Development of a cross laminated, post tensioned bridge deck

Transcription

1 Development of a cross laminated, post tensioned bridge deck Christophe Sigrist 1, Martin Lehmann 2 ABSTRACT: Cross Laminated Timber (CLT) finds many applications in housing construction and multi storey timber constructions transmitting high loads. Alternative applications have been proposed in heavily loaded bridge constructions including bridge decks. In particular in terms of serviceability the longitudinal post tensioning may provide more compact solutions at joints and supports showing no or little deformation. Additional benefit regarding strength could also result. Theoretical calculations as well as first tests on full size, nearly centrically post tensioned deck elements have been carried out. The focus was on a possibly improved lateral distribution of high point loads, less deformation and a good behaviour under cyclic loading. A good correlation of theoretical and measured deformations was found. The determination of the failure load of such systems will be carried out at a later stage. KEYWORDS: Cross Laminated Timber, post tensioning, modelling, full scale testing 1 INTRODUCTION 123 A road bridge for heavy truck traffic (2 x 4 tons) was planned to cover a span of about 34 m. The entire structure was conceived using CLT only. Due to limited width and thickness of gluelam beams the four large I- beams as well as the bridge deck were planned to be realised using this novel material in order to ease construction and to provide stiff, non deformable load carrying timber elements. Due to various reasons, an alternative bridge construction was realised. As no experience and data regarding longitudinally post tensioned CLT is available (part of the reason why the project proposal was not realised) a full scale experiment was designed in order to study the feasibility, service behaviour, behaviour at failure and the suitability for cyclic loading of such solutions. The main focus was to determine if the lateral distribution of high point loads could be achieved and if the post tensioning would have a favourable influence at service condition. Figure 1: Cross section of road bridge for heavy traffic In order to optimise the secondary system and to simultaneously slow down the traffic on the bridge the trafficable surface was divided into two separate (narrow) lanes and a central, elevated and covered walkway. More loads can be assigned to the easy to stabilise central bearers when providing a continuous transverse system. 2 DESIGN CONCEPT AND KEY ISSUES Solutions with two lateral main beams lead to extremely heavy transverse members to carry the high double truck load even providing narrow spacing of the members. 1 Christophe Sigrist, Bern University of Applied Sciences, Architecture, Wood and Civil Engineering BFH-AHB, Solothurnstrasse 12, P.O.Box, CH-25 Biel-Bienne 6, christophe.sigrist@bfh.ch 2 Martin Lehmann, Bern University of Applied Sciences, Architecture, Wood and Civil Engineering BFH-AHB, Solothurnstrasse 12, P.O.Box, CH-25 Biel-Bienne 6. martin.lehmann@bfh.ch Figure 2: View and dimensions of I-beam in CLT and longitudinal section of the post tensioned deck with critical truck loading positions

2 The bridge deck is formed by cross laminated panels arranged between the transverse system, a continuous Kerto panel protected by a waterproof membrane and the bituminous service layer. In order to reduce relative movements (opening and closing gaps, rotation at the support) between the transverse beam and the deck panel due to high truck loads (Q k,1 = 3 kn and q k = 9 kn/m 2 ) and to provide an improved continuity of the longitudinal system the deck was post tensioned (PT) over the full length. Spacers required to transmit normal forces across the transverse system in order to avoid stresses perpendicular to the