Potential Upgrade of Timber Frame Buildings in the ...
Potential Upgrade of Timber Frame Buildings in the UK Using Timber-Concrete Composites Rob GRANTHAM Senior Consultant Building Research Establishment, Watford, U.K. [email protected] Vahik ENJILY Massimo FRAGIACOMO Research Engineer University of Trieste, Trieste, Italy. [email protected] Claudio NOGAROL Ivan ZIDARIC Claudio AMADIO
Graduated in Civil & Structural Engineering in 1991. Worked in Timber Engineering at BRE since 1996. Currently lead a team of 5 conducting R&D projects on timber engineering, building science and environmental impacts of timber products. Born in Trieste (Italy) in 1967. Graduated in Civil Engineering in 1992. Ph.D. in structural engineering in 2001. Research Engineer since 1999. Author of about 50 papers focused on timber-concrete and steelconcrete composite beams, FE modelling, seismic engineering.
Summary Many modern timber framed buildings in the UK are constructed using platform frame techniques with the addition of floating floors for acoustic separation. This type of construction, commonly used for domestic multi-occupancy buildings, may be upgraded for reuse as an office building without the need for demolition of the existing building and replacement with a new steel or concrete framed structure. To achieve this, the authors propose replacement of the acoustic floor with a 60 mm lightweight concrete slab, connected to the existing joists with shear connectors to form Timber-Concrete Composite beams (TCC’s). This scenario has been tested on the TF2000 building at BRE Cardington and suggests that structural upgrading using TCC’s can achieve the required stiffness and up to six times the design load for satisfying design limit states for office loading in the UK. Keywords: collapse test, composite structures, creep, framed buildings, lightweight concrete, longterm behaviour, mechano-sorptive effect, rheological phenomena, shrinkage, timber.
1.
Introduction
Timber-concrete composite beam is a recent technique much used for strength and stiffness upgrading in both existing and new structures. It consists of connecting an existing or new timber beam with a concrete slab cast above a timber deck by means of a connection system [1]. The timber deck can then be employed as a formwork for the slab which is nominally reinforced with a simple anti-crack steel mesh. The slab should be cast after having propped the structure, to decrease stresses in the timber due to the structural dead load before concrete hardening. Experimental results demonstrate that the proposed upgrade technique is suitable for reuse of domestic buildings for office buildings. Both serviceability and ultimate limit states are fully satisfied. Thanks to the concrete slab, the floor is characterized by a larger stiffness, which reduces deflections and vibrations. The influence of the real restraint condition for the floor has also been
investigated. It is highlighted that the type of collapse is quite different with respect to the case of ideal hinge support.
2.
Test Set-up
125 476.5 600 600 600 600 600 600
An existing timber floor, with timber joists and Oriented Strand Board (OSB) decking, was converted to a TCC floor on the experimental TF2000 building [2] at BRE Cardington, UK. Two floor areas (A1 and A2 in Fig. 1) were upgraded using SFS VB-48-7.5x100 mm connectors in predrilled holes in the 38x225 mm timber joists, at the spacings and inclinations shown in Fig. 2. A plastic membrane laid on the existing 15 mm thick OSB decking provided the formwork for casting a 60 mm thick lightweight concrete slab. Nominal reinforcement for anti-crack purposes was provided by a 200x200 mm I6 steel mesh located 20 mm from the decking. Prior to casting the slab, the A2 floor was propped at one-third and two-thirds of the span along each joist. The A3 floor was left as original without casting a concrete slab. For each floor a reference joist was chosen and monitored in time starting from the concrete casting in order to evaluate the influence of rheological phenomena during the construction and loading history.
A1
A3
(Unpropped floor)
(Timber floor)
(Propped floor)
REFERENCE JOIST: J1
REFER. JOIST: J3
REFERENCE JOIST: J2
3440
2934
A2
3644
45 50 77 45 45
1720 585
765
108
55
No.4 connectors Inclination: 90°
No.14 connectors Inclination: 45° Spacing: 45 mm
No.9 connectors Inclination: 45° Spacing: 85 mm
Fig. 1 Plan view of the tested floors (dimensions Fig. 2 Connector arrangement for the in mm) unpropped floor (dimensions in mm)
3.
