COMPARISON OF STRAIN HARDENING BEHAVIOUR OF ......TRADITIONAL REINFORCED CONCRETE BEAMS C. A. Muir 1, D. K. Bull2, E. Au 3 and S. Pampanin 4 ABSTRACT 1234 Research conducted during

COMPARISON OF STRAIN HARDENING BEHAVIOUR OF NON-TEARING AND TRADITIONAL REINFORCED CONCRETE BEAMS

C. A. Muir1, D. K. Bull2, E. Au3 and S. Pampanin4

ABSTRACT1234

Research conducted during an undergraduate project is presented and placed in the context of a larger research program. The study analytically investigates the effect that increased plasticity in the bottom longitudinal reinforcement has on the overstrength development in slotted beams. Emphasis is placed on comparison with traditional monolithic reinforced concrete beams. Two 16 storey moment resisting frames formed the basis of this investigation with one being a traditional monolithic system and the other implementing slotted beam connections. A multispring representation of a slotted beam was created to more accurately model the joint subassembly response. Non-linear time history analyses were performed on these frames using a suite of earthquake records for the Wellington region. Interstorey drift was observed to be larger in the slotted structure due to the reduced stiffness of the system. The slotted beam-column connections investigated exhibited comparable rates of strain hardening to equivalent monolithic connections under realistic seismic excitation. This is due primarily to large strain penetration experienced in the slotted connection when using large diameter bars necessitated by large moment demand. Characteristic overstrength factors of 1.22 and 1.25 were obtained for Grade 300 reinforcement for the slotted beam subassemblies tested. The on-going doctoral research into the behaviour and design of slotted beams by the first author under the supervision of the second and forth authors is introduced.

1 PhD Candidate, University of Canterbury, Christchurch, New Zealand. 2 Professor, University of Canterbury, Christchurch, New Zealand. Technical Director, Holmes Consulting Group Limited, Christchurch, New Zealand. 3 Engineer, Structex Limited, Christchurch, New Zealand. 4 Associate Professor, University of Canterbury, Christchurch, New Zealand.

INTRODUCTION

Background Seismic design requirements were introduced to New Zealand in 1935. Since this time and throughout subsequent revision of these requirements, reinforced concrete has remained a popular construction material [1]. The use of precast concrete construction gained popularity in the 1960’s through the introduction of precast flooring systems. The 1980’s saw a rapid increase in the use of precast elements as part of the primary lateral load resisting system. Extensive experimental and historical earthquake data has shown that well detailed monolithic reinforced concrete structures perform well at the ultimate limit state. Given the comparative lack of data on precast concrete, construction methods have evolved to primarily involve the joining together precast elements to achieve comparable levels of performance to an equivalent monolithic system. However during recent seismic events traditional monolithic reinforced concrete structures have had to be demolished due to prohibitive repairing costs. The main contributors to this cost are residual drift and structural damage. Residual drift occurs when the building does not return to its original, plumb, position following earthquake attack. The building will need to be righted because it can not only impair the seismic performance of the structure during subsequent events but also impose severe serviceability issues. Structural damage within a monolithic reinforced concrete moment resisting frame primarily stems from plastic hinge zones. These zones are chosen, and suitably detailed, to undergo large inelastic deformation which limits the amount of force a structure must be designed for. In a traditional monolithic structure the energy is dissipated through alternative tension and compressive yielding of the top and bottom reinforcement over cyclic reversals. This mechanism can result in there being a difference in the neutral axis depth at each beam end. This causes a geometric contribution to beam elongation. Shear transfer through a plastic hinge zone is by way of an equivalent truss mechanism. The horizontal component of the diagonal shear strut means that tensile forces and hence strains tend to be larger than compressive. These accumulating tensile strains cause a material contribution to beam elongation. The net effect of combined geometric and material beam elongation is the potential to cause significant damage to the structure through the formation of undesirable inelastic mechanisms in neighbouring columns and tearing of the floor diaphragm. Floor diaphragm damage has been shown to inhibit lateral force transfer and in extreme cases cause complete loss of the gravity carrying capacity leading to total collapse [2, 3]. An example of this sort of collapse was observed at the Meadows Apartment car park structure following the 1994 Northridge earthquake, as shown in Figure 1.