grain are not represented in the drawings in Figure 2. For this novel concept the following key issues have to be solved: production of large and long CLT panels with various layups, the maximum length currently available in Switzerland is 3,4 m x 13,7 m optimise the panel layup design in situ assembly of large panels solve the connection between flange and web conceive the construction stages and the installation and tensioning of the PT strands on site study the short and long term behaviour of the post tensioned timber deck under static and cyclic loading. Figure 3: Full scale prototype joint for in-situ assembly of large CLT members During a workshop the site assembly of prefabricated panels was investigated and turned out to be rather efficient and simple. Protruding longitudinal layers which are slightly profiled at the end to ease the assembly and inserted boards or panels in the cross layer of adjacent members are brought in contact using the previously installed PT strands. The precise machining of the panels using CNC technology and optional lateral guides leads to a self alignment of the members. The overlap allows placing the required mechanical connectors to transfer internal forces. This technique may be used for both flat and vertical members. 3 EVALUATION OF LAYUPS 3.1 ANALYSIS OF LAYUPS The layups of cross laminated timber may be determined randomly in order to suit the planed application. Panel thickness, number of layers, the thickness and strength class of the individual boards may be determined to suit the needs. For a bridge deck high strength and stiffness in the longitudinal direction as well as good stiffness in the perpendicular direction in order to spread the high point loads are required. Various layups showing different arrangements of layers and various eccentricities of the post tensioning cable were investigated theoretically. The plywood analogy was used attributing the cross layers a stiffness of MOE = N/mm 2 for the six (P3) or seven layer panels of a total thickness of 21 mm considered. The eccentricity e z of the strand is given by the position of one of the longitudinal layers close to the neutral axes where the strand can be easily installed in a gap between the boards. The optimisation was carried out modelling the bridge deck using a plate programme varying section properties, support conditions and changing loading positions (point loads only). A plate on two lineal supports loaded by a point load was analysed. Table 1: Investigations on panel types, E* = apparent MOE of panel P1 P2 P3 P4 I x [1 6 x mm 4 ] I y [1 6 x mm 4 ] E* x [N/mm 2 ] E* y [N/mm 2 ] e z [mm] m x,max [knm/m ] m y,max [knm/m ] δ z,max [mm] P1: top layers in x-direction, 7 layers t = 3 mm layup is // // // // standard setup, good longitudinal and transverse behaviour P2: 2 top layers in x-direction, 7 layers t = 3 mm layup is // // // // // optimised for longitudinal behaviour, limited transverse action P3: top layers in x-direction, 6 layers 4 layers t = 3 mm, 2 cross layers t = 45 mm layup is // // // // improvement of transverse behaviour, acceptable deformations P4: top layers in y-direction, 7 layers t = 3 mm layup is // // // // optimised for transverse behaviour, substantially reduced longitudinal stiffness (high deformation) The composition of the layers was such to optimise longitudinal load carrying behaviour or to optimise the transverse distribution of point loads. 3.2 SELECTED LAYUP A regular layup, where every consecutive layer is rotated by 9, formed by 7 layers of C24 boards (spruce) usually used for the gluelam production was finally selected. The panel is 21 mm thick. The panel shows appropriate static behaviour and stiffness. The transverse distribution of loads is comparable to built ups where thicker layers or outer layers run across the main load carrying direction.