Material Test
Some laboratory tests were performed at BRE to fully characterise the component materials of the composite floors: lightweight concrete, timber and connection system. 3.1
Lightweight Concrete
A novel lightweight (LW) concrete was adopted for the slab incorporating re-processed sewage sludge for the course aggregate, similar to Lytag. The mix design ensured a strength class of LC 25/30 although standard compression tests on 100 mm cubes suggested a higher average strength of 46.7 MPa. The concrete fresh density was measured as 2020 kg/m3, its dry density as 1760 kg/m3 and the slump measured prior to pouring was 120mm. The medium values were evaluated for the following using two specimen results: Young’s modulus Elcm , cylindrical compression strength f lcm , and flexural tensile strength f lctm . The Young’s modulus of Lytag concrete was found to be about 60% that of a normal weight (NW) concrete with the same strength. Creep and shrinkage were measured on 100 mm by 200 mm diameter cylinders. Results are displayed in Fig. 3 for shrinkage strain and in Fig. 4 for the creep function at different times of loading t 0 . Dashed lines refer to predictions using Model Code 90 [3] for NW concrete. The shrinkage strain for Lytag concrete is much larger than for NW concrete (about four times). The creep function J (t , t 0 ) , is also larger compared to NW concrete. The creep coefficient for Lytag concrete ranges from 1 to 2.5 times that for NW concrete. Although further investigation is required, these tests highlight the larger sensitivity of Lytag LW concrete to rheological phenomena compared to NW concrete.
1000
200
Hcs(t)x106 [-]
800
J(t,t0)x106 [MPa-1]
160 t0=2
600
120 Exper. CEB 90
400
80
200
t [days]
0 0
20
40
60
80
t0=6
0
100
Fig. 3 Trend in time of shrinkage strain for Lytag lightweight concrete 3.2
Exper. CEB 90
40 t0=21 t0=35
0
20
40
t0=54 60
t [days] 80
100
Fig. 4 Trend in time of creep functions for Lytag concrete at different times of loading
Timber
The joists of the composite floor are classified as C16 European redwood kiln dried timber, according to BS EN 5268-2 [4]. Flexural strength f m and Young’s modulus E w were measured through four-point bending tests on eight 15x20x400 mm clear timber specimens cut from a joist in a region with few defects. Thermal expansion coefficients were obtained from seven specimens (5x5x210 mm) subject to relative humidity variations in a climate chamber and agreed closely with the value proposed by Toratti [5]. 3.3
Connection system
The connection system was tested at collapse in order to evaluate shear stiffness, shear strength and shear-slip relationship. The collapse test was performed according to EN 26891 [6] on three pushout specimens. The experimental shear-slip curves are displayed in Fig. 5 for one SFS VB-487.5x100 mm connector. A large stiffness can be realised up to 75% of the collapse load, beyond which an increase in plastic strains occurs up to the collapse, due to bond failure at the connectortimber interface. 16
1
V [kN]
M t [-]
0.8
12
Conn.-exp. Conn.-Kenel Timber-Toratti
0.6 8
0.4 4
0.2
slip [mm]
0 0
2
4
6
8
t [days]
0 10
0.01
0.1
1
10
100
1000
Fig. 5 Experimental shear vs. slip curves for Fig. 6 Creep coefficient for connection system connection system To investigate the rheological behaviour of the connection system, two push-out specimens were loaded with 25% of the collapse load applied through a lever mechanism. Such a load corresponds to the shear force that may be produced by the quasi-permanent load combination in a composite beam. The creep coefficients M are displayed in Fig. 6 for constant environmental conditions (70% RH, 24 °C). The results are in good agreement with those obtained by Kenel and Meierhofer [7] on the same type of connector differently arranged. The creep coefficient is about 60% larger
1003
compared to that of timber, as proposed by Toratti [5]. This confirms the importance of taking into account also the rheological behaviour of the connection [8].
4.
