Prefabricated enduring composite beams based on innovative shear transmission

Transcription

1 Prefabricated enduring composite beams based on innovative shear transmission (PRECO-BEAM) Research and Innovation EUR EN

2 EUROPEAN COMMISSION Directorate-General for Research and Innovation Directorate G Industrial Technologies Unit G.5 Research Fund for Coal and Steel rtd-steel-coal@ec.europa.eu RTD-PUBLICATIONS@ec.europa.eu Contact: RFCS Publications European Commission B-1049 Brussels

3 European Commission Research Fund for Coal and Steel Prefabricated enduring composite beams based on innovative shear transmission (Preco-Beam) G. Seidl, E. Viefhues SSF Ingenieure AG, Leopoldstraße 208, München, GERMANY (SSF) J. Berthellemy Service d Étude Technique des Routes et Autoroutes, Avenue Aristide Briand 46 BP 100, Bagneux, FRANCE (SETRA) I. Mangerig, R. Wagner Universität der Bundeswehr München, Werner Heisenberg Weg 39, Neubiberg, GERMANY (UBWM) W. Lorenc, M. Kozuch Politechnica Wrocławska, Wybrzeze st. Wyspianskiego 27, Wroclaw, POLAND (TUWRO) J.-M. Franssen, D. Janssen Université de Liège, Plac du 20 Aout 7, 4000 Liège, BELGIUM (ULGG) J. Ikäheimonen, R. Lundmark RamböllSverige AB, PO Box 4205, Stockholm, SWEDEN (RAMBSV) O. Hechler, N. Popa Arcelor Profil Luxembourg S.A., rue du Luxembourg 66, 4009 Esch-sur-Alzette, LUXEMBOURG (Arcelor) Grant Agreement RFSR-CT July 2006 to 30 June 2009 Final report Directorate-General for Research and Innovation 2013 EUR EN

4 LEGAL NOTICE Neither the European Commission nor any person acting on behalf of the Commission is responsible for the use which might be made of the following information. The views expressed in this publication are the sole responsibility of the authors and do not necessarily reflect the views of the European Commission. Europe Direct is a service to help you find answers to your questions about the European Union Freephone number (*): (*) Certain mobile telephone operators do not allow access to numbers or these calls may be billed. More information on the European Union is available on the Internet ( Cataloguing data can be found at the end of this publication. Luxembourg: Publications Office of the European Union, 2013 ISBN doi: /9363 European Union, 2013 Reproduction is authorised provided the source is acknowledged. Printed in Luxembourg Printed on white chlorine-free paper

5 Table of contents SUMMARY... 5 SCIENTIFIC AND TECHNICAL DESCRIPTION OF THE RESULTS... 9 OBJECTIVES OF THE PROJECT... 9 DESCRIPTION OF ACTIVITIES AND DISCUSSIONS ANALYSIS OF PRECO-BEAM SYSTEM EXISTING INVESTIGATIONS CONSTRUCTION MATERIAL Consideration of materials for Preco-Beam Fatigue of steel due to cutting Influence of residual stresses Fire resistance of composite girders EVALUATION OF APPLICATION AREAS Preco-Beam in buildings Preco-Beam in bridges ECONOMIC EFFICIENCY AND DURABILITY INVESTIGATIONS IN BEARING BEHAVIOUR Mechanical models Actions in the steel part of the dowel Description of the FE model Conclusions for steel part Fatigue General: Fatigue in steel construction state of the art Investigations in fatigue by FEA Conclusions for fatigue Concrete part and geometrical influences General D Modelling Verification to tests / studies D Modelling for beams Behaviour of girders STATIC AND CYCLIC LABORATORY TESTS GENERAL Best shape TEST EXECUTION Test setup Push Out tests Pull Out tests Beam tests Load application EVALUATION OF TESTS Push Out Tests / Pull Out Tests Testing of shear transmission strip Testing of geometrical influences Conclusions Beam tests T-Beams steel part Conclusions for steel part Slabs geometrical influences Conclusions for geometrical influences

6 3 PRECO-BEAM EXPOSED TO FIRE GENERAL FEA FOR FIRE LOADS Thermal analysis Mechanical analysis LABORATORY TESTS General Test execution Evaluation of test results DESIGN OF PRECO-BEAM STRUCTURAL DESIGN FIRE DESIGN VALIDATION OF PRECO-BEAM SYSTEM IN PRACTICE CONCLUSIONS EXPLOITATION AND IMPACT OF THE RESEARCH RESULTS LIST OF FIGURES AND TABLES APPENDICES APPENDIX 1: DESIGN GUIDE DISTRIBUTED IN THE FRAME OF PRECO-BEAM APPENDIX 2: VALIDATION OF PRECO-SYSTEMS IN PRACTICE