Figure 1: Total collapse of hollow-core flooring unit following 1994 Northridge Earthquake [4].

It follows that these deficiencies with current reinforced concrete design need to be rectified. Efforts to date have primarily focussed on developing low damage connections using dry jointed ductile connections. Whilst these systems address connection damage they neglect floor and column damage caused by beam elongation.

Slotted Beam Development A solution that addresses these issues is the slotted beam. The slotted beam concept was first proposed during the PREcast Seismic Structural Systems (PRESSS) research program [5] in the form of the UT-GAP connection, shown in Figure 2. This connection provided a 1” slot extending ¾ down the column face. This meant rotation occurred about the bottom of the beam, concentrating plasticity in the top longitudinal reinforcement. The reaction force to form a moment couple was provided by vertical rods extending from the shear corbel. The connection performed satisfactorily and was developed throughout the PRESSS program to include post-tensioning to provide a more direct force transfer. Whilst this connection reduces beam elongation, it is likely that damage to the floor diaphragm would result due to the gap opening at the top of the beam. It was noted during the PRESSS program that inverting the connection would resolve this issue [6].

Figure 2: UT-GAP connection tested during PRESSS program [6].

The slotted beam solution for cast-in-situ reinforced concrete frames was proposed at a similar time in Japan. An early design of the reinforced concrete slotted beam is shown in Figure 3. A slot extending approximately ¾ up the column face is provided which constrains rotation to occur about the top concrete hinge. The top longitudinal reinforcement is stronger than the bottom to limit strain and hence cracks. The plasticity in the connection is limited to the bottom longitudinal reinforcement and debonding is provided to limit plastic strain. Shear transfer is provided by diagonal hangers anchored into the columns.

Figure 3: Early design of reinforced concrete slotted beam [7].

The rotation occurring about the top concrete hinge at either end of the beam drastically reduces the geometrical contribution to beam elongation. Similarly, the reduction in cracking and recovery of tensile strains reduces the material contribution to beam elongation. As a result, the beam elongation is approximately 10% that of an equivalent monolithic beam [8, 9]. This reduces the aforementioned undesirable effects that can occur as a result of beam elongation. The first paper on reinforced concrete slotted beams published in English was in 1999 [7]. This investigated ways of improving the shear transfer mechanism. Early tests showed that due to the debonded bottom longitudinal reinforcement the equivalent truss shear transfer mechanism required concrete tension to be complete. When the tensile capacity of the concrete was exceeded large shear cracks occurred. Specimens RCSB3 and RCSB4, shown in Figure 4, tested additional reinforcement designed to prevent these cracks and improve shear transfer. These details performed well.

(a) Details of specimen RCSB3. (b) Details of specimen RCSB4.

Figure 4: Different shear transfer details investigated by Ohkubo et. al. [7].

Subsequent research by Ohkubo and Hamamoto [9] tested cruciform subassemblies with cast-in-situ floor slabs. A benchmark monolithic connection was compared to a slotted to determine the differences in damage. The slotted beam displayed a stable response and showed significantly less damage to the frame and floor. This is shown in Figure 5.

Figure 5: Damage to frame and in-situ floor for slotted (left) and monolithic (right) connection details [9].

The research on slotted beams at the University of Canterbury has built on the original proposal presented in the PRESSS program for precast concrete systems as well as on the progress made by Japanese researchers on cast-in-situ systems. Recent research by Au [8] and Leslie [10, 11] has concentrated on investigating the slotted beam parametrically, analytically and experimentally. These investigations were conducted in parallel and were slightly different in their focus. Research by Leslie [10, 11] focussed on developing and experimentally validating a variety of precast or semi-precast concrete slotted beam details for rapid implementation into New Zealand practice. Different configurations of the slotted beam were investigated, including top hinge material and the use of high strength bottom longitudinal reinforcement. This research showed that fabricated structural steel top hinges could deform excessively in shear, hastening buckling of the bottom longitudinal reinforcement. It was found that high strength lower longitudinal reinforcement could reduce residual drifts. Activation of the floor slab reinforcement in tension and associated contribution to beam overstrength in negative flexure was minimal. Research by Au [8] focussed on developing and testing an appropriate detail for a slotted reinforced concrete connection that can be used as a substitute for conventional monolithic reinforced concrete connections in New Zealand. An extensive theoretical investigation into slotted beam mechanics produced new design recommendations. Four specimens were tested experimentally, one benchmark monolithic (RCB1) and three slotted (SB1-3). Specimen SB1 performed well up to 3.5% drift upon which failure by bar buckling occurred. Specimen SB2 failed by bond slip of the bottom longitudinal reinforcement through the interior column at the 2.5% drift cycle. Damage observed in RCB1 was significantly greater than SB1 and SB2. The slotted beam connections displayed greater hysteretic energy dissipation and the beam elongation was, as anticipated, approximately one tenth of the monolithic. Specimen SB3, shown in Figure 6,

