress in the undamaged state (Simo and Ju, 1987). The effective stress tensor 𝝈 � is transformed into the Cauchy stress tensor 𝝈 by means of the fourth-

(5)

𝝈 �− = 𝝈 �−𝝈 �+ 3

(8)

Therefore, the management of independent damage mechanisms for tension and compression (Faria et al., 1998) is made possible. A spectral decomposition of the effective stress tensor has to be performed in order to obtain its principal values (𝜎�𝑖 ) and their associated unit eigenvectors (𝒑𝒊 ). This spectral decomposition is one of the reasons why it is important to work with a symmetric effective stress tensor, and therefore, with an isotropic damage model.

evolution law and loading-unloading (Kuhn-Tucker) conditions defined in Eqs. (9), (11) and (12), respectively. Thus, it can be concluded that, in addition to define the start of the damage evolution process, the damage function accomplishes two more tasks. The first of them is the identification of the loading or unloading states, while the second is the activation and deactivation of the damage variable growth (Faria et al., 2004).

On the other hand, the initiation of the damage evolution process (when the damage variable 𝜔 ± starts its gradual growth from 0 to 1) is determined by the damage criterion function defined in Eq. (9). This is expressed in terms of an effective stressbased norm, called “equivalent stress” (𝜏̅ ± ), and a threshold variable (𝑟 ± ) which controls the size of the damage surface. 𝑓𝑑 ± (𝝈 � ± , 𝑟 ± ) = 𝜏̅ ± − 𝑟 ±

(9) Figure 3. Cauchy stress evolution law for tensile softening employed in Calayir and Karaton (2005), Faria et al. (1998), Sandhaas et al. (2012) and De Borst et al. (1999).

Due to the anisotropy, complex stresses interaction and failure modes characteristic of timber material, a multidimensional damage function based on Hill’s criterion is adopted in this model. Hence, the associated equivalent stress of the Hill’s damage criterion is defined as: 3

𝜏̅ ± = 𝝈 � 2

±𝑻

𝑷± 𝝈 �±

𝑓𝑑 ± ≤ 0; 𝑟̇ ± ≥ 0; 𝑟̇ ± 𝑓𝑑 ± = 0

References

(10)

Calayir, Y., Karaton, M., 2005. A continuum damage concrete model for earthquake analysis of concrete gravity dam-reservoir systems. Soil Dynamics and Earthquake Engineering, 25(11), 857–869. De Borst, R., Pamin, J., Geers, M. G. D, 1999. On coupled gradient-dependent plasticity and damage theories with a view to localization analysis. European Journal of Mechanics, A/Solids, 18(6), 939–962. Faria, R., Oliver, J., Cervera, M., 1998. A strain-based plastic viscous-damage model for massive concrete structures. International Journal of Solids and Structures, 35(14), 1533–1558. Faria, R., Oliver, J., Cervera, M, 2004. Modeling Material Failure in Concrete Structures under Cyclic Actions. Journal of Structural Engineering, 130(12), 1997–2005. Karagiannis, V., Málaga-Chuquitaype, C., Elghazouli, A. Y., 2016. Modified foundation modelling of dowel embedment in glulam connections. Construction and Building Materials, 102(2), 1168–1179. Karagiannis, V., 2016. Structural response of hybrid steel-timber joints. PhD Thesis (in preparation). Khelifa, M., Khennane, A., El Ganaoui, M., Celzard, A., 2015. Numerical damage prediction in dowel connections of wooden structures. Materials and Structures. Sandhaas, C., Van de Kuilen, J.-W., Blass, H. J., 2012. Constitutive Model for Wood Based on Continuum Damage Mechanics. In World Conference on Timber Engineering, Auckland. Simo, J. C., Ju, J. W., 1987. Strain- and stress-based continuum damage models—I. Formulation. International Journal of Solids and Structures, 23(7), 821–840. Valliappan, S., Yazdchi, M., Khalili, N., 1999. Seismic analysis of arch dams—a continuum damage mechanics approach. International Journal for

where the equivalent stress, 𝜏̅ ± , is defined for both, tension and compression failure, in terms of the corresponding effective stress tensor 𝝈 � ± and ± the strength parameters matrix 𝑷 . The variation of the damage variable 𝜔 ± in function of the threshold variable 𝑟 ± is ruled by the damage evolution law, as defined in Eq. (11). The mathematical expression of this monotonically increasing function is related to the post-elastic stress behaviour (e.g. softening, perfect plasticity, hardening) by means of Eq. (6). 𝜔 ± = 𝑔𝑑 ± (𝑟 ± )

(12)

(11)

Four damage evolution laws employed elsewhere to model tensile softening behaviour are graphically reproduced in Fig. 3 for the Cauchy stress 𝜎. The versatility of one of them (De Borst et al., 1999) for modelling different exponential softening behaviours due to the inclusion of the parameter 𝑛 can be appreciated. For 𝑛 smaller than 1, asymptotically exponential softening is noticed. On the contrary, for values of 𝑛 higher than 1, there is no asymptotic behaviour and the stress rapidly decays to zero. For this reason, and based on the behaviour observed in Fig. 2, this damage evolution law was selected to model the timber damage evolution for tensile softening. In summary, the damage model is completely defined by the damage function, the damage 4

Numerical Methods in Engineering, 45(11), 1695– 1724. Xu, B., Bouchaïr, A., Racher, P, 2014. Appropriate Wood Constitutive Law for Simulation of Nonlinear Behavior of Timber Joints. J. Mater. Civ. Eng. (ASCE), 26, 1–7.

5

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 15

Behaviour of Glued-in BFRP Rods under combined Axial Force and Bending Moment C. O’Neill, D. McPolin and S.E. Taylor Civil Engineering Research Centre, Queen’s University Belfast A.M. Harte College of Engineering and Informatics, National University of Ireland Galway ABSTRACT Glued-in rods (GiR) present a viable alternative to traditional steel moment connections in both new build and retrofit of timber structures. A simplified experimental set-up was developed to study the behaviour of the glued-in rod system under a combination of axial force and bending moment. The testing method allowed controlled adjustment of embedded length and edge distance to assess their influence on performance of the system. A clear increase in pull-out strength was observed with an increase in embedded length. An increase in pullout capacity of 213% was observed between the shortest embedded length of 80mm and the longest length of 600mm. Increasing edge distance did not significantly affect failure strength however a difference in behaviour was observed with a shift to a more favourable failure mode with increased edge distance. For axially loaded systems the stress distribution along a GiR is generally considered to show a peak at the loaded end with dissipation of the stress occurring along the length of the rod. At a fixed loading rate it appears that for short embedded lengths the entire length of the rod reaches peak stress at once whereas at longer embedded lengths failure is more gradual, with one end reaching peak stress before the other leading to a higher failure load and lower peak stress. This research aims to determine if this assumption of stress distribution is valid also for systems under a combination of axial and bending forces. Electrical Resisting Strain gauges and Draw Tower Grating fibre optic sensors were used to capture the stress profile along the length of the GiR. Embedded length and edge distance were varied to investigate the effect of these variables on stress distribution. Specimens were tested under a pull-bending test set-up. Generally, a linear increase in strain was observed at each measured location along the glued length until failure. In all specimens, the loaded end recorded the maximum strain. This proves that failure initiates primarily at the loaded end. This paper presents the full results of this experimental program.

and floor beams in buildings subject to decay (Smedley et al. 2006; Schober & Rautenstrauch 2005). In new build, five areas were identified where glued-in rods may be used for connections: frame corner, beam-post connection, beam-beam joint, supports and hinged joints (Gehri 2010).

1. INTRODUCTION The development of lightweight, corrosion resistant and sustainable moment resistant timber connections using glued-in rods would facilitate the adoption of timber elements in large construction projects. However the basic principles of behaviour under both axial force and bending moment must first be established. Glued-in rods (GiR) present a sustainable, aesthetically pleasing alternative to the cumbersome conventional steel moment connections that are often encountered in timber construction. Not only do connections with bondedin rods look better than conventional connections, they also have enhanced fire protection as the rods which transfer moment are embedded inside, and are therefore protected by, the timber.

Since the late 1980s there have been many research projects commissioned on the use of bonded-in rods in timber construction e.g. GIROD and LICONS (Broughton & Hutchinson 2004; Bainbridge & Mettem 1999). In spite of this, no universal standard exists for their design. There had been an informative annex in the pre-standard PrBS ENV 1995-2:1997 which provided limited coverage of the design of bonded-in rods using steel bars however this document was replaced by BS EN 1995-2:2004 and no guidance is included in this current document.

Glued-in rods have great potential in a wide range of both new build and restoration projects. Successful renovation has been carried out in roof

Considerable research can be found in the literature on the behaviour of such connections however, the majority of this research is focused on 1

steel rods under purely axial loading. In service, these moment resisting connections will be subject to a combination of axial and bending forces rather than exclusively axial force. Performance of glued-in Basalt Fibre Reinforced Polymer (BFRP) rods under combined axial and bending was appraised by considering both the joint performance in terms of strength, failure mode and deflection and the nature of the stress distribution along the joint interface. This was achieved by altering both embedded length and edge distance in a controlled manner and monitoring the effect this had on performance. Joint performance was determined by measuring the force, deflections and strain as well as observing behaviour during loading and failure mode.

Test set-up Pull-out capacity can be used as a measure of the strength of a glued-in rod. The pull-out test system used was a pull-bending set-up, as pictured in Figure 1. The pull-bending system allows bending strength of the glued-in rod connection to be evaluated by removing the timber in the section being loaded so that the only resistance is from the BFRP bars glued-in to the timber.

Figure 1. Pull-bending test set-up

The stress distribution along a GiR is generally considered to show a peak at the loaded end with dissipation of the stress occurring along the length of the rod as suggested by Steiger et al. (2006) for axially loaded systems. At a fixed loading rate it appears that for short embedded lengths the entire length of the rod reaches peak stress at once and hence fails at a relatively low load and with high peak stress. At longer embedded lengths failure is more gradual, with one end reaching peak stress before the other leading to a higher failure load and lower peak stress. This research aims to determine if this assumption of stress distribution is valid for pull-bending mechanisms; that stresses are not distributed evenly along the embedded length of a GiR and that failure arises at the loaded end due to a peak in stress concentration at this location.