7 SUMMARY The major objective of the Preco-Beam project is to elaborate cost effective and sustainable bridge structures of high quality using a new construction method as a combination of filler beam decks and prefabricated composite structures and to prove their durability to obtain competitive composite bridges. The bearing behaviour of a new connector the composite dowel has to be analysed in detail by FEA and experimental tests taking the interactions in a girder, material and geometrical influences, fatigue behaviour and fire into account. Designers and authorities will be conducted through the design process by offering them a comprehensive design guidance including design recommendations and design examples. In some first realised projects references for the clients can be given. This aid is imperative as administrations and clients as well have been reserved when it came down to benefit from this promising bridge type as design questions concerning the behaviour of the composite dowel not addressed by design codes so far need to be analysed and composed to a design guide. Investigations in state-of-art literature show that composite dowels are not state of the art nowadays but gain of importance. Thus little research concerning the bearing behaviour of composite dowels or similar connectors has been done yet. General technical approvals exist for connection systems acting similar to the Preco-Beam composite dowel. Some investigations about the shear transmission of continuous shear connectors have been done in the frame of research projects and first design rules for these connectors have been elaborated. However detailed information about interaction of loads in girders and influences of geometry, reinforcement or material is still missing. For the applicability of the system in buildings more knowledge about the behaviour under fire load is required; just as for application in bridges a detailed analysis of the fatigue behaviour is indispensable. 5

8 Figure 1 Project overview Thus the required extension of knowledge on the one hand concerns the bearing behaviour of Preco- Beam. This means in particular the bearing behaviour of the shear connector. More detailed and realistic FE models have to be elaborated with regard to the interaction between steel and concrete. This as well as wide ranging experimental tests should help to analyse the shear transmission in a local and the bearing behaviour a global way and to develop mechanical models to explain the bearing process in a comprehensible way. To force the application of Preco-Beam a design guide for the composite dowel in girders for the clients and designers is required. Of high importance is the increase of information about the fatigue behaviour. In the frame of the Preco-Beam project some Pilot Projects should have been elaborated but it turned out to be a difficulty to enlist clients for the application of Preco-Beam in bridges as there is no experience with composite dowels under cyclic loads. To enlarge Preco-Beams application area e.g. in buildings further investigation about fire resistance is necessary and possibilities to reach determined fire resistance class have to be developed. On the other hand a major objective is the evaluation of economic efficiency of the system. Application areas such as possible spans and loads have to be determined, as well as the economy of construction time and simplification of the procedure on the construction site. The construction process in detail has to be analysed and recommendations for the fabrication of the composite dowel itself and the whole 6

9 composite girder should be given. Not least the costs of this construction method are of importance to ensure the application of the system in practice. Several activities have been started to improve the existing investigations of the composite dowel and the Preco-Beam system in general regarding as well the bearing behaviour as the economic aspect of the method. Mechanical models are elaborated to analyse the load processes of the Preco-Beam system with regard to shear transmission and the behaviour of girders considering influences like haunches and reinforcement. The existing FEA models are enlarged for in depth investigations. Thus elastic models are evaluated to analyse the processes regarding steel which are verified by the test results subsequently. For the concrete part nonlinearities due to the concrete s material law are introduced in a 3D simulation. The complex load interaction of steel and concrete is represented by contact elements. This results in an advanced FE model with high computation time and partly instability within the calculation process. So a 2D model is evaluated, which shows good accordance to the 3D model and the test results. Thus an effective way it given to study influences of concrete and geometry of the Preco- Beam. For the verification of the initial investigations in the frame of Preco-Beam and to establish the results for the application of Preco-Beam in practice wide ranging tests with regard to the steel and the concrete part and the influences of geometry are performed. The test program contains static and cyclic tests to specify the bearing and the fatigue behaviour of the shear connector. Taking the investigations in FEA into account, first tests using three different dowel shapes result in a recommendation for a best shape. The puzzle shape is chosen to be the best shape due to the production process, the bearing capacity and the fatigue behaviour. The remaining tests are performed afterwards, already considering this recommendation. The test program contains Standard Push Out Tests and beam tests, to analyse the bearing behaviour of the dowel in girders under bending conditions. It can be shown that next to influences like material grade, interaction of steel strips, reinforcement and width of the concrete web the position of the concrete dowel in the girder is of importance, whereas the dowel size does not influence the bearing capacity of this connection system. The ductility of the shear connector can be increased by apply two reinforcement bars at the steel dowel s bottom. The production process and the dowel shape itself are of vital importance for the fatigue behaviour, as notches decrease the fatigue resistance. The cutting line should be designed in a way which enables the production of a dowel by one cut only to avoid stopping of the cutter, as this would result in intensive heat introduction and probably generates notches. During the project s progress and with regard to all the experience gained by the tests an advanced shape has been developed which combines all advantages of the puzzle shape but shows a better fatigue behaviour due to its geometry and the production process. The application of Preco-Beams in buildings heavily depends on the fire resistance of the system. One objective is therefore the investigations of Preco-Beams exposed to fire. Theoretical FE analysis are performed and verified by experimental tests. Thus it can be noted that the shear connection itself is not influenced by fire remarkably; the protection of the beams and their strengthening can be similar to which is well known from composite girders. 7