Long-term test
The monitoring of the full-scale floors began on the 6th of June when the concrete was poured on both propped (A2) and unpropped (A1) floors. For each reference joist the following quantities were monitored: vertical displacements at 5 locations through Linear Voltage Displacement Transducers (LVDT’s), strains of timber for the bottom fibre at mid-span through Demec gauges and Demec points, temperature of timber joist and concrete slab using thermocouples, moisture content of timber using electrical moisture probes, environmental temperature and relative humidity using a digital thermohygrometer. The purpose of monitoring the construction, before the live load application, was: to compare the floor behaviour for different types of construction (propped and unpropped), and to investigate the influence of rheological phenomena (creep and shrinkage) of materials when the concrete is still green, and hence more sensitive to these effects. After removal of the props, 14 days after the concrete casting, the live load of 2.5 kN/m2 required for office buildings was applied, using water pools for dead loading. A comparison between the propped (A2) and unpropped (A1) floor is displayed in Fig. 7 and Fig. 8 for both the mid-span deflection, v and bottom fibre timber strain, H . Comparing the two cases, propped and unproped, indicates a better behaviour for the propped floor with respect to the unpropped one ( v 5.3 mm instead of v 6.5 mm, and H 0.7 mm/m instead of H 1.4 mm/m immediately after the live load application). This agrees with recommendations given by many authors [1] for propping the composite floor to reduce deflections, however the reduction in deflection, for this case, was pretty small warranting use of the unpropped technique, which leads to cost savings for construction. The dead load due to the weight of the concrete slab and the rheological phenomena produced, during the first 34 days of testing, a mid-span deflection v p 3.9 mm ( v p l 1 934 , l being the span length) and v u 5.0 mm ( v u l 1 688 ) for propped and unpropped floors, respectively. The live load application produced an increase in deflection of 'v p 1.4 mm ( 'v p l 1 2603 ) and 'v u 1.5 mm ( 'v u l 1 2293 ) for the composite floors. The limitation v l d 1 333 [4] is met for all structures, however it is expected that the rheological phenomena will increase this ratio over longer periods. 8
1.5
v [mm]
Unpropped Propped
6
H [mm/m]
1
4
Unpropped Propped
0.5 Live load application
Props removal
2
t [days]
0 0
10
20
30
Props removal
0
40
Live load application
t [days]
-0.5 0
10
20
30
40
Fig. 7: Trend in time of mid-span deflection for Fig. 8: Trend in time of the timber bottom fibre the composite floors strain at mid-span for the composite floors
5.
Collapse test
Collapse tests were performed on the unpropped composite floor (A1 in Fig. 1) to evaluate the performance of the upgrade technique in terms of both stiffness and ultimate load and to estimate any contribution offered by the real support conditions for joists ‘clamped’ by wall panels (see Fig. 2). For the investigation, the composite floor was cut making two independent portions for test T1 and test T2; see Fig. 9. This eliminated unwanted restraint effects, allowing easier comparison of
experimental results. For test T1, the slab was completely disconnected from the building and supported under joist ends, on half rounds (Fig. 9). In this way, support conditions for test T1 were similar to a laboratory test assuming perfect hinges and no rotational restraint at supports. Test T2 was instead restrained at joist ends with the real joint, allowing assessment of the rotational restraint by comparison of the two tests. Concrete cut
Load rig
Test T1
Joist B1
146 300
Floor cut
Floor cut
600
Joist B2
300
Joist B1
300
Concrete cut
Test T2
600
Joist B2
Timber joist
Load rig
600
Joist B3
600
Joist B4
300
Concrete cut 1120
1200
330 1120
Fig. 9 Set-up of the collapse tests
Fig. 10 Test T1: Load rig
Fig. 11 Test T2: Load rigs
The loading rigs are depicted in Fig. 9, and pictured in Figs. 10 and 11 for the tests T1 and T2, respectively. For each test, the following quantities were monitored: vertical displacements, relative slip between concrete and timber joists near the supports, longitudinal strains at top and bottom fibres of concrete slab and timber joists. Load was applied according the schedule in EN 380 [9] and results are plotted, respectively, in Figs. 12 and 13 for Test T1 and T2 in terms of concentrate load on each joist vs. mid-span deflection. The curves concerning the case of rigid connection (fully composite behaviour) and no connection (non composite behaviour) between concrete and timber are drawn for comparison. 25
P [kN]
25
20
Joist B1
15 10
Non-composite
5
10
15
20
25
Joist B2 Non-comp.