Stress in the rod was monitored by means of an electrical resistance strain (ERS) gauge placed at mid-span on the exposed BFRP rod on each sample. Vertical deflection at mid-span and slip of the rod were recorded with linear variable differential transformers (LVDTs). A flat plate was clamped at mid-span to allow measurement of overall slip of the rod. Embedded length (l b ) was varied to assess its impact upon the performance of the glued-in rod. Embedded length is thought to be the most influential variable on pull-out capacity of a glued-in rod. Embedded length ranging from 80mm to 600mm was investigated. Two methods were employed to obtain the stress profile along the length of the GiR: Electrical Resisting Strain (ERS) gauges and Draw Tower Grating (DTG) fibre optic sensors. The sensors were attached directly on to the BFRP rod on one specimen in each tested set to assess the how stresses were distributed along the glued length after being transferred through the timber and adhesive.

2. Test Procedure Materials Class C16 Irish Sitka Spruce (Picea sitchensis), sourced from Balcas Sawmill in Northern Ireland with a size of 75mm x 225mm sawn section was used. Material testing revealed that this timber had th 2 a 5 percentile bending strength f m,k = 16.8N/mm , 2 shear strength f v,k =8.7N/mm and a density ρ k = 3 381kg/m .

Samples were loaded in 0.5kN increments to failure using the accurately calibrated 600kN capacity hydraulic actuator. Deflection at mid-span and net horizontal movement of the bar as the sample was loaded was recorded with data acquisition connected to the transducer. Failure load was recorded when the sample could not take any additional load. The mode of failure was recorded also – percentage failure mode was then calculated for each bonded length. Each test was repeated nine times due to the high variability of the timber used.

12mm diameter Basalt Fibre Reinforced Polymer (BFRP) rods were used in this experimental programme. These rods were found to have a 2 tensile strength of 920 N/mm under a low loading rate of 0.2kN/s (Tharmarajah 2010). Unlike steel or some other FRPs, no extensive cleaning of the rods was required prior to bonding as they are sand-coated which provides a good surface for adhesion.

3. Results

A two-part thixotropic gap filling epoxy was used. This adhesive only flows under shear so is ideal for GiR applications such as overhead beam repair or jointing overhead.

Failure mode All specimens failed in a sudden, brittle manner. Two primary failure modes were identified and are pictured in Figure 2: a timber plug pull-out 2

indicative of shear failure in the timber and a ‘clean’ pull-out signifying a failure of the rod/adhesive interface. The most prevalent failure mode observed was a pull-out failure in shear of the timber with a total of 67.6% of all samples failing in this manner. This was as expected due to the timber being the weakest element in the connection. Rod/adhesive failure was thought to have occurred due to the sand coating on the BFRP rod not adhering sufficiently well to the adhesive. The BFRP rod never failed as the force required for the rupture of the rod was never reached.

failure strength. However, as edge distance increased the lever arm was reduced. Thus, moment capacity of the section decreased with increasing edge distance since moment capacity is a function of the force and lever arm. Influence of embedded length A clear increase in pull-out strength was observed with an increase in embedded length as illustrated in Figure 4. An increase in pull-out capacity of 213% was observed between the shortest embedded length of 80mm and the longest length of 600mm. This was as expected since the larger interface area with each increase in embedded length provides additional resistance to the applied loading.

Figure 2. Failure modes observed a) shear in timber, b) rod/adhesive failure Figure 4. Pull-out strength with increasing embedded length

Splitting was evident in 28% of all specimens. Splitting occurred as a consequence of the build-up of stresses approaching failure. When splitting of the timber occurred the length of the split was often equal to the embedded length of the rod. When the timber split the capacity of the section was reduced resulting in a lower failure load, this is evidenced in Figure 3 where strength of specimens without splitting is compared to those where splitting did occur.

Comparing the experimental data obtained in this research to the three most used design guidelines, it can be seen that the data follows the same trend as both the DIN and Riberholt design equations (Figure 5). The experimental strengths are significantly stronger than both the DIN and Riberholt predictions however this is to be expected since the guidelines are designed to give a safe prediction of strength. The GIROD prediction gives a completely linear behaviour. While this is conservative at shorter embedded lengths compared to the experimentally derived data, beyond an embedded length of 330mm the design prediction is significantly higher than the experimentally obtained strengths and therefore unsafe.

Figure 3. Comparison of failure strength of specimens with and without splitting

In an attempt to alleviate this issue and optimise the specimen capacity, edge distance was increased. This principal is similar to increasing the cover to reinforcement in a concrete beam. Splitting was significantly reduced with increasing edge distance with no significant impact upon

Figure 5. Comparison of experimental data with commonly used theoretical predictions

3

4. Conclusions

Stress distribution along glued length Generally, a linear increase in strain was observed at each measured location along the glued length until failure. As shown in Figure 6 and Figure 7, in all specimens, the loaded end (0mm) recorded the maximum strain. This proves the presumption that failure occurs primarily at the loaded end.

• A clear increase in pull-out capacity with increased embedded length of the glued-in BFRP rods was seen, which appears to be reaching a plateau as failure is occurring in the timber as opposed to in the bond. • Increasing edge distance is an effective method of reducing instances of splitting without sacrificing pull-out capacity. • Stress distribution along the length of a glued-in rod under combined axial and bending force is not linear, with the loaded end reaching failure first and this then propagating along the glued length of the rod. • Further work is suggested to assess the performance of such a connection method as a moment connection in a portal frame application. Acknowledgements This research was funded by the Department of Agriculture, Food and the Marine of the Republic of Ireland under the FIRM/RSF/COFORD scheme as part of ‘Innovation in Irish timber Usage’, project ref. 11/C/207. The authors would also like to thank the technical staff in QUB for their assistance.

Figure 6. Stress distribution along glued length with increasing edge distance at a load of 10kN

References Bainbridge, R. & Mettem, C., 1999. Bonded-in rods for timber structures: a versatile method for achieving structural connections. The Structural Engineer, Volume 77(Issue 5), pp.p24–27. Broughton, J.G. & Hutchinson, A.R., 2004. LICONS Task 2-Sub tast 2.2, Gehri, E., 2010. High Performing Jointing Technique Using Glued-in Rods. In 11th World Conference on Timber Engineering 2010, WCTE 2010. Trentino, Italy. Schober, K.U. & Rautenstrauch, K., 2005. Experimental Investigations on Flexural Strengthening of Timber Structures with CFRP. In Proceedings of the International Symposium on Bond Behaviour of FRP in Structures (BBFS 2005). pp. 457–464. Smedley, D., Alam, P. & Ansell, M., 2006. George Street, St. Albans, UK–a case study in the repair of historic timber structures using bonded-in pultruded plates. … of 9th World Conference on Timber …, (2006). Steiger, R., Gehri, E. & Widmann, R., 2006. Pullout strength of axially loaded steel rods bonded in glulam parallel to the grain. Materials and Structures, Vol. 40(8), p.p 69–78. Tharmarajah, G., 2010. Compressive Membrane Action in Fibre Reinforced Polymer (FRP) Reinforced Concrete Slabs. Queen’s University Belfast.

Figure 7. Stress distribution along the glued length with increasing embedded length at a load of 10kN

In a typical specimen stresses were distributed in a triangular fashion with the start of the bond length having the highest stress concentration and this dissipating along the bonded length to a minimum concentration at the unloaded end. Further along the glued length it was observed that behaviour became more linear. This suggests that bending has less of influence on performance of the GiR further along the glued length. As loading increased stresses increased at each location along the rod Increasing edge distance tended to result in a reduction of stresses at each measured point. Increasing embedded length had the opposite effect, with higher stresses being recorded at each point along the rod, suggesting that forces are more effectively being transferred through the timber to the connection element when a longer embedded length is used.

4

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 16

Recycling of fly ash-slag Geopolymer binder in mortar mixes N.A. Chaliasou, A. Heath and K. Paine Department of Architecture and Civil Engineering, University of Bath

ABSTRACT Fly ash-slag based Geopolymer cement (GPC) has demonstrated mechanical properties and environmental advantages that make it one of predominant sustainable alternatives to Portland cement (PC). Despite the fact that numerous environmental analyses about geopolymers are being published, their environmental impact after the end of service-life has barely been explored. Given that construction-waste management is a major sustainability issue, the present study is investigating the potential of recycling fly ash-slag GPC as a fine aggregate in mortar mixes. The major physical properties of the fine recycled aggregates (FRA) were tested and compared to those of PC FRA and natural sand of similar fineness. The effect of incorporating FRA in low (25%) and high (50%) percentage in PC or GPC matrix mortars was investigated. The 28day compressive and flexural strength of mortars were tested. Also the 28day water absorption and flow of mixes incorporating GPC FRA were recorded. GPC FRA exhibited properties similar to those of PC FRA and poorer than those of natural sand. The results of compressive and flexural strength proved that FRA addition had a negligible effect in all cases. The influence of the high water absorption of GPC FRA, relatively to that of natural sand, was prominent on the workability of fresh mixes and possibly affected the water absorption of mortar prisms. The effect of GPC FRA proved to be similar to that of PC FRA on compressive strength, while none of the tested mortar properties appeared to be jeopardised by the incorporation of the GPC FRA in the mix. demolition wastes is an issue that already troubles the construction industry as far as Portland cement (PC) concrete is concerned. Therefore, it has been deemed appropriate to investigate disposal or reuse scenarios for GPC since the potential of its wider adoption is becoming higher. The present study is investigating the potential of recycling fly ash-slag based GPC by utilising it as recycled aggregate in mortar mixes. To do so, fine recycled aggregates (FRA) were produced in the laboratory and their basic physical properties such as water absorption (WA) and particle density were investigated. The potential of recycling GPC in mortar mixes with PC and fly ashslag based GPC matrices was investigated. The effect of GPC FRA on the 28day flexural and compressive strength of mortars was tested and the results were compared to those obtained by similarly produced mortars with natural sand and PC FRA. Additionally, the flow and water absorption of mortars incorporating GPC FRA were recorded.