10 All these investigations are compiled in a design guide which enables designers to dimension the Preco- Beam system and especially the composite dowels. The design guide classifies the composite dowel s resistance by SLS and fatigue limit state (FLS) for the steel part and by ULS for the concrete part. As a result of the research of Preco-Beam under fire it is possible to classify the girders due to their fire resistance. Mechanical models are used to analyse the load processes in the connector, thus an easy application of the design rules is guaranteed. Recommendations regarding the dowel shape, the reinforcement and the design of the girder are given. In another part investigations concerning the economic aspect of the new construction method are done. The range of application is analysed by several studies and it can be shown that for buildings the system is advantageous for spans up to 20 m, as high loads can be carried by a lightweight and slender construction. The material costs itself are not lower than for conventional systems but if the boundary conditions necessitate minimal construction height and high bearing capacity Preco-Beam is the appropriate construction system. For bridges the advantage is more evident, as high prefabrication is not usual for conventional composite structures. Now short construction times and high quality standards are possible. Compensation of formwork at the site by the pre-casted concrete members decreases the construction cost and time considerably; automation of the connector s production by cutting the steel girder minimises the effort of work. By realisation of a first pilot project experience is gained regarding the construction progress and the handling of the girders. It can be highlighted that the production of girders with spans about 26 m does not cause any problems. The steel girders can be prepared with high quality at the steel workshop. The reinforcement can be applied directly on the steel girder at the prefabrication plant. No purpose-built formwork is necessary as the usual formwork for prefabricated concrete elements can be used. Prefabricated elements of high quality and high precision can be delivered to the construction site and enable a construction process without any problems. Due to the cost effectiveness and the smooth project progress the client already applied the Preco-Beam system for other projects. Several other pilot projects have been started, but as at the beginning of the Preco-Beam project investigations and experience regarding fire and fatigue have been missing, clients have been reserved and hence most projects are still at a preliminary state. Nevertheless about ten projects could be started during the Preco-Beam project term. 8

11 SCIENTIFIC AND TECHNICAL DESCRIPTION OF THE RESULTS Objectives of the project Bridges are of importance to the European infrastructure and composite bridges already became a popular solution in many countries and a well-established alternative to concrete bridges [Collin, et al., 1999]. They are gaining in importance due to the economic and versatile application. Since 1998, bridges have been created in a composite pre-fabrication (VFT = Verbund-Fertigteil- Träger = prefabricated composite beam) method of construction [Schmitt V., et alt., 2001]. The prefabricated composite beams consist of a steel beam with a concrete flange, which serves as a compressive chord and formwork element for the site-mixed concrete deck. These not only absorb the compressive stresses while the bridge is under construction, but also stabilise the beam while it is being transported, and render unnecessary the installation of bracing for concreting of the site-prepared concrete deck. Concrete and steel are connected by headed shear studs. This construction method permits extremely short erection times. The Preco-Beam system combines the advantageous of the filler beam plates like robustness and slenderness and VFT girders. Due to the high degree of automation even for the fabrication of the connectors and regarding to the high degree of prefabrication shorter construction times and more efficient constructions are possible. Preco-Beam designs enable the configuration of composite bridges as single span girders or frames with slenderness ratios of with spans of up to 50 m. This feature finds application both in the creation of new structures that conserve resources and also in the replacement of existing bridges. The Preco-System is notable for its efficient use of material. As determined by the design configuration using steel profiles without an upper chord (e.g. halved rolled girders), the steel in the span is loaded in tension. In the column and framework corner region there is similarly sufficient material available in order to accommodate the compressive forces without the introduction of additional plates into the concrete. With the separation technology used it is possible to achieve a high quality for the separating faces with minimum local notch effects. In this manner it is possible to ensure the favourable fatigue characteristics of the composite medium that have been substantiated from tests, so that in the future the concrete dowel technology could also be used for structures under cyclic loads. In general there are no recommendations for composite dowels used in Preco-Beams. Thus the approval of the responsible authority is always necessary for use of the beams, which often introduces a significant loss in time into the building programme. In many cases this actually prevents the use of the beams. Within the Preco-Beam project further investigations will be done regarding the general bearing behaviour of composite dowels in composite girders with halved rolled sections. Influences as dowel shape, geometry of the girder, reinforcement, interaction of steel strips and the influence of fire will taken into account to develop design rules and recommendations for the usage of 9