10
Only timber
5
v [mm] 5
Joist B3 Joist B1
15
Only timber
0
Joist B4
Fully composite
20
Joist B2
0
P [kN]
Fully composite
v [mm]
0
30
0
10
20
30
40
Fig. 12 Test T1: Load vs. mid-span deflection Fig. 13 Test T2: Load vs. mid-span deflection curves curves Both tests demonstrate behaviour close to that of a fully composite floor during the first loadingunloading cycle. The effectiveness of the connection system is very good, with small vertical deflections and high stiffness. Small plastic strains occurred during the cycle, especially for test T1. Collapse was observed when timber joist B1 failed in flexure at a load P 18.3 kN ( q 22.0 kN/m2). Test T2 demonstrated a rather different behaviour. Longitudinal cracks appeared on the timber joists near the supports at a load P 11.6 kN ( q 15.3 kN/m2). These cracks, coincident with the line joining the lower connector ends (see Fig. 2), suddenly spread towards the
1005
mid-span of joists B1 and B3 at a load P 12.5 kN ( q 16.5 kN/m2). Although the serviceability limit state was compromised, the ultimate limit state was not reached until P 22.5 kN ( q 29.7 kN/m2). The increased flexibility, following longitudinal splitting (Fig. 13) changed the behaviour to almost that of a non-composite section. The comparison between tests T1 and T2 highlights that: - the presence of the real panel-to-joist joint worsens the behaviour of the composite floor with respect to the case of a simply supported joist, leading to an early longitudinal cracking of timber joists with loss of stiffness; - the performance of both tests was satisfactory: the collapse occurred for the test T1 under an equivalent uniformly distributed load q that was 8.8 times larger than the live load for office buildings. In the test T2, corresponding to the real floor behaviour, longitudinal timber cracking occurred at 6.1 times the live load, and the structural collapse at 11.9 times the live load. The global safety factor is hence well beyond the values proposed by current regulations [4].
6.
Conclusions
The paper presents a possible upgrading technique for timber frame buildings with platform frame construction, based on the use of TCCs. Experimental studies, both on full scale in-situ structures and laboratory test specimens have enable the characterisation of this type of structural application. Tests performed on materials, laboratory test specimens and real structures demonstrated that: - the lightweight concrete used had a lower Young’s modulus (compared to NW concrete), large shrinkage and creep, but also favourable lower density and high strength; - the connection is characterized by a good stiffness and strength. Rheological phenomena are not negligible being a little larger than those measured on timber; - the larger shrinkage due to the use of lightweight concrete produces important deflections during the first days after the pouring; - the proposed upgrading technique allows satisfaction for the ultimate limit state in the case of live load for office building. The collapse loads are always more than six times the design load; - the connection system allows a high effectiveness of the composite behaviour, which is close to that of a fully composite structure; - the real connection characteristics for joist significantly affects the collapse modality. Since the results of the tests are very promising, especially in terms of collapse load and stiffness, a reduction of the connector number may be considered in order to simplify the construction and reduce the costs.
7.
References
[1]
Ceccotti A., Timber Engineering - Step 2, Centrum Hout, The Netherlands, 1995, pp. E13/1E13/12. Grantham R., Enjily V., et al, Multi-storey timber frame buildings, BR454, BRE, UK, 2003. Comité Euro-International du Béton, CEB Bull. N°213/214, Lausanne, Switzerland, 1993. British Standards Instit., Structural use of timber – Part 2, BS 5268-2, BSI, London, 1996. Toratti T., “Creep of timber beams in a variable environment”, Report 31, Helsinki Univ., 1992. CEN, General principles for the determination of strength and deformation characteristics, EN 26891, Bruxelles, 1991. Kenel A., Meierhofer U., Holz/Beton-Verbund unter langfristiger Beanspruchung, Forschungs-und Arbeitsbericht 115/39 EMPA, Abteilung Holz, Dübendorf, 1998 (in German). Amadio C., Fragiacomo M., Ceccotti A., and Di Marco R., “Long-term behaviour of a timberconcrete connection system”, RILEM Conference, Stuttgart, 12-14 Sept. 2001, pp. 263-272. CEN, General principles for static load testing, EN 380, Bruxelles, 1993.