1. INTRODUCTION Geopolymers were popularised by Joseph Davidovits in the 1970’s and their production relies on minimally processed natural materials or industrial by-products. There are various categories of geopolymer binders (depending on the raw materials), but from a terminological point of view, geopolymer cement is a binding system that hardens at ambient temperature. Given that this relies on addition of calcium (usually through ground granulated blast furnace slag - GGBS), the most appropriate type for high volume construction applications is fly ash-slag based geopolymer (Davidovits, 2015). With the adoption of that type of geopolymer cement (GPC), extreme high temperature kilns with large expenditure of fuel and CO 2 emissions by the decomposition of calcium carbonate are avoided. Therefore a reduction of carbon emissions at the range of 40-80% is considered possible ( Davidovits, 2013). But before proceeding to its wider adoption, the engineering community has to ensure tangible environmental benefits. An increasing number of life cycle assessments and environmental evaluations of GPCs are being conducted. Most of them are focusing on cradle-togate issues relating to the availability of materials or the environmental impact of the activating solutions (Habert, et al., 2011) (Fawer, 1999) (Heath, et al., 2014). Hardly any research concerning the stage after the end of service-life has been conducted. Disposal of construction and

2. Materials and Experimental methodology The GGBS used for the GPC precursor was provided by the Hanson Heidelberg cement group from the Port Talbot works. The fly ash (FA) was CEMEX 450-S (BS EN 450 - 1 Fineness Category S; LOI Category B). The activating solution was produced by mixing sodium hydroxide pellets (NaOH, 98-100.5%) and sodium silicate solution (Na 2 O(SiO 2 )x · xH 2 O, Na 2 O, ~10.6%,SiO 2 , ~26.5%) with distilled water. The chemicals were

1

physical properties of aggregates”. The final mortars were tested according to BS EN 1961:2005 “Methods of testing cement” and BS 1881122:2011 “Testing concrete”. The notation used for the mixes follows the rationale Matrix typeReplacement percentage.

both supplied by Sigma-Aldrich and the mixing of the solution took place 24 hours prior to casting. For the PC elements, General Purpose Portland fly ash cement/ Sulfacrete EN 197-1 –CEMII/BV 32,5R supplied by Tarmac Cement & Lime Ltd was used. The natural aggregate incorporated in the mortars was standard sand complying with BS EN 196-1 specifications. The same materials were used for the fabrication of both the recycled aggregates and the mortars. Although the PC used had fly ash addition, the specimens will be referred to as PC for convenience. The FRA were produced in the laboratory by casting 100mm cubes of pure paste to establish a worst case. For the GPC binder the proportion of FA/GGBS was 1/1, water/GPC solids 0.4, the Na 2 O/SiO 2 ratio 1 and the percentage of Na 2 O in the activating solution was 5% by mass precursor. The water/GPC solids ratio was used and it indicates that the products deriving from the dissolution of the compounds of the activating solution were taken into consideration for the calculation of the free water proportion for the mix. For the PC paste the water/cement ratio was 0.4. The cubes were demoulded one day after casting and then stored in sealed plastic bags at 20±1°C for a period of 28 days before crushing. The resulting FRA was crushed and sieved to achieve fineness and particle distribution similar to that of standard natural sand.

3. Results and discussion The results of the FRA testing are presented in Table 2. The apparent particle density of GPC FRA is about 20% lower than that of natural sand but still at the same order of magnitude. For oven-dried and saturated surface dried condition the difference increases at about 48% and 33% accordingly. When compared to the results obtained from the testing of similarly produced PC FRA, it appears that GPC FRA are less influenced by oven drying and water immersion. Table 2 Particle density and 24hour water absorption of natural sand, GPC and PC FRA Natural GPC Aggregate type PC FRA Sand FRA 3 Particle density (kg/m ): Apparent 2631 2041 2247 Oven-dried 2202 1151 895 Saturated-Surface dried 2367 1588 1497

The obtained results do not present significant divergence from those reported in literature regarding PC recycled aggregates of similar 3 fineness. The usual ranges are 1,970-2,140 kg/m 3 for oven dried and 2,190-2,320kg/m for saturated surface dried density ( Dhir, et al., 1999) (Wai , et al., 2012) (Hansen, 1986) ( Silva, et al., 2014b). Given the nature of the original material (binder without any aggregate) such values are considered reasonable. The most striking observation is that GPC FRA exhibited water absorption (WA) 5 times that of natural sand. The general trends reported in literature are much lower. Specifically, PC recycled aggregates are reported to demonstrate WA 3–6 times higher than that of natural aggregates, while FRA present values at the range of 8-12% (Akash Rao, 2007) (Hansen, 1986). Given that the tested PC FRA exhibited an exceptionally high WA as well, it can be assumed that this significant increase is due to the nature of the original material and the fineness to which the FRA were crushed. When evaluating the obtained results comparatively though, it can be stated that GPC FRA had a better performance than PC FRA. The remarkably high WA of GPC FRA was correlated with the results of the WA of the mortar prisms and the flow of fresh mixes. As it can be observed by the values demonstrated in Table 3, the WA of GPC FRA had a prominent effect on the workability of mortars. Specifically, a gradual reduction of flow was observed with increasing replacement levels for both matrix types. With incorporation of 25% GPC FRA in the mortar, a 13% reduction of flow occurred but the mixes maintained reasonable

Table 1 Amount of materials used for the production of recycled aggregate mortars in kg/m3 Matrix PC PC PC GPC GPC GPC type RA 0% 25% 50% 0% 25% 50% PC 960 960 960 FA 480 480 480 GGBS 480 480 480 NA 2160 1620 1080 2160 1620 1080 RA 540 1080 540 1080 NaOH 19 19 19 Silicate 53 53 53 Water 451 451 451 463 463 463

The replacement percentages of natural sand in the final mortars were 25% and 50% while the reference mixes (0% replacement) incorporated standard sand only. For the GPC mortars water/GPC solids ratio was 0.47 and Na 2 O was 3% by mass of precursor, while all the other parameters were the same as for the original GPC binder. The water/cement ratio for the PC mortars was 0.47. For all mortar mixes the proportions were 2.25 parts of sand to 1 part of binder per mass. No presaturation of aggregates or water compensation took place. Mortar prisms of 40x40x160mm were fabricated and demoulded 24h after casting. Then they were kept at 20±1°C in moist conditions for 28 days. The amount materials used for the casting of the final mortars is presented in Table 1. The recycled aggregates were subjected to water absorption and particle density testing according to BS EN 1097-6:2013 “Test for mechanical and

2

workability. With increased replacement level though, flow demonstrated a reduction of about 50% for both matrix types and the mixes were hard to cast.

increase was observed for 25% GPC FRA, while for 50% the strength is slightly lower than that of the reference mix. For high replacement level, the favourable effect was probably counterbalanced by the dryness of the mix. In the case of GPC mortars, a clear enhancement of strength is observed by the GPC FRA addition. Regardless the FRA level, a strength increase at the range of 50% took place. Apart from the justification given above for the case of PC mortars, an interpretation of the effect of GPC FRA addition could be based on the fact they derived from a binder with alkalinity higher than that of the final mortar. It is possible that alkalis from the FRA leached into the new mix resulting to the observed strength increase.

Table 3 Flow and 28day water absorption of mortar mixes incorporating GPC FRA. The relation to the reference mixes with 0% GPC FRA is stated as reduction % and increase %. Flow Reduction WA Increase PC-0%

144%

-

4.9%

-

PC-25%

125%

13%

6.7%

37%

PC-50%

68%

53%

10.1%

108%

GPC-0%

145%

-

6.8%

-

GPC-25%

126%

13%

9.8%

45%

GPC-50%

77%

47%

10.6%

56%

These results are in accordance with the information reported in literature. Specifically, it is mentioned in most studies that WA of recycled aggregates influences directly the workability of the resulting mixes, as the hardened mortar tends to absorb the available water during mixing, resulting to less free water in the mix (Akash Rao, 2007) ( Zhao, et al., 2015) ( Evangelista & de Brito, 2007) (Pepe, et al., 2014) (Silva, et al., 2014a). Additionally, aggregate porosity and WA have been correlated with the water absorption of the resulting mortar (Zega & di Maio , 2011) ( Evangelista & de Brito, 2010). This remark is confirmed by the results in Table 3. PC mortars appeared to be more prone to the influence of GPC FRA addition, given that, for double replacement percentage, the WA doubled. For the GPC mortars the replacement level did not seem to have a remarkable impact as both cases showed increases of about 50%.

Compressive strength (MPa)

70

35.3 37.4

35.2

30

24.2

35.9

37.0 24.2

20

12.3 11.6

10

12

10 0 OPCMortar

41.9

30

0%-RA 25%-RA 50%-RA

20 Flexural Strength (MPa)

Compressive Strength (MPa)

38.9

40

The results obtained by corresponding mixes incorporating PC FRA are presented in Figure 2. It is observed that in the case of PC mortar the replacement of natural sand had a similar effect, irrespective of the FRA type. For GPC mortars though, it is prominent that PC FRA had the opposite effect of GPC FRA. Summarizing the compressive strength results it could be stated that GPC FRA had a similar or enhancing effect compared to PC FRA.

0%-RA 25%-RA 50%-RA 35.9

50%-RA

50

OPCMatrix GPCMatrix Figure 2 Results of 28day compressive strength testing of PC and GPC mortars incorporating PC FRA

50 40

25%-RA

0

70 60

0%-RA

60

GPCMortar

Figure 1 Results of 28day compressive strength testing of PC and GPC mortars incorporating GPC FRA

In fact, the high WA of FRA in combination with the hardened paste could lead to an overall modification of the nominal w/c ratio and the resulting compressive strength of the mix. This could be related to the obtained results for compressive strength presented in Figure 1. In most cases the addition of recycled aggregates is reported to lead in decrease of mortar and concrete strength (Gerardu & Hendricks, 1985) ( Zhao, et al., 2015) (Hansen, 1986). It was observed that, regardless the replacement level, the addition of GPC FRA had a minor effect on the compressive strength of PC mixes. A slight

10 8

7.83 6.68

6

6.11

6.43 6.56 4.65

4 2 0

OPCMatrix GPCMatrix Figure 3 Results of 28day flexural strength testing of PC and GPC mortars incorporating GPC FRA

A similar trend was recorded for the results of the 28day flexural strength of mortars (Figure 3). 4. Conclusions The overall conclusion of the present research is that no indication that could be considered as

3

preventive for the use of GPCs as recycled aggregates in mortar mixes was observed. GPC

Evangelista, L. & de Brito, J., 2007. Mechanical behaviour of concrete made with fine recycled concrete aggregates. Cement & Concrete Composites 29, p. 397–401. Evangelista, L. & de Brito, J., 2010. Durability performance of concrete made with fine recycled concrete aggregates. Cement & Concrete Composites 32 , pp. 9-14. Silva, R., de Brito, J. & R.K. Dhir, 2014a. Properties and composition of recycled aggregates from construction and demolition waste suitable for concrete production. Construction and Building Materials 65, pp. 201217. Zhao, Z., Remond, S., Damidot, D. & Xu, W., 2015. Influence of fine recycled concrete aggregates on the properties of mortars. Construction and Building Materials 81, p. 179–186. Akash Rao, K. N. J. ,. S. M., 2007. Use of aggregates from recycled construction and demolition waste in concrete. Resources, Conservation and Recycling 50, p. 71–81. Davidovits, J., 2015. Environmental implications of Geopolymers. [Online] Available at: http://bit.ly/21Z1byQ [Accessed 15 January 2016]. Fawer, M. C. M. R. W., 1999. Life cycle inventories for the production of sodium silicates. The International Journal of Life Cycle Assessment, 4(4). Gerardu , J. & Hendricks, C., 1985. Recycling of road pavement materials in Netherlands. Habert, G. d. L. J. B. N., 2011. An environmental evaluation of geopolymer based concrete production: reviewing current research trends. Journal of Cleaner Production, 19(11). Hansen, T. C., 1986. Recycled aggregates and recycled aggregate concrete second state-of-theart report developments 1945-1985. Materials and Structures, 19(3), pp. 201-246. Heath, A. P. K. M. M., 2014. Minimising the global warming potential of clay based geopolymers. Journal of Cleaner Production, Volume 78. Pepe, M., Koendersb, E. A., Faellaa, C. & Martinellia, E., 2014. Structural concrete made with recycled aggregates: Hydration process and compressive strength models. Mechanics Research Communications 58, p. 139–145. Silva, R., de Brito, J. & Dhir, R., 2014a. The influence of the use of recycled aggregates on the compressive strength of concrete: a review. European Journal of Environmental and Civil Engineering, p. DOI 10.1080/19648189.2014.974831. Wai , H. K., Mahyuddin , R., Kenn , J. K. & Mohd , Z. S., 2012. Influence of the amount of recycled coarse aggregate in concrete design. Construction and Building Materials,, Volume 26, p. 565–573. Zega, C. & di Maio , A., 2011. Use of recycled fine aggegate in concretes with durable requirements. Waste Management 31, pp. 2336-2340.