3 longitudinal perpendicular longitudinal perpendicular longitudinal perpendicular longitudinal Figure 4: Layup of deck and position of strand, the outermost layer runs parallel to the deck span mm space for post tensioning strand / in the third layer --> eccentricity (e ) of the strand 3.3 POST TENSIONING VSL technology is based on the principle of posttensioning (PT) usually used for reinforced concrete structures where the prestress is permanently introduced into the structure. This is achieved by the stressing of suitably arranged, high-strength prestressing tendons. VSL post-tensioning generates favourable stress conditions in the structure, enabling efficient use of building materials while controlling deformations under service conditions. Single strands or tendons composed by multiple strands are positioned in the cross section in gains. Composite action may be achieved when grouting the gains once the prestressing force is applied ensuring the bond between the cable and the concrete. For the application presented here the single strands were used which remain unbonded in guide pipes. As timber shows equally good strength in tension and in compression, PT is not really needed. However the additional, mostly centric compression stress might help to bridge defects in the tension zone. Only a small eccentricity of the compressive force is applied helping to balance some of the dead load acting on the bridge deck. Due to the continuity of the deck too much eccentricity is not desired in order not to overstress the deck at the intermediate support. One of the benefits anticipated at ultimate limit state (ULS) is to increase the effective width of the deck portion located under the concentrated load as lateral load distribution through the various layers composing the deck is very limited. The main idea to applying post tensioning to the bridge deck is to improve the behaviour at service limit state (SLS). Superposing transverse and longitudinal systems or loosely connecting those leads to substantial rotation, longitudinal and lateral movements at the supports due to the heavy, travelling and dynamic load. Post tensioning of the entire deck results in a rather monolithic block which for both short term dynamic loads and long term service should lead to a much better behaviour. The prefabricated panels had a dimension of 3.6 m (span) x 2. m (width) and were produced using vacuum bags. The panels were to be tested as a partially continuous two span beam as the connection at the intermediate support was of large interest. A lateral 3 z distance between neighbouring boards of the longitudinally arranged timber layer allows to create longitudinal openings / gaps and to positioning the guide pipes in the panel during production. Four strands of diameter.6 inch (A p = 15 mm 2 ) were positioned in the third layer from the bottom with an eccentricity of 3 mm as indicated above. The strands were fed through the guide pipes in order to achieve the desired continuity. A post tensioning force of 1 kn per strand (VSL post tensioning system) was applied and monitored using load cells throughout the experiment. Standard VSL hardware like anchor heads and profiled steel wedges as presented in Figure 5 and tensioning systems (monostrand hydraulic press) were used to permanently apply and anchor the PT force. The stresses due to the post tensioning lead to a centric compression of 1,7 MPa and a negative bending stress of 1,1 MPa on the bridge deck. Figure 5: VSL hardware (press, anchor head and wedges) to post tension the bridge deck 4 METHODS 4.1 EXPERIMENTS Due to the size of the specimen no climatisation was undertaken however the panels were produced using boards at a moisture content of 1% w 15%, then delivered and installed in the test rig. A more or less uniform moisture content around 12% can be expected. As several different experiments were performed using the same specimens the load was kept well within the elastic domain of the CLT. Two different loading configurations were tested: Centric loading (2 kn) at midspan in each span Eccentric loading (5 mm across; 2 kn) at midspan in each span. For both loading configurations static as well as cycling loading was performed. Due to time constraints and as the CLT will be used for further investigations the cycling loading was limited to 2 cycles. In order to determine the influence of post tensioning regarding load distribution and deformation the loads were applied before and after post tensioning for each