was similar to SB2. However, it had a precast floor slab and incorporated improvements in design following reduction and analysis of SB1 and SB2 test results. Such as, improved; buckling restraint, beam torsion restraint, joint shear reinforcement and bottom longitudinal beam reinforcement bond requirements.

(a) Front elevation. (b) Side elevation. Figure 6: Specimen SB3 tested by Au [8].

The performance of specimen SB3 was extremely satisfactory with a stable and fat hysteresis to 4.5% beam drift upon which failure occurred by bar fracture, likely due to low cycle fatigue. Experimental and analytical investigations have shown that the increased plasticity in the bottom longitudinal reinforcement in slotted beams causes increased cyclic strain hardening and subsequently increased connection overstrength [8]. This programme investigated numerically the effect that the increased plasticity in the bottom longitudinal reinforcement has on characteristic overstrength factors when subject to realistic seismic excitation. Direct comparison with traditional monolithic reinforced concrete beams was undertaken to provide a performance benchmark. RESEARCH PROGRAM

The basis of this numerical investigation was two 16 storey reinforced concrete perimeter moment resiting frames. One structure used slotted beams and the other traditional monolithic beams. The generic structure is shown in Figure 7. These three by three bay office buildings are situated in Wellington on soil class C and designed for a 50 years life. The structures were designed to not be affected by near fault effects. Because the effect of cyclic strain hardening is being assessed, far field seismic records with a long periods and typically longer durations are considered in this case the worst case scenario. The design and analysis of the structures was undertaken using the equivalent static method, modal response spectrum and displacement based design. The base shear determined using modal response and equivalent static were in good agreement. The displacement based method yielded values slightly lower. The more conservative force based values were designed for. For analysis, it was assumed that both the slotted beams and traditional monolithic beams had similar stiffness characteristics and hence effective stiffness factors outlined in NZS3101:2006 [12] could be used in both structures. This has been shown by experiment to be realistic [8] and simplified analysis for design to one generic model.

Figure 7: 16 storey prototype structure with subassembly location highlighted.

Both structures were detailed according to NZS3101:2006 [12]. However, in the case of the slotted structure some design outside of this Standard was undertaken based on recent research [8, 10, 11]. Section capacities were decreased up the height of the structures according to expected demand determined from force based analysis to promote the formation of a favourable inelastic mechanism. The analysis of monolithic reinforced concrete sections was undertaken using traditional flexural techniques based on Euler-Bernoulli theory. However the debonding of the bottom longitudinal reinforcement in slotted beams and the presence of a physical gap means that the key assumption of this technique is no longer applicable through the whole depth of the section. Analysis was thus instead performed using moment rotation techniques. Three variations on beam depth in both the slotted and monolithic structures were examined. Beam depths of 800, 900 and 1000mm were designed. This was to more fully examine the effect of cyclic strain hardening and to increase the general applicability of the results. This variation meant that six structures were designed in total. In order to undertake detailed analysis and compare hysteretic response between the different structure designs a common subassemblage was chosen. The forth storey exterior beam-column joint, shown in Figure 7, was chosen because preliminary analysis showed this to be the location of peak beam rotation demand. EARTHQUAKE SELECTION AND SCALING

The choice and scaling of earthquake records for use in numerical integration time history analysis is fundamental to the applicability of the results generated. Preliminary recommendations of appropriate records to be used for time history analyses have been made by the Institute of Geological and Nuclear Sciences. The selected records for the Wellington region, on soil class C are shown in Table 1. In line with NZS1170.5:2004 [13] a minimum of three earthquake records were chosen. All records chosen were far-field. Only the principle earthquake directions were applied.