FRA demonstrated suitable physical properties for many applications. Although they were inferior to those of natural sand they were still within acceptable limits for recycled aggregates. Their high water absorption, which was attributed to their fineness and the nature of the original material, influenced most of the final mortar properties in both direct and indirect ways The physical properties of similarly produced PC FRA proved to be inferior, signifying that, despite the values obtained in the present case, the use GPC FRA is viable. The workability of final mixes was influenced in a way acceptable for the case of recycled aggregate mortars. With adaptation of the FRA percentage and appropriate mix design this effect could be easily overcome. The 28day water absorption of PC mixes appeared to increase proportionally to the replacement level, while in GPC mixes the increase of FRA percentage did not result to significant fluctuations. Increase of mortars’ water absorption is generally accepted to occur with FRA use. In both matrices though the increase for low replacement level was not significantly high. The compressive strength of both mortar times exhibited slight enhancement with 25% GPC FRA addition. For high replacement level the effect was less favorable but still resulted to values almost equal to those of the reference mixes. The same trend was observed for flexural strength. Overall, the effect of GPC FRA on both PC and GPC mortars was similar, while when compared to PC FRA they did not appear to be inferior in any aspect. For low replacement levels, all the investigated properties presented acceptable deviations from the reference mix. Considering that in most real life cases the percentage of recycled aggregates in mixes does not exceed 30-35%, with adaptation of the mix design the above results indicate that their use will be feasible. The present study was limited in the investigation of specific physical and mechanical properties on mortars alone. Therefore the obtained results should be expanded in order to take into consideration the intrinsic structure and chemistry of geopolymer cements. Chemical aspects and microstructural parameters should be included into the investigation, while the potential of recycling GPC mortars and concretes with various matrix types has to be evaluated. References Davidovits, J., 2013. Geopolymer Cement:A Review, France,Saint-Quentin: Institute Geopolymere. Dhir, R. K., Limbachiya, M. C. & Leelawat, T., 1999. Suitability of recycled concrete for use in BS 5328 designated mixes. Proc. Instn Civ.Engrs Structs & Bldgs,, Augoust , pp. 257 - 274.

4

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 17

Use of superabsorbent polymers (SAP) to reduce autogenous shrinkage in mortars with ground granulated blast-furnace slag (GGBS) F.C.R. Almeida and A.J. Klemm School of Engineering and Built Environment, Glasgow Caledonian University, UK G. Moriconi and M. Cingolani Facoltà di Ingegneria, Università Politecnica delle Marche, ITALY

ABSTRACT The proposed study argues that superabsorbent polymers (SAP) can be used for internal curing and reduction of cracking susceptibility, due to their high capacity of water control in fresh state. SAP efficiency in autogenous shrinkage reduction in blended cements containing GGBS has been evaluated. The study was focused on mortars with four levels of cement replacement by GGBS (0%, 25%, 50% and 75%) and two types of SAP with different water absorption capacity. Tests of consistency (flow table method), air content of fresh mortar (pressure method) and autogenous shrinkage (corrugated tubes method) have been carried out. The results showed that SAP can significantly reduce autogenous shrinkage for any studied GGBS content in comparison with the reference mortar. Beside the positive environmental impact by usage of GGBS, the reduction of autogenous shrinkage by SAP may decrease the cracking susceptibility and hence increase the sustainability level of the material for more durable constructions.

application of superabsorbent polymers (SAP) as internal curing agents (Mechtcherine & Reinhardt 2012). SAP has high capacity to absorb water from fresh mix and to release it in either fresh or hardened state. Figures 1 and 2 show SEM micrographs of two types of SAP used in the experimental study in dry and wet conditions respectively.

1. INTRODUCTION Cementitious materials (concrete and mortars) are the most widely used construction materials in the world. However, the production of Portland cement demands high energy and leads to emission of significant amount of carbon dioxide into the atmosphere. In an attempt to reduce its impact on the environment and improve sustainability of Portland cement-based materials, supplementary cementitious materials, such as ground granulated blast-furnace slag (GGBS), are often used in concrete mixes. Utilization of these by-products not only prevents them from being land-filled and reduces cement contents, but also improves workability, increases strength and enhances durability of concrete (Siddique & Khan 2011).

Figure 1. SEM micrographs of SAP I in dry and wet conditions respectively.

In addition, a sustainable material may be understood as one with minimum societal impact during its entire life cycle, considering its short-term and long-term effects. In this way, improving sustainability means producing more durable materials (Naik & Moriconi 2005). In case of concrete, one of the major concerns about durability loss is related to cracking formation. In general, cementitious materials are very susceptible to cracking provoked by autogenous shrinkage due to their associated selfdesiccation processes. However, this negative effect of shrinkage may be prevented by

Figure 2. SEM micrographs of SAP II in dry and wet conditions respectively.

This ability to provide water over the hydration process can lead to enhancement of some properties of concrete, such as, reduction in 1

autogenous shrinkage. (Hasholt et al. 2012; Klemm et al. 2012; Mechtcherine et al. 2013; Snoeck et al. 2015).

3. RESULTS Figure 3 shows the results of flow test (consistency) and air content. In general, SAP addition reduced consistency and increased air content of fresh state mortars with GGBS.

Although the use of SAPs as internal curing agents in cementitious materials has been previously investigated (Jensen & Hansen 2001; Mechtcherine & Reinhardt 2012; Mechtcherine et al. 2013; Klemm & Sikora 2013; Beushausen et al. 2014) their application in blended cements with GGBS is not clearly articulated and cohesive.

0% GGBS

Therefore, this study aimed to evaluate efficiency of two types of SAPs (with different water absorption capacities) in autogenous shrinkage reduction in mortars with blended cements containing GGBS. 2. MATERIALS AND METHODS Different mix proportions of mortars were produced for the experimental study, considering four levels of Portland cement (CEM I 52.5N) replacement by GGBS: 0%, 25%, 50% and 75% by mass.

25% GGBS

For each mix, two different types of SAP have been used in the proportion of 0.25% by mass of binder: SAP I (Figure 1) and SAP II (Figure 2) respectively with water absorption capacity of 25-30 g/g and 35 g/g in cement paste solution. Fine sand contained 99.95% of particles below 0.6 mm (EN 13139:2013). Proportion of 1:2 (binder : sand) and water/binder ratio (w/b) of 0.5 have been adopted for all mixes. Table 1 shows identification of mortar samples based on the type of SAP and GGBS content.

50% GGBS

Table 1. Identification of mortar samples. Sample name

Type of SAP

CEM I (%)

GGBS (%)

R-0

-

100

0

R-25

-

75

25

R-50

-

50

50

R-75

-

25

75

I-0

I

100

0

I-25

I

75

25

I-50

I

50

50 75

I-75

I

25

II-0

II

100

0

II-25

II

75

25

II-50

II

50

50

II-75

II

25

75

75% GGBS

Tests of consistency (flow table method, EN 10153:1999) and air content (pressure method, EN 1015-7:1998) were carried out for fresh mortar o mixes in laboratory environment (T = 21±2 C and RH = 40±5%). Autogenous shrinkage was tested by the corrugated tubes method (ASTM C1698:2009) from the final setting time until the age of 14 days, using a digital bench dilatometer.

Figure 3. Flow test and air content results – comparison by the same level of cement replacement by GGBS.

All samples with SAP had their consistency reduced significantly in comparison with the 2

respective reference mortar with same GGBS content (grey dashed reference lines in Figure 3). This reduction is due to high water absorption capacity of SAP. The driest mortars were those with SAP II which has the highest absorption capacity between both SAPs studied. Regarding air content testing results, all mortars modified by SAP have incorporated air in the fresh mixture compared to reference samples. When SAP absorbs water from the mix it swells up and becomes a gel. This process leads to introduction of air into the mix, resulting in greater air content values. All samples with SAP presented values above the respective black dotted reference lines in Figure 3. Mortars with SAP II contain more air in comparison with samples modified by SAP I. It may indicate that higher absorption capacity of SAP II is able to have a greater swelling ability and hence to increase air content of the mix.

0% GGBS

This effect of SAP on air content can be related to flow results in an inversely proportional way. The greater consistency the lower air content values, regardless GGBS level. Reference samples had the highest flow values and the lowest air content. Therefore, water absorption by SAP may cause the mixture to be drier and contain more air.

25% GGBS

Figure 4 shows the results of autogenous shrinkage during the first 14 days. Overall, SAPs clearly reduced autogenous shrinkage compared to reference mortars. Considering samples without SAP, autogenous shrinkage steadily increased for different GGBS contents. The higher the level of cement replacement by slag, the greater is autogenous shrinkage. At the end of 14 days, there was shrinkage about 300, 350, 400, 500 µm/m for reference mortars with 0%, 25%, 50% and 75% of GGBS, respectively. This behaviour was also verified by other authors; the greater values of autogenous shrinkage with GGBS may be due to its higher degree of hydration, and therefore, a greater degree of self-desiccation (Jiang et al. 2014). According to these authors, the greater chemical shrinkage led to faster and greater selfdesiccation, and results in larger autogenous shrinkage.