[2] [3] [4] [5] [6] [7] [8] [9]
Graduated in Civil & Structural Engineering in 1991. Worked in Timber Engineering at BRE since 1996. Currently lead a team of 5 conducting R&D projects on timber engineering, building science and environmental impacts of timber products. Born in Trieste (Italy) in 1967. Graduated in Civil Engineering in 1992. Ph.D. in structural engineering in 2001. Research Engineer since 1999. Author of about 50 papers focused on timber-concrete and steelconcrete composite beams, FE modelling, seismic engineering.
Summary Many modern timber framed buildings in the UK are constructed using platform frame techniques with the addition of floating floors for acoustic separation. This type of construction, commonly used for domestic multi-occupancy buildings, may be upgraded for reuse as an office building without the need for demolition of the existing building and replacement with a new steel or concrete framed structure. To achieve this, the authors propose replacement of the acoustic floor with a 60 mm lightweight concrete slab, connected to the existing joists with shear connectors to form Timber-Concrete Composite beams (TCC’s). This scenario has been tested on the TF2000 building at BRE Cardington and suggests that structural upgrading using TCC’s can achieve the required stiffness and up to six times the design load for satisfying design limit states for office loading in the UK. Keywords: collapse test, composite structures, creep, framed buildings, lightweight concrete, longterm behaviour, mechano-sorptive effect, rheological phenomena, shrinkage, timber.
1.
Introduction
Timber-concrete composite beam is a recent technique much used for strength and stiffness upgrading in both existing and new structures. It consists of connecting an existing or new timber beam with a concrete slab cast above a timber deck by means of a connection system [1]. The timber deck can then be employed as a formwork for the slab which is nominally reinforced with a simple anti-crack steel mesh. The slab should be cast after having propped the structure, to decrease stresses in the timber due to the structural dead load before concrete hardening. Experimental results demonstrate that the proposed upgrade technique is suitable for reuse of domestic buildings for office buildings. Both serviceability and ultimate limit states are fully satisfied. Thanks to the concrete slab, the floor is characterized by a larger stiffness, which reduces deflections and vibrations. The influence of the real restraint condition for the floor has also been
investigated. It is highlighted that the type of collapse is quite different with respect to the case of ideal hinge support.
2.
Test Set-up
125 476.5 600 600 600 600 600 600
An existing timber floor, with timber joists and Oriented Strand Board (OSB) decking, was converted to a TCC floor on the experimental TF2000 building [2] at BRE Cardington, UK. Two floor areas (A1 and A2 in Fig. 1) were upgraded using SFS VB-48-7.5x100 mm connectors in predrilled holes in the 38x225 mm timber joists, at the spacings and inclinations shown in Fig. 2. A plastic membrane laid on the existing 15 mm thick OSB decking provided the formwork for casting a 60 mm thick lightweight concrete slab. Nominal reinforcement for anti-crack purposes was provided by a 200x200 mm I6 steel mesh located 20 mm from the decking. Prior to casting the slab, the A2 floor was propped at one-third and two-thirds of the span along each joist. The A3 floor was left as original without casting a concrete slab. For each floor a reference joist was chosen and monitored in time starting from the concrete casting in order to evaluate the influence of rheological phenomena during the construction and loading history.
A1
A3
(Unpropped floor)
(Timber floor)
(Propped floor)
REFERENCE JOIST: J1
REFER. JOIST: J3
REFERENCE JOIST: J2
3440
2934
A2
3644
45 50 77 45 45
1720 585
765
108
55
No.4 connectors Inclination: 90°
No.14 connectors Inclination: 45° Spacing: 45 mm
No.9 connectors Inclination: 45° Spacing: 85 mm
Fig. 1 Plan view of the tested floors (dimensions Fig. 2 Connector arrangement for the in mm) unpropped floor (dimensions in mm)
3.