Table 1: Earthquake records applied in time history analyses.

Record Name Date Magnitude Distance (km)

Depth (km)

Duration (s) Mechanism Near

Fault Scaled PGA (g)

El Centro, Imperial Valley, USA 19-May-1940 7.0 6 10 63.48 Strike-Slip No 0.613

La Union, Mexico 19-Sept-1985 8.1 16 15 72.94 Subduction No 0.477 HKD085 Hokkaido,

Japan 26-Sept-2003 8.3 46 33 78.00 Subduction No 0.611 Scaling of earthquake records was performed according to NZS1170.5:2004 [13]. The prototype structures were designed using force based design to achieve 2.5% interstorey drift under a design level earthquake. It was found that the earthquake records scaled according the New Zealand Standard [13] were not sufficiently strong to reach target deformations. To be able to fully access the primary investigation objective, significant deformation was required at the beam ends to induce overstrength development. To achieve this, the earthquake records were scaled to induce design level interstorey drifts. The scaled peak ground accelerations of the records applied to the models are shown in Table 1. ANALYTICAL MODELLING

All analysis was undertaken using the non-linear time history analysis software Ruaumoko2D [14]. Monolithic Frame The monolithic frame was assembled from beam-column reinforced concrete frame linear elements, with plasticity concentrated at both ends. All beams and the ground floor columns were specified non-linear with a modified Takeda hysteresis rule. Rigid end blocks were used to represent the beam-column joint panel zone. The elements were calibrated using axial-moment interaction data determined through sectional analysis using Response-2000 [15]. Effective stiffness’ were used in the model according to NZS3101:2006 [12]. Dummy pin-ended columns were specified to represent the P-delta effects of the interior gravity bays. Slotted Frame The slotted beam frame was assembled from a combination of rotational spring and beam-column reinforced concrete elements. This model is termed rotational spring model in the following. Only the column bases and rotational springs were specified inelastic. The column base properties, rigid end blocks and P-delta dummy column were identical to that used in the monolithic structure. The slotted beam behaviour was represented by a rotational spring assigned a Dodd-Restrepo hysteresis rule. Because the plasticity in a slotted connection is constrained to the bottom longitudinal reinforcement a steel hysteresis rule was used to model the connection behaviour. Calibration was undertaken using the results of a moment-rotation analysis via spreadsheet [16]. This allowed unbonded length, strain penetration and moment contributions from the hanger to be accounted for. Multispring Model The multispring model was constructed with steel and concrete spring elements arranged to reflect the physical geometry of the section, as shown in Figure 8. This model draws on concepts developed by Peng [17] and Au [8]. The top, bottom and shear springs were assigned Dodd-Restrepo hysteresis rules. These were calibrated using monotonic test data for reinforcement [16].

Figure 8: Subassembly multispring model.

The concrete hinge was modelled using 6 concrete springs, distributed on the basis of a Lobatto integration [18]. This enabled the provision of greater local stiffness in the centre of the hinge than near the edges resulting from decreasing confinement. The concrete springs were assigned the Peng hysteresis rule with default variables. Shear was resisted by a steel shear spring, analogous to the diagonal shear reinforcement. NUMERICAL ANALYSIS PROCEDURE

The scaled earthquake records were applied to both the monolithic and slotted frame systems for all beam depths and non-linear time history analyses run. The multispring was not used directly in the slotted beam frame structures due to computational cost and on-going element numerical stability issues. A multispring model was instead used to more accurately represent the slotted beam behaviour at a subassembly level. Using a hybrid approach, the rotational history of the slotted beam subassembly recorded in the frame response during the earthquake records was extracted and formed the loading protocol for the analysis of the multispring model subassembly. The rotation was applied at one end of the element whilst the other was fixed. Validation of the simplified rotational spring model when compared with a more refined multispring model could thus be performed. Following thus validation, comparison between the monolithic and slotted multispring subassembly responses could be undertaken to examine differences in response. A schematic of this overall procedure is presented in Figure 9. In total 27 time history analyses were performed.

Figure 9: Schematic of experimental procedure.