50% GGBS

75% GGBS

However, in the presence of SAPs, this behaviour was sharply altered showing a significant reduction in autogenous shrinkage for all levels of cement replacement by GGBS. Autogenous shrinkage was mitigated when using SAPs as internal curing in mortars with GGBS. It was also verified by others authors who studied the effect of SAP on cement paste with slag (Snoeck et al. 2015). The principle of internal curing by means of SAP relates to the provision of water-filled cavities in hardened mortars, reducing autogenous shrinkage (Beushausen et al. 2014).

Figure 4. Autogenous shrinkage results – comparison by the same level of cement replacement by GGBS.

Besides the effect of water provision for internal curing, autogenous shrinkage may also be influenced by more porous structure caused by SAP in fresh mixtures, as seen in Figure 3. The incorporation of air can lead to less compacted mixtures, resulting in more porous structure of matrices. Consequently, autogenous shrinkage is expected to be reduced due to the larger capillaries 3

that lead to lower tensile stresses generated by water menisci in the capillaries. The surface tension of water, caused by the loss of water in the capillary pores, leads to increased attraction forces between pore walls, resulting in shrinkage (Valcuende et al. 2015).

References ASTM Standard, 2009. Standard test method for autogenous strain of cement paste and mortar. ASTM C-1698. Beushausen, H. et al., 2014. The influence of superabsorbent polymers on strength and durability properties of blended cement mortars. Cement and Concrete Composites, 52:73–80. BSI, 1998. Methods of test for mortar for masonry – part 7: determination of air content of fresh mortar. BS EN 1015-7. BSI, 1999. Methods of test for mortar for masonry – part 3: determination of consistence of fresh mortar (by flow table). BS EN 1015-3. BSI, 2013. Aggregates for mortar. BS EN 13139. Hasholt, M.T. et al., 2012. Can superabsorbent polymers mitigate autogenous shrinkage of internally cured concrete without compromising the strength? Construction and Building Materials, 31:226–230. Jensen, O.M. & Hansen, P.F., 2001. Waterentrained cement-based materials. Cement and Concrete Research, 31(4):647–654. Jiang, C. et al., 2014. Autogenous shrinkage of high performance concrete containing mineral admixtures under different curing temperatures. Construction and Building Materials, 61:260–269. Klemm, A.J. et al., 2012. The effect of super absorbent polymers on the performance of immature cementitious mortars. In Brittle Matrix Composites 10. Warsaw: Elsevier, pp. 21–31. Klemm, A.J. & Sikora, K.S., 2013. The effect of Superabsorbent Polymers (SAP) on microstructure and mechanical properties of fly ash cementitious mortars. Construction and Building Materials, 49:134–143. Mechtcherine, V. et al., 2013. Effect of internal curing by using superabsorbent polymers (SAP) on autogenous shrinkage and other properties of a high-performance fine-grained concrete: results of a RILEM round-robin test. Materials and Structures, 47(3):541–562. Mechtcherine, V. & Reinhardt, H.W., 2012. Application of Superabsorbent Polymers (SAP) in Concrete Construction, Springer. Naik, T.R. & Moriconi, G., 2005. Environmentalfriendly durable concrete made with recycled materials for sustainable concrete construction. In: CANMET/ACI Int. Symposium on Sustainable Development of Cement and Concrete, Toronto. Siddique, R. & Khan, M.I., 2011, Supplementary cementing materials, Springer. Snoeck, D. et al., 2015. The influence of superabsorbent polymers on the autogenous shrinkage properties of cement pastes with supplementary cementitious materials. Cement and Concrete Research, 74:59–67. Valcuende, M. et al., 2015. Shrinkage of selfcompacting concrete made with blast furnace slag as fine aggregate. Construction and Building Materials, 76:1–9.

Moreover, concrete with SAP has less “free water” to contribute to these tensile stresses. This water is absorbed by SAP resulting in drier mixes (Figure 3). Therefore, the higher air content and the lower “free water”, the tensile stresses in the capillaries are lower and mortar is less prone to deformation. Regarding the studied samples (Figure 4), SAP II seemed to have better performance than SAP I, indicating different efficiencies for different polymers. Mortars with both SAPs have varied shrinkage values between 100 and 200 µm/m at 14 days. These values represent reductions in order of 50% to 70% for SAP I and 65% to 75% for SAP II, in comparison with reference mortars with same GGBS content. It seems that SAP II, which has higher water absorption capacity, is able to retain and provide more water for binder hydration. It contributes to a more effective internal curing and hence, results in a decreased autogenous shrinkage. The reduction in autogenous shrinkage is more pronounced as the slag content is increased. It suggests that SAP has a positive effect on autogenous shrinkage reduction, especially for high contents of slag when this diminution is maximized. 3. CONCLUSIONS Superabsorbent polymers (SAP) have been efficient in controlling internal curing in fresh state mortars. SAP absorbs water from the fresh mix and releases it during hydration process, aiding to mitigate autogenous shrinkage. This water absorption leads to drier mixtures (with lower consistency) as well as it triggers off a slight swelling of SAP, resulting in greater air content in fresh state mortar. Mitigation in autogenous shrinkage is more significant for higher contents of GGBS, where the degree of self-desiccation is greater in relation to Portland cement. Therefore, the reduction in autogenous shrinkage by SAP may decrease the cracking susceptibility and hence increase the sustainability level of the material for more durable constructions. Acknowledgment The authors acknowledge CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico - Brazil) for the financial support and BASF for SAPs supply.

4

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 18

Finnish Clay Based Pozzolan and Dehydrated Cement Paste as Cement Replacement Materials K. Loimula Lifetime Management, VTT Technical Research Centre of Finland Ltd

ABSTRACT Cement manufacturing is responsible for a serious amount of CO 2 -emissions worldwide. Emissions related to concrete structures can be reduced by a partial replacement of cement with a more environmentally friendly material when constructing concrete structures. In this study, a Finnish clay based pozzolan, two kaolins and dehydrated cement paste were used as cement replacement materials. Heat treatments of 3h at 800°C and 2h at 750°C were applied to the clay and cement paste, respectively. The kaolins were heat treated 3h at 600°C. All heat treated materials performed well as cement replacement materials. The heat treated clay had the required (EN 450-1) activity index at 28 days of age. Dehydrated cement paste and the heat treated kaolins met the requirements at all test ages (28 and 91 days).

treated in order to regenerate reactive calcium silicates. The heat treated product was used as a cement replacement material and its performance was evaluated.

1. INTRODUCTION It has been estimated that up to 5% of the global CO 2 -emissions are caused by the cement manufacturing industry. Cement is needed for building concrete structures and can be partially replaced to reduce emissions related to these structures. (Worrell et al. 2001)

2. Experimental The kaolins and clay were characterized with X-ray fluorescence (XRF) and X-ray diffraction (XRD) techniques. Elemental and mineralogical compositions were obtained. The thermal behaviour of the kaolins and clay was characterized with thermogravimetry (TG-DTA). The heat treatment temperatures for these materials were chosen based on the TG-DTA curves.

Pozzolans can be used as cement replacement materials. Pozzolans are siliceous or aluminous and siliceous materials that can react with calcium hydroxide to form calcium silicate hydrate (C-S-H). Calcium hydroxide is a hydration product in normal cement hydration and C-S-H is the binding phase in concrete, giving the structure its strength. (Taylor 1997)

The kaolins and clay were decided to give heat treatments of 3h at 500, 600, 700 and 800°C. The hydrated cement paste was decided to give heat treatments of 2h at 250, 450 and 750°C. The heat treatment durations of 2 and 3 hours and heat treatment temperatures for hydrated cement paste were chosen based on literature. The heat treated products were further examined with XRD to determine the optimal heat treatment temperature for each material.

Metakaolin is a well-known pozzolan. It is manufactured by heat treating kaolin clays (clays containing kaolinite). Kaolinite is a crystalline clay mineral and heat treatment at sufficient temperatures breaks down its crystalline structure to give an amorphous material with pozzolanic properties. Other clay minerals, than kaolinite, can also be heat treated to obtain materials with pozzolanic properties. (Siddique 2008, He et al. 1995)

The optimal heat treatment temperature for both kaolins was concluded to be 600°C while the optimal heat treatment temperature for clay was found to be 800°C. Hydrated cement paste was heat treated in 750°C based on XRD analysis and literature.

In this study, two kaolins of different quality were heat treated to obtain metakaolins of different quality. Finnish clay containing no kaolinite was also heat treated in order to obtain a material with pozzolanic properties. These pozzolans and a commercial metakaolin were used as cement replacement materials and their performance was evaluated. Hydrated cement paste was also heat

The heat treated products along with commercial metakaolin were used as cement replacement materials replacing 25% of the cement used in a reference mortar (REF1). Another reference mortar 1

(REF2) was made with 25% replacement with inert material (quartz). The mortars were tested for compressive strength at 7, 28 and 91 days of age (EN 196-1). Activity indexes were calculated (EN 450-1) and statistical analysis was performed on the compressive strength results. (Loimula 2013)

Activity index of 100% equals the performance of mortar without any cement replacement (REF1). The requirement for activity indexes in standard EN 450-1 is 75 and 85% at 28 and 91 days, respectively. REF2 with inert cement replacement material fails to meet these requirements, as expected. CLAY with heat treated Finnish clay as a cement replacement material meets the requirements at 28 days of age but performs slightly below the requirement at 91 days of age. KAO1, KAO2, META and HCP meet the requirements at both test ages.

3. Results The mix designs, abbreviations and compressive strength results are presented in Table 1. Compressive strength (CS) values are averages of 9 measurements at all test ages.

Analysis of variance was performed on the compressive strength averages at all test ages to investigate if the differences between mortars are statistically significant. At 7 days, HCP and REF1 were statistically the same (95% confidence level). No statistical difference was observed between KAO2 and HCP either. At 28 days, HCP, REF1 and KAO1 were statistically the same. The difference between HCP and KAO1 at the age of 91 days was not statistically significant.

Table 1. Mix design data and compressive strength results. Cement CS CS CS replacement 7d 28d 91d Mix material (MPa) (MPa) (MPa) REF1

-

44.8

51.1

56.9

REF2

quartz

20.2

24.4

30.9

KAO1

heat treated kaolin no1

37.8

50.2

52.5

KAO2

heat treated kaolin no2

47.3

63.3

68.7

META

commercial metakaolin

53.8

69.0

71.6

CLAY

heat treated clay

35.3

45.1

47.4

HCP

heat treated cement paste

46.1

52.5

52.4

4. Conclusions Heat treatment was successfully applied to kaolins, Finnish clay and hydrated cement paste to obtain cement replacement materials. Heat treated Finnish clay performed adequately at 28 days of age as a cement replacement material but slightly failed to meet the requirements at 91 days of age. Heat treated hydrated cement paste performed well as a cement replacement material at all test ages.