4 loading configuration. Partial continuity over the inner support was reached after installation and tensioning of the strands. Table 2: Tests performed during the experimental study # type post tensioned configuration 1 static no centric 2 static yes centric 3 cycling yes centric 4 static no eccentric 5 static yes eccentric 6 cycling yes eccentric 36 deformations of the supports (stiffness of testing rig) were measured during the tests and considered in the theoretical analyses. The theoretical study was limited to the single span deck without post tensioning. Further numerical investigations using Ansys will be undertaken in the near future. 5 RESULTS The deck wasn t loaded to failure as more detailed and additional investigations are planned. The stresses due to the post tensioning lead to a centric compression of 1,7 MPa and a bending stress of 1,1 MPa. The external forces created bending stresses of 17 MPa in the deck. Design stresses in tension and compression were approximately reached under the load of 2 kn at the extreme fibre. 2kN 2kN 5.1 CENTRIC LOADING A B C Figure 6: Static system and geometry of the experiments with and without post tensioning 7.2mm 11.9mm 13.5mm 11.9mm 25.5kN 7.6mm 12.8mm 14.8mm 23.6kN 7.6mm 12.8mm 14.8mm 23.6kN Multiple measurements were taken during the various loadings. Vertical deformations of the deck were taken at mid-span and two intermediate positions across the deck (Figure 7). One linear support (support C in Figure 6) was replaced by 4 load cells in order to determine the reactions due to loading. Two load cells were used to monitor the post tensioning force in the unbonded PT strand during installation and under load condition mm 11.9mm 13.5mm 7.9mm 11.3mm 13.1mm 8.3mm 12.3mm 13.9mm 8.4mm 12.1mm 14.2mm 25.5kN 19.kN 29.4kN 28.2kN C4 B4 A4 X4 K4 RK2 7.9mm 11.1mm 12.9mm 21.9kN C3 B3 A3 K3 RK1 C2 B2 A2 K2 6.3mm 9.1mm 1.9mm 9.1mm 16.9kN C1 B1 A K1 7.1mm 1.4mm 12.1mm 7.6mm 1.8mm 12.7mm 26.4kN 25.8kN deformation measurement load cell on strands force (35mm x 35mm) support with load cell 7.4mm 9.7mm 11.6mm 2.3kN Figure 7: Position of the different measurements 4.2 CALCULATIONS For the calculations for the CLT panels a simple 2D-FEprogram designed to calculate concrete decks was used. The program allows for analysing orthotropic materials varying stiffness properties E*I eff and the appropriate coefficient of Poisson ν. The MOE parallel to the fibre direction was set to 11 GPa the remaining elastic constants of Picea Abies were taken from Neuhaus [1]. The bending stiffness of the CLT in each direction was determined as described in chapter 3.1. The Figure 8: Deformations and support reactions for the model (top), single span deck (middle, no post tensioning) and the post tensioned, partially continuous deck (bottom) The symmetric loading of the post tensioned deck would lead to a reduction of the deformation of about 2%, probably due to the rather stiff connection and continuity of the strands at the inner support (Figure 8). Furthermore the reaction at the outer supports was reduced by about 1% for the same reason. For the

5 centrically loaded deck a good correspondence between measured and calculated deformations are observed (Figure 8 to Figure 1). deformation at 2kN [mm] 16 A1 A2 A3 A calculated 4 single span 2 prestressed position over the width [mm] Figure 9: Deformation across the deck for symmetrical and central loading deformation at 2kN [mm] C2 B2 A2 calculated single span prestressed position over the length [mm] Figure 1: Deformation along the deck for symmetrical and central loading (only one half of the deck is shown) which could lead to serviceability problems regarding the superimposed layers for water tightness and the bituminous service layers. This could be overcome by installing a centric PT load composed by two strands positioned in top and bottom layers. During the first loading cycles (n = 3) the deformation slightly increased but afterwards the deformation and the hysteresis were constant (Figure 12). The loading of 2 kn lead to an increase of the post tensioning force in the strand (1 kn) of about 5% which is 5 kn. Nearly no permanent shift of the force in the tendon was recognised during the maximum 2 cycles performed. load [kn] deformation gauge A2 [mm] Figure 12: Load as a function of the deformations at midspan for 2 cycles of the post tensioned and centric loaded deck 5.2 ECCENTRIC LOADING The eccentric loading confirmed the trends obtained from the tests applying the centric loading. The calculated support reactions are quite different from the measured values. The transverse load distribution seems to be over estimated by the theoretical analysis. The deformations in the three sections A to C considered and the support reactions are about symmetric. Due to the restricted capability of the cross layers to distributing the loads in transverse direction more support action on the inner supports is measured. The measurements confirm that the (about centric) post tensioning still leads to a certain degree to a continuous system. Figure 13: Post tensioned bridge deck under eccentric load configuration Figure 11: Central support of the deck. Left: Post tensioned but unloaded; right: post tensioned and loaded However a post tensioning strand in the upper part of the deck would be needed for avoiding the rotation of the section at the (Figure 11). This rotation leads to an opening gap of about 1 mm during loading For the eccentric loading the agreement between the calculated and measured deformations was not as good as for the centric loading. Again the transverse load distribution seems to be overestimated by the theoretical analysis. This effect is even more pronounced regarding the reactions at the inner supports considering eccentric loading. The more distant support reaction with respect to the load position were hardly loaded or even showed uplift (Figure 14) during the experiments. It seems as if a very limited effective width of the deck is mobilised.