Material Test
Some laboratory tests were performed at BRE to fully characterise the component materials of the composite floors: lightweight concrete, timber and connection system. 3.1
Lightweight Concrete
A novel lightweight (LW) concrete was adopted for the slab incorporating re-processed sewage sludge for the course aggregate, similar to Lytag. The mix design ensured a strength class of LC 25/30 although standard compression tests on 100 mm cubes suggested a higher average strength of 46.7 MPa. The concrete fresh density was measured as 2020 kg/m3, its dry density as 1760 kg/m3 and the slump measured prior to pouring was 120mm. The medium values were evaluated for the following using two specimen results: Young’s modulus Elcm , cylindrical compression strength f lcm , and flexural tensile strength f lctm . The Young’s modulus of Lytag concrete was found to be about 60% that of a normal weight (NW) concrete with the same strength. Creep and shrinkage were measured on 100 mm by 200 mm diameter cylinders. Results are displayed in Fig. 3 for shrinkage strain and in Fig. 4 for the creep function at different times of loading t 0 . Dashed lines refer to predictions using Model Code 90 [3] for NW concrete. The shrinkage strain for Lytag concrete is much larger than for NW concrete (about four times). The creep function J (t , t 0 ) , is also larger compared to NW concrete. The creep coefficient for Lytag concrete ranges from 1 to 2.5 times that for NW concrete. Although further investigation is required, these tests highlight the larger sensitivity of Lytag LW concrete to rheological phenomena compared to NW concrete.
1000
200
Hcs(t)x106 [-]
800
J(t,t0)x106 [MPa-1]
160 t0=2
600
120 Exper. CEB 90
400
80
200
t [days]
0 0
20
40
60
80
t0=6
0
100
Fig. 3 Trend in time of shrinkage strain for Lytag lightweight concrete 3.2
Exper. CEB 90
40 t0=21 t0=35
0
20
40
t0=54 60
t [days] 80
100
Fig. 4 Trend in time of creep functions for Lytag concrete at different times of loading
Timber
The joists of the composite floor are classified as C16 European redwood kiln dried timber, according to BS EN 5268-2 [4]. Flexural strength f m and Young’s modulus E w were measured through four-point bending tests on eight 15x20x400 mm clear timber specimens cut from a joist in a region with few defects. Thermal expansion coefficients were obtained from seven specimens (5x5x210 mm) subject to relative humidity variations in a climate chamber and agreed closely with the value proposed by Toratti [5]. 3.3
Connection system
The connection system was tested at collapse in order to evaluate shear stiffness, shear strength and shear-slip relationship. The collapse test was performed according to EN 26891 [6] on three pushout specimens. The experimental shear-slip curves are displayed in Fig. 5 for one SFS VB-487.5x100 mm connector. A large stiffness can be realised up to 75% of the collapse load, beyond which an increase in plastic strains occurs up to the collapse, due to bond failure at the connectortimber interface. 16
1
V [kN]
M t [-]
0.8
12
Conn.-exp. Conn.-Kenel Timber-Toratti
0.6 8
0.4 4
0.2
slip [mm]
0 0
2
4
6
8
t [days]
0 10
0.01
0.1
1
10
100
1000
Fig. 5 Experimental shear vs. slip curves for Fig. 6 Creep coefficient for connection system connection system To investigate the rheological behaviour of the connection system, two push-out specimens were loaded with 25% of the collapse load applied through a lever mechanism. Such a load corresponds to the shear force that may be produced by the quasi-permanent load combination in a composite beam. The creep coefficients M are displayed in Fig. 6 for constant environmental conditions (70% RH, 24 °C). The results are in good agreement with those obtained by Kenel and Meierhofer [7] on the same type of connector differently arranged. The creep coefficient is about 60% larger
1003
compared to that of timber, as proposed by Toratti [5]. This confirms the importance of taking into account also the rheological behaviour of the connection [8].
4.