TIME HISTORY ANALYSIS RESULTS

For the sake of brevity, only the results for the La Union earthquake record for structures with 900mm deep beams are herein presented. These results are considered representative. The overall match between the lumped plasticity rotational spring and multispring model was good, as shown in Figure 10 (a). The overstrength development in the lumped plasticity model is larger than that developed in the multispring. This is due to constraining the behaviour of the slotted beam to model reinforcing steel hysteresis. This rule cannot take into account the subtleties of slotted beam behaviour that the multispring can, such as neutral axis variation, shear interaction and concrete cracking. The match between stiffness of the lumped plasticity and multispring models was close. This is essential to obtaining representative frame response due to the critical role of element stiffness in global deformation.

(a) Hysteretic response of rotational spring and

multispring models for La Union record. (b) Interstorey drifts between forth and fifth stories for La

Union record.

(c) Hysteretic response of monolithic and slotted subassemblies during La

Union record.

Figure 10: Results for structures with 900mm deep beams during La Union record.

The slotted beam structure in general displayed larger peak and residual interstorey drifts than the traditional monolithic beam structure. This can be seen in Figure 10 (b). It is important to note that the larger peak drift occurs only in one direction. This suggests a permanent offset. This is undesirable as residual drifts can significantly increase building rehabilitation costs. The monolithic subassembly displays a stiffer elastic response than the slotted, as shown in Figure 10 (c). However, it should be noted that recent experiment [8] has shown that although this is generally true, the difference is small. In this case the relatively large difference can be attributed to the significant amount of strain penetration assumed in the bottom longitudinal reinforcement. This is a result of the large diameter bottom longitudinal reinforcement necessitated by the large moment demand. Strain penetration was assumed to be 0.022fydb [19]. Also evident in Figure 10 (c) is that the

reloading stiffness of the slotted beam is larger than the monolithic. The lower elastic stiffness of the slotted beam structure means that it is more easily deformed initially. The lower reloading stiffness of the monolithic structure means that deformations can be more easily recovered over successive cycles. The cumulative effect of these factors is that the slotted structure suffers larger permanent and peak drifts in that direction than the monolithic for the earthquake records examined. The slotted and monolithic subassemblies display comparable rates of strain hardening. The absence of more significant cyclic strain hardening can be attributed to insufficient strain being generated in the lower longitudinal reinforcement. This is due to the unbonded length and large amount of strain penetration. Furthermore the deformation favouring one direction due to permanent set could have limited the ability to achieve larger cyclic plastic strains and associated material enhancement. Characteristic 95th percentile overstrength factors of 1.22 and 1.25 were computed for positive and negative flexure respectively. This is within the code specified overstrength factors for design of monolithic beams with Grade 300 reinforcement of 1.25 [12]. However it has been shown experimentally that the overstrength in slotted beams can be as high as 1.3 for Grade 300 reinforcement [8]. The amount of post yield cyclic strain hardening depends heavily on the strain history. It is acknowledged that due to the small pool of data and limitations of the models employed these results can be considered indicative only. MORE RECENT NUMERICAL ANALYSES

Recent numerical investigation by Au [8] has presented conflicting results. This research employed a similar methodology to compare the responses of two five storey structures, one using a slotted beam detail and the other traditional monolithic. These models are shown in Figure 11 (a) and (b). Where this investigation differs from that described above is that the multispring representation of the slotted connection is implemented directly in the prototype structure. Furthermore the multispring element was calibrated based on experimental data. Good agreement was obtained between the numerical model and experimental results of the slotted beam specimen SB1, as shown in Figure 11 (c). In addition, a more realistic representation of monolithic connection detail was used in the form of a multispring element developed by Peng [17].

(a) Slotted structure. (b) Monolithic structure.

(c) Comparison of experimental response and numerical modelling of slotted beam specimen SB1.

Figure 11: Five storey prototype frame structures investigated by Au [8].