Activity indexes were calculated according to equation (1).

References Worrell, E et al., 2001. Carbon dioxide emissions from the global cement industry. Annual Review of Energy and the Environment, 26:303-329.

(1) where, AI = activity index CS = compressive strength CS REF = compressive strength of REF1

nd

Taylor, H.F.W., 1997. Cement chemistry. 2 edition, Thomas Telford, London. Siddique, R., 2008. Waste materials and byproducts in concrete. Springer.

Activity indexes for all test mixes at all test ages are presented in Figure 1.

He, C.et al., 1995. Pozzolanic reaction of six principal clay minerals: activation, reactivity assessments and technological effects. Cement and Concrete Research, 25:1691-1702. Loimula, K., 2013. Applicability of Finnish clay as a raw material for pozzolans. Master’s Thesis, Aalto University, School of Chemical Technology.

Figure 1. Activity indexes for all test mixes at all test ages. Connection lines between data points do not have physical meaning.

2

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 19

Potentials of Ultra-Fine Fly Ash Concrete B. K. T. Kandie Department of Civil and Structural Engineering, Masinde Muliro University of Science and Technology Abdul Mullick Associates Limited, Consulting Engineers K. Pilakoutas Department of Civil and Structural Engineering, University of Sheffield ABSTRACT This paper presents an experimental investigation on the potentials of Ultra-Fine Fly Ash (UFFA) application up to 50% partial replacement of cement in concrete. UFFA is a new generation fine coal fly ash, which is double-classified to obtain the finer particles from the classified bulk. The effects on fresh (water demand, setting time, flow and air), porosity of mid to high strength, medium workability concrete mixes (designed for equal 28-days strength) and compared with concrete mixes with Ordinary Portland Cement (OPC), MicroSilica (MS) and Seven BS 3892 Part 1 Pulverized Fly Ashes (PFA). The research study shows that UFFA has an immense significant higher pozzolanicity than all United Kingdom (UK) fly ashes tested and significantly reduced the water demand and decreased air content and porosity of concrete. The compressive strength results indicated that a partial replacement of OPC with UFFA can not only reduce the total binder content required to obtain target strength at specific ages irrespective of curing, but also increased the efficiency of combination binder by up to 50% at latter ages. A high strength of 2 130 N/mm was attained at 28 days for blend of UFFA/MS/OPC concrete with a combination of selected aggregates sizes proportions. The incorporation of UFFA improved durability due to the enhanced resistance to Chloride attack and Carbonation compared to equal strength concrete made with OPC and UK PFA, irrespective of curing environment. very finest fly ash particles, typically with a mean 1. INTRODUCTION particles size of 2.32 μm and a specific surface 2 area of 30,000 cm /g. Portland cement is the principle hydraulic binder used in modern concrete practice and 2. AIM AND OBJECTIVE manufactured by energy intensive process that consumes up to 5.29 GJ/t, mostly obtained from The aim of this investigation was to characterize burning of fossil fuel, is consumed [1]. The the UFFA powder, fresh, hard engineering and production of every tonne of Portland Cement durability performance of UFFA concrete in emits 0.89 to 1.1 tonne of carbon dioxide, thus comparison to concrete made with OPC, a range of contributing to the greenhouse gas emission that is UK fly ashes and Micro Silica. responsible for the global warming [2]. The objectives that were defined are as follows: The current production of fly ash at United 1. To assess the effects of UFFA on water Kingdom power plants is in the range of 8 – 10 Mt demand of concrete. per annum. Approximately 50% of this is used in 2. To optimize the aggregate grading used in industrial applications and the rest require disposal. combination with flowing concrete paste to The world situation is similar with an average of create cohesive flowing UFFA concrete 40% usage [3]. 3. To minimize the cement used in very high performance UFFA concrete mixes. The use of green concrete worldwide is being 4. To measure the compressive strength adopted in reducing both CO 2 emission and cost of development of UFFA concrete mixes. energy consumption and using of industrial waste To compare the economic and performance of as cement replacement is embraced by many UFFA concrete mixes with concrete made with governments. other cementitious materials of equal mix proportions. Ultra-Fine Fly Ash is a new style of Class F Fly Ash to ASTM 311. Which is double classified at the source to produce a product consisting at only the 1

Table 1. Chemical Properties of the cementitious materials used PERCENTAGE COMPOSITION Chemical composition

UK BS 3892 Part 1: PFAs 3

4

5

6

7

20.7

53.4

55.1

53

54.3

55.6

48.6

51.2

50.6

92

Aluminum oxide (Al2O3)

5.7

32.3

25.1

25.1

23.1

25

24.6

26.1

26.1

1

13

Ferric oxide (Fe2O3)

2.3

4.09

10.7

10.2

9.2

10.4

13

7.8

9.4

1

0.55

Calcium oxide (CaO)

64.8

5.05

1.6

3.8

4.1

1.9

2.8

6.4

3.7

0.3

39.53

1.1

1.29

1.5

2

2

1.6

1.7

1.9

2.2

0.6

8.28

0.19

45(µm) 0 6.5 11.1 10.1 6.3 14.3 9.3 14.2 0 0.5

Particle Diameter D[v,0.1] D[v,0.5] D[v,0.9] 0.4 1.5 3.37 2.14 1.58 2.43 2.24 3.65 0.25 0.9

2.32 8.47 15.82 12.18 9.2 14.97 12.79 15.28 1.03 6.64

5.28 36.81 46.33 44.56 36.5 52.1 42.91 52.19 22.27 19.52

Span 2.1 4.17 2.72 3.48 3.8 3.32 3.18 23.05 21.36 2.8

polyethylene sheet to prevent evaporation. Specimens were cast in steel in 100 mm steel cube, 100 mm diameter by 200 mm cylinder and 100 mm by 100 mm by 500 mm prism moulds and were cured in water to air environment after demoulding after 1 day.

3. EXPERIMENTAL INVESTIGATION 3.1. Materials The cement was Portland Cement confirming to BS 2 12 1996, class 42.5 N/mm . Seven UK Class F fly ashes confirming to BS 3892, Part 1: 1997. Micro Silica was an aqueous slurry (50/50 by weight) and UFFA used a commercially available Class F fly ash confirming to BS 3892 Part 1: 1997. The chemical compositions are shown in Table 1. The aggregates were 10 and 20 mm river aggregates and sand. A Naphthalene-base superplasticizer to BS 5075, Part 3 was used to achieve a concrete workability level at 125 mm slump. Figure 1 shows the various Class F ashes researched on.

BS3892: Part 1 PFAs 3 4 5 2

6 7

1

SPFA 3 µm

Figure 1. UFFA and various types of UK PFAs.

3.2. Concrete mixes

3.3. Curing conditions

A series of concrete mixes were prepared for grade 2 40-80 N/mm , with w/b ration of 0.21 – 0.56 and a 125 mm slump target. UFFA was used in the proportion of 0 - 50% of total cementitious materials. The cementitious materials tested included OPC 42.5 N [BS 12:1992], commercially available UFFA to 3892: Part 1, 7 UK fly ashes to 3892: Part 1 and Micro Silica in slurry form (Table 1). The fresh samples were covered with

The curing conditions used are listed in Table 2. 3.4. Determination of particle distribution A laser Malvern Mastersizer E was used to measure the particle size and surface area of the various pozzolans and cementitious materials.

2

3.5. Determination of water demand This test was undertaken in accordance to BS 3892: Part 1: 1993. Fine, medium and coarse sand was mixed in accordance to BS EN 196: Part 1: 1987. 3.6. Determination of compressive strength The compressive strength in accordance to BS 1881: Part 116: 1983 on three 100 mm cubes at 7 and 28 days of full water immersion samples. The air-cured cubes were saturated prior to test by immersion in water tank for 24 hours (7 day test) or 48 hours (28 day test).

Specific Surface Area m2/kg

Table 3 and figure 2 shows the particle size distribution of the various pozzolans used. UFFA has higher specific surface area than the UK. BS 3892 Part 1 PFA and GGBS which was in the range 2 2 of 340 to 715 m /kg and 1,448 m /kg respectively, but SM had the highest surface area. UFFA had a fineness of 0% with a mean particle size of 2.32 µm and 90% of the particles were 5.25 µm, this was less than that obtained from BS 3892 Part 1 PFA and GGBS but similar to MS. The fineness of the various fly ashes showed relationship with its content in respect to particle fineness. Figure 3 (a) shows that there is a linear relationship between the mean particle size and the fineness of fly ash and similarly with every increase in fineness there is an increase in the carbon content (Figure 3 (b)). The relationship between the specific surface area and the mean particle size was not linear but a curve, which was decreasing with the increase in the mean particle size due to the fly ash particle being more coarser in it morphology.

1 10 Particle Size, µm

% Passing

10

2000

8

1500

6 4

1000

2

500

0

0

-2 5

10 15 20 Mean Particle Size (ADM) (a)

25

3000 2500 2000 1500 1000 500 0 0

2

4 Carbon content % (b)

6

4.2. Water demand Figure 4 shows the water demand for cement, various pazzolans and the cementitious materials that were used in this investigation. UFFA had the lowers water demand and 29% was the maximum water reduction at 95% replacement level of PC42.5N. The seven UK fly ashes had a variation of water demand and these were between the upper and lower bound zone. GGBS showed a slight reduction in the water demand up to a maximum reduction of 4% while MS increased the water demand as its partial replacement level of PC 42.5N was increased, with an increase of 20% and 71% for 5% and 10% replacement level.

UFFA PFA01 PFA02 PFA03 PFA04 PFA05 PFA06 PFA07 GGBS MS

0.1

12

Figure 3. Fly ash physical characteristics.

SAMPLE PARTICLE SIZE ANALYSIS

0 0.01

14

3500

4.1. Particle size distributions

50

ADM

2500

0

4. RESULTS AND DISCUSSIONS

100

16 Sp. S. Area

3000

Fineness >45 microns

Specific Surface Area m2/kg

3500

4.3. Workability

100

All the concrete mixes were designed for a slump of 125 mm for an equal 28 days compressive strength. The workability was measured at different rates using the Tattersall two-point workability test and the flow table. From table 5, we note that as the percentage of UFFA replacement of cement increased, there was a reduction in yield value (g) and plastic viscosity (h). This reduction of g and h was due to UFFA replacement levels is as a result of an UFFA supplying additional lubricant round the grains of aggregates, due to the “ball bearing effect” of the ultra fine particle size, spherical shape and low carbon content of UFFA. The application of superplasticizers also aided the decrease in both g and h value for all the cementitious materials.