Long-term test
The monitoring of the full-scale floors began on the 6th of June when the concrete was poured on both propped (A2) and unpropped (A1) floors. For each reference joist the following quantities were monitored: vertical displacements at 5 locations through Linear Voltage Displacement Transducers (LVDT’s), strains of timber for the bottom fibre at mid-span through Demec gauges and Demec points, temperature of timber joist and concrete slab using thermocouples, moisture content of timber using electrical moisture probes, environmental temperature and relative humidity using a digital thermohygrometer. The purpose of monitoring the construction, before the live load application, was: to compare the floor behaviour for different types of construction (propped and unpropped), and to investigate the influence of rheological phenomena (creep and shrinkage) of materials when the concrete is still green, and hence more sensitive to these effects. After removal of the props, 14 days after the concrete casting, the live load of 2.5 kN/m2 required for office buildings was applied, using water pools for dead loading. A comparison between the propped (A2) and unpropped (A1) floor is displayed in Fig. 7 and Fig. 8 for both the mid-span deflection, v and bottom fibre timber strain, H . Comparing the two cases, propped and unproped, indicates a better behaviour for the propped floor with respect to the unpropped one ( v 5.3 mm instead of v 6.5 mm, and H 0.7 mm/m instead of H 1.4 mm/m immediately after the live load application). This agrees with recommendations given by many authors [1] for propping the composite floor to reduce deflections, however the reduction in deflection, for this case, was pretty small warranting use of the unpropped technique, which leads to cost savings for construction. The dead load due to the weight of the concrete slab and the rheological phenomena produced, during the first 34 days of testing, a mid-span deflection v p 3.9 mm ( v p l 1 934 , l being the span length) and v u 5.0 mm ( v u l 1 688 ) for propped and unpropped floors, respectively. The live load application produced an increase in deflection of 'v p 1.4 mm ( 'v p l 1 2603 ) and 'v u 1.5 mm ( 'v u l 1 2293 ) for the composite floors. The limitation v l d 1 333 [4] is met for all structures, however it is expected that the rheological phenomena will increase this ratio over longer periods. 8
1.5
v [mm]
Unpropped Propped
6
H [mm/m]
1
4
Unpropped Propped
0.5 Live load application
Props removal
2
t [days]
0 0
10
20
30
Props removal
0
40
Live load application
t [days]
-0.5 0
10
20
30
40
Fig. 7: Trend in time of mid-span deflection for Fig. 8: Trend in time of the timber bottom fibre the composite floors strain at mid-span for the composite floors
5.
Collapse test
Collapse tests were performed on the unpropped composite floor (A1 in Fig. 1) to evaluate the performance of the upgrade technique in terms of both stiffness and ultimate load and to estimate any contribution offered by the real support conditions for joists ‘clamped’ by wall panels (see Fig. 2). For the investigation, the composite floor was cut making two independent portions for test T1 and test T2; see Fig. 9. This eliminated unwanted restraint effects, allowing easier comparison of
experimental results. For test T1, the slab was completely disconnected from the building and supported under joist ends, on half rounds (Fig. 9). In this way, support conditions for test T1 were similar to a laboratory test assuming perfect hinges and no rotational restraint at supports. Test T2 was instead restrained at joist ends with the real joint, allowing assessment of the rotational restraint by comparison of the two tests. Concrete cut
Load rig
Test T1
Joist B1
146 300
Floor cut
Floor cut
600
Joist B2
300
Joist B1
300
Concrete cut
Test T2
600
Joist B2
Timber joist
Load rig
600
Joist B3
600
Joist B4
300
Concrete cut 1120
1200
330 1120
Fig. 9 Set-up of the collapse tests
Fig. 10 Test T1: Load rig
Fig. 11 Test T2: Load rigs
The loading rigs are depicted in Fig. 9, and pictured in Figs. 10 and 11 for the tests T1 and T2, respectively. For each test, the following quantities were monitored: vertical displacements, relative slip between concrete and timber joists near the supports, longitudinal strains at top and bottom fibres of concrete slab and timber joists. Load was applied according the schedule in EN 380 [9] and results are plotted, respectively, in Figs. 12 and 13 for Test T1 and T2 in terms of concentrate load on each joist vs. mid-span deflection. The curves concerning the case of rigid connection (fully composite behaviour) and no connection (non composite behaviour) between concrete and timber are drawn for comparison. 25
P [kN]
25
20
Joist B1
15 10
Non-composite
5
10
15
20
25
Joist B2 Non-comp.