This investigation found, similarly, that the monolithic connection was stiffer initially but degraded rapidly compared to the slotted connection. The main difference between the investigations was the energy dissipated. Au [8] found that the hysteretic energy dissipation of the slotted connection was greater than the monolithic. This removed energy from the system and resulted in slightly smaller peak and residual drifts in the slotted structure compared to the monolithic. This effect was not captured accurately by the research presented above. FURTHER DEVELOPMENTS AND PROPOSED RESEARCH

Whilst significant progress in the performance of the slotted beam has been achieved through the concerted efforts of researchers in Japan and subsequently at the University of Canterbury, there remain unsolved issues. Research is on-going at the University of Canterbury and involves the efforts of industry, doctoral and masters researchers. The contribution proposed by the first author as part of his doctoral research work can be broadly subcategorised into superassembly and, subassembly experimental tests and numerical investigations. Superassembly test A large scale three-dimensional test is required to provide data on the interaction between structural elements in a realistic New Zealand building geometry. It will serve to confirm whether the damage to the floor diaphragm is reduced, and to what extent, when the slotted beam detail is used. Global hysteretic response data will be used to calibrate subsequent numerical models. This will also serve to assess the practicalities and performance of the system as designed using current knowledge. The superassembly test will also serve as a showpiece to increase familiarity of the construction industry with this emerging detail. At the time of writing the detailed design of the superassembly is complete and documentation has been completed and issued for pricing. The prototype structure for this investigation is a seven storey moment resiting frame designed for the Wellington region, founded on soil class C and affected by near field effects. The structure is shown in Figure 12 (a). Analysis of the structure was conducted using modal response spectra and non-linear time history analysis.

(a) Prototype structure. (b) Superassembly extracted from prototype. Figure 12: Three-dimensional test specimen and origin.

The two storey, two by one bay, superassembly was extracted from the first and second stories of the prototype building and is shown in Figure 12 (b). The specimen is scaled geometrically at 2/3 according to a ‘practical real model’ philosophy [20]. The floors of both stories are prestressed precast concrete units. Design was according to NZS3101:2006 [12]. However, given that the Standard was not written with the intent of being applied to this type of structure, some aspects of the design were based on recommendations from recent research and first principles and as such lie outside of the current Standard [12]. The specimen has been designed to closely replicate how a typical structure in New Zealand would be designed and constructed using the slotted beam configuration.

The specimen will be founded on universal joints to represent the points of contraflexure in the columns. Loading will be biaxial through the frame ends at a constant force ratio of 2:1 between the top and bottom floors respectively. The protocol will be quasi-static following a cloverleaf displacement trace. The loading rig is based on a set-up prepared and used for large scale testing as part of a University of Canterbury research project on innovative timber structures. However a full redesign was necessary due to the force and geometry requirements. The forces required in this test are at the limit of what is achievable given the current laboratory facilities. Subassembly tests Based on the performance of the three-dimensional superassembly, a number of two-dimensional subassembly quasi-static cyclic tests will be undertaken to assess and develop changes to design recommendations. The performance of the superassembly will direct what, and how many, subassembly tests need to be undertaken. Regardless of how positive the superassembly results could be two-dimensional testing will be required to assess details that could not be incorporated into the superassembly test and to determine the statistical reliability of previous results. The subassemblies may include floor slabs if the three-dimensional test shows the diaphragm is important to the response of the element being investigated. Three-dimensional testing is to be undertaken before the two-dimensional in order to assess the performance of the system as it stands at present and highlight any deficiencies in a realistic specimen and loading scenario. Armed with the knowledge of the issues encountered and boundary conditions present in the superassembly, the subassembly tests can be used to verify new design details to rectify any undesirable behaviour in a representative manner. Numerical investigation Given the conflict between the two-dimensional time-history analyses undertaken to date and the need assess the ideal structural configuration for a slotted beam system the numerical analyses presented herein will be extended to three-dimensions. This will enable the overall structural response of structures incorporating slotted beams to be compared to current and emerging details. In this manner, the respective merits of each system can be compared. The performance of different structural configurations incorporating the slotted beam can be evaluated to determine the ideal system. CONCLUSIONS

The following conclusions can be drawn from the presented research; - Initial stiffness of slotted beams is less than the equivalent monolithic. However experiment

has shown the difference to be small [8]. - Strength degradation is greater in the monolithic beams when compared to an equivalent

slotted beam. - These two combined effects cause peak and residual interstorey drifts to be greater in

structures with slotted beams than structures with equivalent monolithic beams for the structural configurations and earthquake records studied.

- Overstrength factors of 1.22 and 1.25 were obtained for positive and negative flexure respectively for the structural configurations and earthquake records studied. The amount of post yield cyclic strain hardening depends on the strain history. Experiment has shown overstrength can be as high as 1.3 [8].