12 Preco-Beams. The validation of the research results will be done by realising some first pilot projects to evaluate the feasibility and the economic efficiency of this new construction method. 10

13 Description of activities and discussions 1 ANALYSIS OF PRECO-BEAM SYSTEM 1.1 Existing investigations At the beginning of the Preco-Beam project the existing investigation concerning composite dowels had been assembled. Three fundamental investigations had been considered within the further work of the project: Stuttgart University - Perfobond strip: Several tests have been done [Andrä H.-P., 1990] and the Technical Approval has been reached in 1991 [Zulassungsbescheid Z , 1991]. The application is allowed in buildings and bridges (except railway lines). The influence of cyclic loads can be considered by a reduction of 2/3. Another Technical Approval has been reached in 2000 by KOMBI TRAGWERK [Zulassungsbescheid Z , 2000]. University of Federal Armed Forces Munich: At UBWM push-out-tests have been performed and a design-guidance has been elaborated by Zapfe [Zapfe C., 2001] which arises from the investigations of Wurzer [Wurzer O., 1998]. The design rules are applicable for concrete dowels in concrete plates. Minas Gerais (Brazil) and Guimaraes (Portugal): Within this project several push out tests have been performed [Veríssímo G., et al., 2006]. In comparison to the other tests the introduction of the load is different. That is why the tests result could not been taken into account. In the scope of the research project P621 [Hechler O., et al., 2007] the Standard-Push-Out-Test (POST) and composite beam tests have been used to investigate in the fatigue behaviour of shear connectors in high strength concrete. Also the puzzle shape has been investigated as continues shear connector. The test evaluation shows, that no fatigue damage could be identified in the high strength concrete and the reinforcement. However fatigue cracks have been verified in the steel of the puzzle strip with a crack initiation from the gas cut edge. Detailed information about interaction of loads in girders and influences of geometry, reinforcement or material is little. Therefore more realistic FE models have to be elaborated as well as wide ranging experimental tests to analyse the shear transmission in detailed and global load bearing behaviour. Mechanical models are desirable to explain the bearing process in a comprehensible way. To force the application of Preco-Beam a design guide for the composite dowel in girders is required. For the applicability of the system in buildings more knowledge about the behaviour under fire load is required; just as for application in bridges a detailed analysis of the fatigue behaviour is indispensable. In the frame of the Preco-Beam project some Pilot Projects should have been elaborated but it turned out to be a difficulty to enlist clients for the application of Preco-Beam in bridges as there is no experience with composite dowels under cyclic loads. A major objective is the evaluation of economic efficiency of the system. Application areas such as possible spans and loads have to be determined, as well as the economy of construction time and 11