1000

Figure 2. Particle size distribution Analysis

A global fly ash characteristic may be predicted from figure 3 by knowing one of the elements of any type of fly ash. 3

Table 4. Composition and the compressive strength of concretes. Mix composition kg/m3 MIX CODE

OPC UFFA

MS

OPC-

522

UFFA/MS 1*-

368

79

79

UFFA/MS 2*+

368

79

79

Cement Save

w

Compressive Strength (Mpa) Aggregates

w/b

Age of testing (days)

120

20mm 15mm 10mm 5mm Sand 0.23 770 0 340 0 778

153

100

0.19

770

0

340

0

702

97

111.3

153

100

0.19

0

770

340

0

702

98.4

116.6

546

102.5

129.5

368 79 79 153 100 0.19 0 1110 0 156 UFFA/MS 3*+ * This were blend of 70% OPC 15%UFFA and 15%MS; + crushed gravel ; - river gravel

Table 5. The fresh properties of concrete.

Grade 40 N/mm2 OPC UFFA-05 UFFA-10 UFFA-30 UFFA 50 BS 3892 Part 1 PFS-30 MS-10

2.96 1.95 1.59 1.22 0.71 1.4 2.04

0.48 0.39 0.83 0.38 0.34 0.4 0.39

0.964 0.983 0.987 0.982 0.977 0.963 0.94

0.069 0.048 0.042 0.049 0.055 0.07 0.88

429 488 527 603 653 578 490

2.2 1.8 1.6 1.4 0.8 1.7 1.5

Grade 80 N/mm2 OPC UFFA-05 UFFA-10 UFFA-30 BS 3892 Part 1 PFS-30 MS-10

3.34 0.72 0.65 0.29 0.42 1.25

0.47 0.28 0.28 0.27 0.29 0.42

0.966 0.937 0.943 0.966 0.986 0.954

0.068 0.91 0.87 0.67 0.43 0.78

463 629 633 745 668 589

4.6 2.8 2.4 2 2.5 2.3

5. CONCLUSSION 1. UFFA as a cement binder improved the workability of concrete and this increased with the partial replacement levels of OPC as shown by reductions in ‘g’ and ‘h’ of these blended concrete. 2. UFFA had a maximum water reduction of 29% at 95% partial replacement level of PC42.5N, which was higher than that achieved by UK BS3892 Part 1 fly ashes and MS, which was in the upper and lower limit range and an increase of 71% respectively. 3. At the upper limits of conventional replacement level (40%) UFFA reduced water content by 23% whilst the best UK PFA tested reduced water demand by 11%. 4. UFFA has 0.4 μm and 5.28 μm of its particles size less than 10% and 90% by volume respectively. This was finer that the particles sizes of the seven UK fly ashes, GGBS and MS. 5. The highest compressive strength achieved was 2 129.5 N/mm by a blend of UFFA and MS at 15% partial cement replacement by weight. The w/b ratio was 0.19 and the aggregates used were crushed gravel of maximum size 15 mm.

120 UFFA BS 3892: Part 1. Fly Ash GGBS MS Upper bound

% Water required of OPC

110

100

UK PFA Water demand

90

Lower bound 80

70 0

20

40

60

80

28 95.1

(2002) at 28 days. The 5 mm coarse aggregates seemed to fill the voids between the coarse aggregates, a view that is supported by Kaplan (1959) who stated that compressive strength of concrete is higher than that of mortar and indicated that the mechanical interlocking of the coarse aggregate contributes to this. The binder of 3 UFFA/MS had 153 kg/m cement saving.

Two-point Test Results % Air Flow h g Std Error Content R2

Mixes Grade/Type

7 84.6

100

REFERENCES

Pozzolan replacement (%)

Figure 4. The water demand

Khurana, S., Banerjee, R. and Gaitonde, U. 2002 `Energy balance and cogeneration for a cement plant', Applied Thermal Engineering, Volume 22, page 485-494. Bouzoubaâ, N. Zhang, M. H. and Malhotra, V. M. 1998. Laboratory-produced high-volume fly ash blended cements: physical properties and compressive strength of mortars. Cement and Concrete Research, Volume 28, Issue 11, Pages 1555-1569. McCarthy M. J. and Dhir R.K 1999. Towards maximising the use of fly ash as a binder l, Cement and Concrete Research Volume 78, Issue 2, Pages 121-132.

4.6. Concrete compressive strength The mixes designs and the strength results are given in Table 4. Incorporation of cementitious material resulted in a reduction in water-to-binder ratio of the UFFA/MS concrete mixes. The strength of UFFA/MS concrete is higher than those of OPC concrete. This strength development showed that crushed gravel is better that river gravel. Due to the river gravel shearing faster than the crushed gravels. A combination of 15 mm crushed gravel and 5 mm of sub sand resulted in the highest 2 3 strength which was 129.5 N/mm at 526 kg/m binder at 28 days of testing. This was higher than 2 120 N/mm attained by Price (1996) and Tsartsari 4

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 20

Ground expanded perlite as an effective additive to common cements E. Kapeluszna, Ł. Kotwica, W. Nocuń-Wczelik Faculty of Materials Science and Ceramics, Department of Building Materials Technology, AGH University of Science and Technology, al. Mickiewicza 30, 30-059 Kraków, Poland ABSTRACT In the work the results of the experiments with ground perlite, by-product of expanded perlite production, as mineral addition to cement, are presented. The expanded perlite is used in the building materials technology mainly as a lightweight filler improving the thermoinsulation properties. Unfortunately, a huge amount of fine-grained by-product is delivered when the natural perlite is manufactured and processed. This by-product is a material of very low bulk density (50-150 kg/m3) and is a burdensome waste. Four cements were produced for experiments: industrial cement CEM I 42.5R and this cement with replacement of 5%, 20% and 35% by ground perlite by-product. The addition of perlite increased cement water demand, however this increase was significant only for the largest addition. The cement with perlite addition has longer setting time, but still in accordance with the EN-197-1 requirements. The compressive strength after 2 days was lower, however, after 28 days was higher, especially with 20% addition of perlite. Cement CEM I 42.5R with 5% and 20% addition of ground expanded perlite by-product complied the standard requirement for class 42.5R. The result is beneficial, particularly in the case of 20% addition.Therefore this material is potentially reactive pozzolan and hence attractive from the cement technology point of view. as a result of pozzolanic reaction. The expanded perlite reveals chemical composition similar as the commonly known pozzolans. It shows vitreous character, with high amount (about 70%) of silica and about 10% alumina content. There are some reports dealing with the expanded perlite as a lightweight filler in cement mortars, as well as with perlite rock used as a pozzolanic additive. The synthesis of zeolite from the expanded perlite waste was also studied.

1. INTRODUCTION Perlite is a naturally occurring volcanic glass containing about 5% by mass of water. The expandation process, performed at high temperature results in swelling of material because of the water vapour pressure impact. This phenomenon leads to the increase of volume and the structural changes, as compared to the crude material. In such a way the so-called expanded perlite is produced. The microstructure of expanded perlite, shown in fig. 1a, reveals the presence highly porous microstructure comprised of thin walls.

The presented work focused on the use of expanded perlite by-product as an additive playing the role of main and secondary component of common cements.

The expanded perlite is used in the building materials technology mainly as a lightweight filler improving the thermoinsulation properties. Unfortunately, a huge amount of fine-grained byproduct is delivered when the natural perlite is manufactured and processed. This by-product is a material of very low bulk density (50-150 kg/m3). Therefore several problems appear, dealing with the hazard of dusting, as the transport and storage are concerned.Pozzolanic materials used in cement production, such as silica fume, fly ash, natural and industrial pozzolans, have positive impact on the durability and mechanical characteristics of mortars. This is the consequence of improved pore structure in C-S-H phase formed

Figure 1. Microstructure of expanded perlite by-product under SEM – (a). Microstructure of expanded perlite by-product after 4h grinding in a ball mill examined under TEM – (b).

1

2.

Experimental

2.1.

Materials

The perlite – cement blends with 5, 20 and 35% of the ground expanded perlite, used as cement replacement, were produced. The binder with 5% perlite corresponds to the standard cement type CEM I, the binder with 20% additive corresponds to the standard cement type CEM II/A and the binder with 35% additive corresponds to the standard cement type CEM II/B respectively. Therefore the properties of cements with perlite additive could be related to those of the commercial standard cement. The standard Portland cement CEM I 42,5R was used as reference. The chemical composition and calculated phase composition, according to Bogue formulae is given in Table 1.

Figure 2. Grain size distribution of the ground expanded perlite by-product and commercial CEM I42.5R.

Basing on the grain size distribution curve (fig. 2) one can find that 25% of the ground expanded perlite by-product is composed of 0.06 ÷ 1.0 µm grains and 75% is composed of 1.0 ÷ 60 µm. There is a significant difference between the fineness of perlite and cement. Cement is much coarser – a few % only is in the range 0.06 ÷ 1.0 µm.

Table 1. The chemical and phase composition of CEM I 42,5R. Component, % by mass Phase, % by mass (Bogue) Na 2 O 0.4 C3S 52.2 MgO 1.4 Al 2 O 3 6.6 SiO 2 19.4 β- C 2 S 16.6 P2O5 0.3 SO 3 5.2 K2O 1.2 C3A 11.6 CaO 61.2 TiO 2 0.4 Fe 2 O 3 3.3 C 4 AF 10.6

The standard sand according to PN-EN 196-1:2006 and drinking water were used in the experiments. The cements were produced by mixing Portland cement CEM I 42.5R with the ground expanded perlite by-product. Standard CEM I 42.5R commercially available was used as reference. The mixtures were homogenized in a laboratory roller mixer during 12h. The cements thus produced are described based on the perlite content as CEM 5%, CEM 20% and CEM 35% respectively. The composition of mixtures is given in Table 3.

Because of very low bulk density the expanded perlite material was subjected to the following treatment: after the initial consolidation by pressing it was ground during 4 hours. The morphology of ground material is shown in fig. 1b. There are the flat, plate-like particles. The chemical composition and density of disintegrated expanded perlite was determined. The grain size distribution was characterized by means of laser granulometer MALVERN 2000.

Table 3. Cements composition. Cement Content [%] CEM I Ground expanded 42.5R perlite CEM I 100 -

Table 2. Chemical composition of the ground expanded perlite by-product. Component Content [ %] L.o.i.

0.6

SiO 2

76.2

Fe 2 O 3

1.3

Al 2 O 3

12.5

TiO 2

0.4

CaO

0.5

MgO

0.7

Na 2 O

3.1

K2O

4.5

CEM 5%

95

5

CEM 20% CEM 35%

80

20

65

35

2.2.