- More sophisticated modelling has shown that increased hysteretic energy dissipation by the slotted beam can cause slotted structures to display smaller peak and residual drifts when compared to an equivalent monolithic structure [8].

In response to limitations in the above research, and building on recently presented research [8, 10], doctoral research to be undertaken by the first author under the supervision of the second and forth authors was presented.

ACKNOWLEDGEMENTS

The first author would like to acknowledge the New Zealand Concrete Society for financial support to present this paper through the 2009 Concrete Prize. Also the New Zealand Foundation for Research, Science and Technology and the University of Canterbury for financial assistance to undertake the on-going research. REFERENCES

1. CAE, Guidelines for the use of structural precast concrete in buildings. Report of a study group of the NZ Concrete Society and the NZ National Society for Earthquake Engineering. 1999, Christchurch, New Zealand: Centre for Advanced Engineering.

2. Bull, D.K., Understanding the complexities of designing diaphragms in buildings for earthquakes. Symposium to celebrate the lifetime achievements of Tom Paulay and Bob Park, 2003. Christchurch.

3. Matthews, J.G., Hollow-core floor slab performance following a severe earthquake, in Department of Civil Engineering. 2004, University of Canterbury: Christchurch, New Zealand.

4. Norton, J.A., et al., Northridge Reconnaissance Report. Bulletin of the New Zealand National Society for Earthquake Engineering, 1994. 27(4).

5. Priestley, M.J.N., The PRESSS program - Current status and proposed plans for phase III. PCI Journal, 1996. 41(2): p. 22-40.

6. Palmieri, L., et al., Ductile connections for precast concrete frame systems. ACI, 1996. 162: p. 313-356.

7. Ohkubo, M., et al., Shear transfer mechanism of reinforced concrete beams with a slot at the beam-end. Proceedings of Japan Concrete Institute, 1999. 21(3): p. 301-306.

8. Au, E., The mechanics and design of a non-tearing floor connection using slotted reinforced concrete beams, in Department of Civil and Natural Resources Engineering. 2010, University of Canterbury: Christchurch, New Zealand.

9. Ohkubo, M. and T. Hamamoto, Developing reinforced concrete slotted beam structures to reduce earthquake damage and enhance seismic structural performance. Proceedings of the 13th annual World Conference on Earthquake Engineering. Vancouver., 2004.

10. Leslie, B.J., The development and validation of a non-tearing floor precast concrete structural system for seismic regions, in Department of Civil and Natural Resources Engineering. 2010, University of Canterbury: Christchurch, New Zealand.

11. Leslie, B.J., D.K. Bull, and S. Pampanin, The Seismic Performance of a Non-Tearing Floor Precast Concrete Structural System, in New Zealand Society for Earthquake Engineering Conference. 2010: Wellington, New Zealand.

12. Standard Association of New Zealand, Concrete Strucutures Standard, in New Zealand Standard NZS3101. 2006: Wellington, New Zealand.

13. Standard Association of New Zealand, Structural Design Actions, in New Zealand Standard NZS1170. 2004: Wellington, New Zealand.

14. Carr, A.J., Ruaumoko Manual Volume 2: User Manual for the 2:Dimensional Version Ruaumoko2D. 2005, Department of Civil and Natural Resources Engineering, University of Canterbury, Christchurch, New Zealand.

15. Bentz, E., Response-2000 User Manual. 2001, University of Toronto, Canada. 16. Au, E., Personal Communication. 2009. 17. Peng, B.H.H., Seismic performance assessment of reinforced concrete buildings with precast

concrete floor systems, in Department of Civil and Natural Resources Engineering. 2009, University of Canterbury: Christchurch, New Zealand. p. 495.

18. Spieth, H.A., et al., Modelling of Precast Prestressed Concrete Frame Structures with Rocking Beam-Column Connections. 2004: Research Report 2004-01, Department of Civil and Natural Resources Engineering, University of Canterbury, Christchurch, New Zealand.

19. Paulay, T. and M.J.N. Priestley, Seismic design of reinforced concrete and masonry buildings. 1992, New York: John Wiley & Sons, Inc.

20. Harris, H.G. and G.M. Sabnis, Structural modelling and experimental techniques. 2 ed. 1999: CRC Press, Boca Raton, London, New York and Washington D.C. 789.