14 simplification of the procedure on the construction site. The construction process in detail has to be analysed and recommendations for the fabrication of the composite dowel itself and the whole composite girder should be given. Not least the costs of the construction method are of importance to ensure the application of the system in practice. 1.2 Construction material Consideration of materials for Preco-Beam At the UBWM former investigations concerning concrete dowels are done mainly with normal strength concrete. Test series with static loads and with cyclic loads have been done. The tests have shown a strong correlation between the bearing capacity and the shape of the dowel. Outcomes of these tests result in design equations for mainly normal strength materials. Only in few existing investigations high strength concrete were used. These tests led to an increase of the bearing capacity. As well FEM calculations have shown an increase of the bearing capacity for high strength steel. Therefore two combinations of materials were recommended within the Preco-Beam project. For bridges with high loads and therefore high shear forces combination of C70/58 and S460, for buildings with fewer loads C45/55 and S460 is used Fatigue of steel due to cutting A basic application of Preco-Beams will be bridges thus the fatigue behaviour of the structure is of particular importance. As it has been pointed out in [Hechler O., et al., 2007] the steel part of the composite dowel is the critical part regarding fatigue. The fatigue behaviour can be influenced mainly by the design of the dowel shape and the fabrication of the steel girder for Preco-Beams the cutting process itself. On the one hand sharp notches in the dowel shape should be avoided to minimize the stress concentration due to the redirection of the force. On the other hand the cutting process of the steel girder itself influences the fatigue behaviour of the steel girder strongly. In the frame of Preco-Beam the first steel specimens were done by oxy-fuel cutting at ArcelorMittal s workshop. Oxy-fuel cutting is a combustion process. A mixture of oxygen and fuel gas is used to preheat the metal to its ignition temperature but below its melting point. A jet of pure oxygen is then directed into the preheated area setting off an exothermic chemical reaction between the oxygen and the metal to form iron oxide, slag. The oxygen jet blows away the slag enabling the jet to pierce through the material and continue the cut through the material. The oxidation of iron by this method is highly exothermic. The oxy-fuel process is the most widely applied industrial thermal cutting process principally because it can cut thicknesses from 0.5 to 2500 mm, the equipment is low cost and can be used manually or mechanised. Thermal cutting processes inevitably induce localised expansion and contraction exceeding yield point strains. The degree of thermal distortion is related to both the heat input of the cutting process and the physical and mechanical properties of the material being cut. Lowest distortion is achieved with lower heat input processes or cold cutting processes such as machining and abrasive water jet cutting. It has to be noted that stopping of the flame cutter should be avoided, as this decreases the fatigue strength significantly [Sonsion C.M., et al., 1992]. If a connector shape requires stopping of the flame cutter or a second cut with the flame passing twice the same 12

15 location the location should be designed at an insignificant area in respect to the fatigue loading. Figure 2 shows some influences of the cutting process. Important to the quality is next to others the edge roughness and the edge squareness, which define the cosmetic appearance and the angularity tolerance. By the permeation of heat due to the cutting process the width of the heat affected zone is influenced. This means a detectable micro structural change in the area of the cut and thereby a distortion and a hardening of the steel. The width of this distortion and hardening zone influences the fatigue behaviour of the steel as well. The main criteria used to assess the quality of a cut are as follows. kerf - defined as the width of the cut at its widest point. The kerf gives an indication of the minimum internal radius or feature that can be cut. Cut Edge Roughness (Rz, mm) - used to define the cosmetic appearance of a cut and can give an indication of whether subsequent machining operations are necessary. Cut edge roughness is determined by an Rz value in microns. This is a measure of the surface roughness transverse to the cut edge produced by traversing at 2/3 depth with a stylus and taking an average value. Cut edge Squareness, (U, mm) - defined in terms of the perpendicularity and angularity tolerance. This is a measure of how much the cut edge deviates from a perfect square edge. Edge squareness affects the fit-up between two components and determines whether any post cutting machining operations will be necessary. Heat Affected Zone Width (HAZ) - defined as the width of a detectable microstructural change measured perpendicular to the cut edge face. HAZ width is only applicable to alloys that undergo microstructural changes during the heating and cooling cycle of the cutting operation, i.e. alloys that are hardenable or heat treatable. The width and properties of the HAZ are important due to the region s possible greater brittleness or susceptibility to crack initiation and the possibility of the need for removal of this material before final assembly of the product. Figure 2 Influences of the cutting process In consideration of these facts it has to ensure that the visible signs from edge discontinuities due to the cutting process have to be removed e.g. by machining. According to [EN , 2005] the fatigue category of a gas cut edge should be taken as 140; conservatively as 125. For both categories repair by 13