Methods

The homogenized cements were subjected to the determination of density by picnometric method (PN-EN 1097-7:2008), as well as to the specific surface measurements by Blaine method (PN-EN 196-6:2011). The heat evolution was followed directly in the non-adiabatic – non-isothermal BMR type calorimeter. The strength development was characterized after 2, 7 and 28 days maturing, according to the PN-EN 196-1:2006 standard. The microstructure of hardened mortars was observed under the scanning electron microscope.

2

3.

evolved after 40 h hydration is lower and the induction period is somewhat longer.

Results

In order to find the effect of the ground expanded perlite additive on the strength characteristics of mortars the compressive and flexural strength measurements were done. The mortars were prepared at cement/sand ratio is 3:1 and w/c=0.5, following the conditions given in the PN-EN 1961:2006 standard.

As one can see in table 4, the replacement of Portland cement clinker by the ground expanded perlite by-product results in the reduction of density. 3 The density of perlite (2.4 g/cm ) is about 30% lower than the density of reference cement CEM I 3 42.5R (3.1 g/cm ). The specific surface of blended cement grows with the percentage of the ground expanded perlite additive.

The results of compressing and flexural strength measurements vs. time and perlite contents are plotted in figs 5 and 6.

Table 4. Density and specific surface area of cements. Cement Density Specific surface [g/cm3] [cm2/g] CEM I 3.2 3740 CEM 5%

3.0

4010

CEM 20% CEM 35%

2.9

5240

2.8

5610

The heat of hydration process was examined with help of calorimetric method. The pastes were prepared at water to cement ratio 0.5 [figure 3]. Figure 5. Compressive strength of cement mortars.

The ground expanded perlite used as an additive in amount 5% does not reduce the compressive strength of mortars neither after 2, 7 nor after 28 days of maturing; the results do not decline from those for reference material. The higher percentage of additive brings about the strength reduction after 2 and 7 days. In the case of CEM 35% mortars the strength is about 30% lower at the same early age. On the other side, the 28-day strength of CEM 20%, as well as the strength of CEM 35% mortars is higher than the reference value. This strength increase is undoubtedly the consequence of pozzolanic reaction of perlite and higher C-S-H phase content.

Figure 3. Heat evolution curves of cements with different perlite addition.

At the same time it can be concluded that the cement CEM 5% and 20% comply with standard BS EN 197-1: 2012 for strength class 42.5R.

Figure 4. Total heat evolution of cements.

As one can see in figs 3 and 4 the ground expanded perlite by-product plays a role of accelerating additive from the beginning of hydration process. The additional peak in the time range between 12 h and 22 h reflects is presented. The total heat evolved value after 40 hydration is higher at 5 and 20% additive content [fig 4.]. In case of cement with 35% additive the total heat

Figure 6. Flexural strength of cement mortars.

The results of flexural strength measurements are well correlated with the compressive strength ones 3

Mladenovič A., Šuput J.S., Ducman V., Škapin A.S., 2004, Alkali–silica reactivity of some frequently used lightweight aggregates, Cem. Concr. Res. 34, 1809–1816. Lanzón M. Torres,. García-Ruiz P.A, 2009, Lightweight pozzolanic materials used in mortars: Evaluation of their influence on density, mechanical strength and water absorption, Cem. Concr. Comp. 31, 114–119. Sinan Turhan Erdoğan, Aslı Ünsal Sağlık, 2013, Early-age activation of cement pastes and mortars containing ground perlite as a pozzolan, Cem. Concr. Comp. 38, 29–39. Yu L.-H., Ou1 H., Lee L.-L., 2003, Investigation on pozzolanic effect of perlite powder in concreto, Cem. Concr. Res. 33, 73–76. Erdem T.K., ¸ Meral C., Tokyay M., Erdoğan T.Y., 2007, Use of perlite as a pozzolanic addition in producing blended cements, Cem. Concr. Comp. 29, 13–21. Pichór W., Mozgawa W., Król M., Adamczyk A., Synthesis of the zeolites on the lightweight aluminosilicate fillers, Mat. Res. Bull. 49, 210215.

[figure 6]. However, the flexural strength is slightly lowered at 5% perlite addition. The CEM 35% mortars reveal about 33% lower 2-day strength but after 7 days the decrease is only about 17%. This growth should be attributed to the effect of pozzolanic reaction. It is worthwhile to notice that after 28-day maturing the flexural strength of both CEM 20% and CEM 35% mortars is higher than in case of the other mortars. One can presume in this case again that this result can be attributed to the effect of pozzolanic reaction. The microstructure of mortars with CEM20% was observed under the scanning electron microscope [figure 7]. The active participation of perlite in the hydration process was found. This can be proved by the shortage of well visible boundaries between the perlite grains and C-S-H; this indicate that the product is formed as a layer covering the grains of expanded perlite.

Figure 7. Microstructure of mortar from cement CEM 20% cured 180 days in water at 20oC; a- perlite particles evenly distributed in cement matrix; b- perlite particles, for example 1 and 2.

4. Summary and conclusions The ground expanded perlite by-product can be successfully applied as an attractive additive to Portland cements. The ground expanded perlite by-product exhibit pozzolanic properties and brings about the standard 28-day compressive strength increase. Cements with 5% and 20% replacement of ground expanded perlite comply with standard requirements for class 42.5R. The result is very favourable particularly in the case of 20% replacement. Acknowledgement: The financial support from the Polish National R&D Centre is acknowledged (PBS I 177206) References Lanzón M., García-Ruiz P.A., 2008, Lightweight cement mortars: Advantages and inconveniences of expanded perlite and its influence on fresh and hardened state and durability, Constr. Build. Mat. 22, 1798–1806. 4

Young Researchers’ Forum III Innovation in Construction Materials 12 April 2016

Paper Number 21

Nanoscale characterization of CSH gel pores with scanning transmission electron microscopy 1

1

1,2

Olivia Wenzel , Matthias Schwotzer , Andreas Gerdes Institute of Functional Interfaces, Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany 2 University of Applied Sciences Karlsruhe, Institute for Prevention in Construction, Karlsruhe Germany 1

3

3,4

Torsten Scherer , Venkata S. K. Chakravadhanula Karlsruhe Nano Micro Facility, Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany 4 Helmholtz Institute Ulm (HIU), Ulm, Germany

3

ABSTRACT Transport processes between aqueous solutions and cement-based materials are a key component of reactions which lead to a reduction of durability. The transport processes within capillary pores are already well established as opposed to the influence of the pore system on the nanoscale. To determine the potential impact of the gel pores of the calcium silicate hydrate on material properties, a realistic picture of the entire pore structure is fundamental. However, due to the materials complexity and sensitivity, most analytical techniques have individual shortcomings and provide only a limited understanding of the system. In order to gain a more comprehensive understanding of the nanostructure of cement-based systems, a new approach using scanning transmission electron microscopy with focused ion beam prepared lamellas was developped. The challenge of this approach was the vulnerability of the material, in particular regarding the high content of instable hydrate mineral phases. Using this approach, it was possible to image the structure on the nanoscale including the pore system with a high depth of field and showing the connectivity of the CSH’s gel pores. This is discussed with results of other common investigation techinques for nanoporous systems, such as gas sorption and mercury intrusion porosimetry with regard to material performance aspects.

analytical approach in order to gain an in-depth model of the CSH using scanning transmission electron microscopy.

1. INTRODUCTION Cement-based materials are often exposed to different aqueous solutions during service-life, which lead to structural and chemical changes and, in turn, a reduction of their durability. Transport processes are a key component of these interactions. Generally, only capillary pores are considered in this context. However, chemical attacks on cement pastes result not only in a leaching of Ca(OH) 2 but also effect calcium silicate hydrate. Chen et al (2006) observed changes to the chemical composition and in turn, to structural properties. Therefore, one can assume such processes additionally alter the gel pore system. Current models describe the nanostructure of the CSH as an agglomeration of fine needles, which surround their clinker core. Some differentiate between an “inner” product, which is in the immediate vicinity of the clinker and an “outer” product further away (e.g. Rossen et al, 2015; Taylor, 1986). This inner product has a higher density attributable to progressive fractal crystallization (Jennings (2000)). Due to this heterogeneous nature, assessments of the nanostructure regarding chemical stability are barely possible. Common analytical techniques can only provide fragments of detailed information. Thus, the goal of this study is to develop a refined

2. EXPERIMENTAL APPROACH A nanostructural characterization of a model cement paste was performed to study the CSH in detail. This material was synthesized by hydrating pure C 3 S. The components were studied with conventional mineralogical techniques, such as xray diffraction, mercury intrusion porosimetry (MIP), thermogravimetric analysis, infra-red spectroscopy and environmental scanning electron microscopy. Scanning transmission electron microscopy (STEM) along with gas adsorption measurements (multi-point BET) were used to analyze the gel pores. STEM is an operating mode of conventional transmission electron microscopes where the electron beam is focused and scanned over a sample area with x/y scan coils between the lower condenser and objective lens. The transmitted electrons are detected according to their angle. In order to perform TEM or STEM investigations, the sample has to be electron transparent, which is approximately 50 nm. The observed pore structure is heterogeneous. Parts of the FIB lamella appear to have a lower density and a higher amount 20-50 nm gel pores as opposed to other regions close to the clinker. This fits the idea of an inner and outer product.

Figure 2. Top image: Characterization of the pores using BET and MIP (bottom image).

4. CONCLUSION These findings show another aspect of the CSH’s structure compared to conventional TEM studies, which depict finely intertwined foils building a dense structure, however, do not visualize the porous honeycomb structure through which transport seems feasible. These discretions can be attributed to the scanning mode and differing preparation technique. For conventional TEM investigations, the samples undergo arduous preparation steps including pressing epoxy resin into the sample, which could damage the pore structure. Further research will be devoted to the impact of the nanostructure on material properties with a focus on chemical functionalization of mineral surfaces. References Chen, J.J., Thomas, J.J.; Jennings, H.M., 2006. Decalcification shrinkage of cement paste. Cement and Concrete Research, 36(5): 801-809. Jenning, H. M., 2000. A model for the microstructure of calcium silicate hydrate in cement paste. Cement and Concrete Research, 30(1):101-116. Rossen, J.E., Lothenbach, B., Scrivener, K.L. 2015. Composition of C-S-H in pastes with increasing levels of silica fume addition. In: Cement and Concrete Research 75: 14-22 Taylor, H.F.W. 1997. Cement Chemistry. (2nd ed.) Thomas Telford.

Figure 1. STEM image of CSH.

The dominant pore width was measured around 15 nm. These gel pores touch and intersect each other producing a complex sponge-like pore network. Often smaller pores are visible within larger pores as shown in figure 1 and can appear spherical or slit-shaped depending on the angle. These