10
Only timber
5
v [mm] 5
Joist B3 Joist B1
15
Only timber
0
Joist B4
Fully composite
20
Joist B2
0
P [kN]
Fully composite
v [mm]
0
30
0
10
20
30
40
Fig. 12 Test T1: Load vs. mid-span deflection Fig. 13 Test T2: Load vs. mid-span deflection curves curves Both tests demonstrate behaviour close to that of a fully composite floor during the first loadingunloading cycle. The effectiveness of the connection system is very good, with small vertical deflections and high stiffness. Small plastic strains occurred during the cycle, especially for test T1. Collapse was observed when timber joist B1 failed in flexure at a load P 18.3 kN ( q 22.0 kN/m2). Test T2 demonstrated a rather different behaviour. Longitudinal cracks appeared on the timber joists near the supports at a load P 11.6 kN ( q 15.3 kN/m2). These cracks, coincident with the line joining the lower connector ends (see Fig. 2), suddenly spread towards the
1005
mid-span of joists B1 and B3 at a load P 12.5 kN ( q 16.5 kN/m2). Although the serviceability limit state was compromised, the ultimate limit state was not reached until P 22.5 kN ( q 29.7 kN/m2). The increased flexibility, following longitudinal splitting (Fig. 13) changed the behaviour to almost that of a non-composite section. The comparison between tests T1 and T2 highlights that: - the presence of the real panel-to-joist joint worsens the behaviour of the composite floor with respect to the case of a simply supported joist, leading to an early longitudinal cracking of timber joists with loss of stiffness; - the performance of both tests was satisfactory: the collapse occurred for the test T1 under an equivalent uniformly distributed load q that was 8.8 times larger than the live load for office buildings. In the test T2, corresponding to the real floor behaviour, longitudinal timber cracking occurred at 6.1 times the live load, and the structural collapse at 11.9 times the live load. The global safety factor is hence well beyond the values proposed by current regulations [4].
6.
Conclusions
The paper presents a possible upgrading technique for timber frame buildings with platform frame construction, based on the use of TCCs. Experimental studies, both on full scale in-situ structures and laboratory test specimens have enable the characterisation of this type of structural application. Tests performed on materials, laboratory test specimens and real structures demonstrated that: - the lightweight concrete used had a lower Young’s modulus (compared to NW concrete), large shrinkage and creep, but also favourable lower density and high strength; - the connection is characterized by a good stiffness and strength. Rheological phenomena are not negligible being a little larger than those measured on timber; - the larger shrinkage due to the use of lightweight concrete produces important deflections during the first days after the pouring; - the proposed upgrading technique allows satisfaction for the ultimate limit state in the case of live load for office building. The collapse loads are always more than six times the design load; - the connection system allows a high effectiveness of the composite behaviour, which is close to that of a fully composite structure; - the real connection characteristics for joist significantly affects the collapse modality. Since the results of the tests are very promising, especially in terms of collapse load and stiffness, a reduction of the connector number may be considered in order to simplify the construction and reduce the costs.
7.
References
[1]
Ceccotti A., Timber Engineering - Step 2, Centrum Hout, The Netherlands, 1995, pp. E13/1E13/12. Grantham R., Enjily V., et al, Multi-storey timber frame buildings, BR454, BRE, UK, 2003. Comité Euro-International du Béton, CEB Bull. N°213/214, Lausanne, Switzerland, 1993. British Standards Instit., Structural use of timber – Part 2, BS 5268-2, BSI, London, 1996. Toratti T., “Creep of timber beams in a variable environment”, Report 31, Helsinki Univ., 1992. CEN, General principles for the determination of strength and deformation characteristics, EN 26891, Bruxelles, 1991. Kenel A., Meierhofer U., Holz/Beton-Verbund unter langfristiger Beanspruchung, Forschungs-und Arbeitsbericht 115/39 EMPA, Abteilung Holz, Dübendorf, 1998 (in German). Amadio C., Fragiacomo M., Ceccotti A., and Di Marco R., “Long-term behaviour of a timberconcrete connection system”, RILEM Conference, Stuttgart, 12-14 Sept. 2001, pp. 263-272. CEN, General principles for static load testing, EN 380, Bruxelles, 1993.
[2] [3] [4] [5] [6] [7] [8] [9]