6 Figure 14: Deformations and support reactions for the model (top), single span deck (middle, no post tensioning) and the post tensioned deck (bottom) During the first 7 of the 2 cycles a slight shift of the hysteresis can be detected as shown in Figure 15. Long term loss of PT / shrinkage and swelling of the CLT panels as well as possible fatigue problems are to be investigated for final applications. load [kn] mm 16.1mm 19.3mm 16.7mm 8.2mm 13.7mm 16.1mm 6.9mm 11.1mm 12.3mm 5.8mm 8.7mm Figure 15: Load as a function of the force increase in the post tensioning strand for two hundred cycles for the eccentric loaded deck 6 CONCLUSIONS 9.1mm 13.2mm 2.4mm 25.4mm 19.3mm 9.9mm 14.7mm 16.5mm 6.4mm 9.6mm 1.9mm 2.5mm 3.6mm 4.9mm 11.6mm 18.4mm 23.3mm 18.1mm 9.mm 13.3mm 15.mm 6.3mm 8.5mm 9.9mm 3.3mm 3.7mm 4.6mm force increase in strand [kn] 49.6kN 23.8kN 21.6kN 3.4kN 49.1kN 44.6kN 6.4kN no contact 48.2kN 37.4kN 6.7kN no contact The investigations and calculations seem to confirm a very interesting behaviour of post tensioned deck structures. The strands can be easily installed once the CLT is positioned. The tension force shows quite some variations. It increases and decreases again by about 5 % when the deck is loaded respectively unloaded. Long term losses of the tension force are not expected as the CLT panels show virtually no deformation due to shrinkage and swelling. The long term behaviour of post tensioned CLT decks have to be investigated thoroughly as transversally post tensioned decks have shown substantial problems and require frequent re-tensioning of the PT bars throughout their service life [2,3]. The crucial connection detail at the support allowing the prefabrication of the deck panels, optimal erection of the timber components and easy installation of the PT strands once the bridge deck is fully assembled must still be studied. Special steel brackets or load carrying distance holders directly transferring the loads from one deck element to the other and not to load the transverse beam perpendicular to the grain is still to be developed and designed. Applications for such systems in floors and walls of multi-storey buildings may also be possible. Of particular interest are (ductile) bracing systems to transfer the high horizontal loads from wind and earth quake action from wall to wall segments and to the foundation. Analogies to post tensioned brick wall constructions may be of particular interest for such developments. In terms of durability the use of PT strands should not create particular problems. The greased strands with an additional PVC protection extruded onto the strand (technique used for strands in stay cables) are further protected by the guide pipes required for prefabrication and installation of the PT system. Combined with the high requirements regarding the protection of the timber components reliable and durable solutions are achievable. ACKNOWLEDGEMENT The investigations could be carried out during a semester project in summer 211 thanks to M. Eggenberger, R. Illi and D. Roder. The PT strands and the tensioning equipment were sponsored by VSL (Switzerland) AG, Subingen ( and valuable support given by A. Gnägi and U. Meier. CLT was partially sponsored by Schilliger Holz AG, Küssnacht, Switzerland. REFERENCES [1] Neuhaus H.: Elastizitätszahlen von Fichtenholz in Abhängigkeit von der Holzfeuchtigkeit. Dissertation, Inst. für konstruktiven lngenieurbau, Ruhr-Universität Bochum, Bochum, [2] Ritter M. A., Hilbrich Lee, P. D.: Recommended Construction Practices for Stress-Laminated Wood Bridge Decks. In: International Wood Engineering Conference (IWEC'96), [3] Crews K.: Recommended Guide for the Design of Stress Laminated Timber Plate Bridge Decks. Design Procedures & Commentary, University of Technology, Sydney, 1995.