16 weld refill is not allowed. Re-entrant corners are to be improved by grinding appropriate stress concentration factors. The influence of hammering (an effect which may occur due to the hammering of the continuous shear connector in the gaps from which the plastified concrete may have disappeared) has additionally been investigated. Hammering in parallel load direction is decreasing the fatigue strength about 30%, hammering in an angle of 45 at the edge decreases the fatigue strength about 20%. The difference may be explained by the differing residual stresses induced with the diverging angles. Thus hammering is not resulting in a change of the crack initiation and can be neglected. Further the warming before cutting have no influence on the fatigue strength. The material strength for steel S355 up to S890 has hardly an influence. Thus in respect to the following influences on the fatigue strength as: Influence of the shape of connector Material dependencies Influence of the surface roughness (due to cutting process) Residual stresses (due to cutting process) Influence of flaws at the cutting edge Influence of hammering due to gaps from the destroyed concrete As the oxy-fuel cutting is the most widely applied industrial thermal cutting process all test specimens for this project are manufactured by oxy-fuel cutting to ensure a practical fabrication. However short stopping of the flame cutter is decreasing the fatigue strength significantly Influence of residual stresses As a result of modification of eigenstresses after the beam is longitudinally cut, two T-beams are obtained, which deformations are usually similar to required structure precambering, but very often they need to be additionally straightened. The problem of redistribution of residual stresses existing in rolled beams appears during cutting and precambering. This problem is very complicated as distribution of eigenstresses in rolled girders can differ depending on many factors and it is not easy to estimate however we can expect some usual distributions that one can find in literature. As residual stress system is rather unknown in rolled girder, it is hard task to estimate value of stress distribution after separation by cutting. It results in addition with different deformations and precambering makes the situation even more complicated. The problem is described on the basis of experience from realization of VFT-WIB railway bridge supported by basic numerical calculations by FEM. Steel beams HE1000x438-S460ML have been cut with MCL shape to realize composite girders. Beam lengths was 17m, eight beams have been cut in two as 4 simply supported spans needed to be realized. After separation the vertical deformation of T-shapes at midspan differed from 60 up to 120mm. As all beams were straight before cutting and the same cutting technology was used, one can notice that residual stresses had to be different in the beams. 14

17 a) b) c) Figure 3 HE1000x438 a) before cutting, b) cutting process, c) after separation Simple FE calculations taking into consideration residual stress distribution during separation and precambering with yielding of material have been conducted to present the crucial points. The geometry of the beam is the same as the one used in the bridge, the other aspects should be treated virtual and they are assumed to present crucial points in simple way. For example precambering in reality is realised with 3-point bending at many beam points here is modelised like bending of total beam supported at the ends. The process of separation is modelised with FEM, heating influence is not included in the model as it looks that there is no big difference comparing standard cutting along straight line. No dowels are taken into consideration straight line is assumed, ¼ symmetry model was used, eigenstresses have been modelised by strains and connecting springs have been removed from model after load implementation. Assumed system of strains is of course one of many possible solutions. Deformations have been compared with experimental ones during cutting. Figure 4 FE model and load by strains*10e-3 15

18 Figure 5 FE study of cutting process deformation under eigenstresses modelised by strains in rolled girder before and after separation. Results are presented by stress layouts: Figure 6 Normal stress σ1 and deformation before and after cutting (blue = tension, red = compr.) a) b) c) d) Figure 7 Stress values at the end of beam: before separation: a) σx, b) σy and after separation: c) σx, d) σy 16

19 Displacement at midspan after cutting is 26mm and comparing to 120mm leads to conclusion that residual stress value had to be really high; on the other hand one can conclude that residual stress system can be assumed different way. One can notice that value of σy at the end of beams is high and corresponding to FE calculations on the basis of crack obtained during cutting stress value during cutting was so high that steel part was left to keep T-shapes together was not strong enough and it failed: Figure 8 Beginning of cutting: failure of steel plate due to high σy value As the beams are usually deformed after separation and displacements are too high comparing to required geometry, precambering is needed. Precambering process is modelised with FE in following way: 1) T beam is simply supported what results in deflection of 21.4mm at midspan, 2) uniform load is implemented towards flange to get displacement at midspan 320mm and yielding at dowel region, 3) force is removed to get final deformation and stress layouts. This process reflects situation when initial deformations after cutting need to be decreased. No residual stress system after rolling is included. Elastic-perfectly plastic model for steel is used with yield limit 460 MPa, ¼ symmetry is used in the model, brick elements are used. a) b) Figure 9 Model of the beam: a) general view, b) finite elements at midspan region 17

20 a) b) c) Figure 10 Reduced stress: a) step 1 beam under dead load, b) step 2 force applied: yielding of web is visible, c) step 3 stage after force removed: new residual stress layout is visible One can notice that web was yielded and flange worked elastic (274 MPa). After load was removed displacement at midspan was 71mm and such a cambering process resulted in complicated residual stress layout; plastic strains are combined with big tension stress values. The biggest plastic strain values would appear at critical region at the beginning of dowel root such a situation needs to be expected during real precambering also. 18

21 a) b) Figure 11 Plastic strains at longitudinal direction: a) step 2, b) step 3 (midspan region) With help of FEM and realisation experience problem of redistribution of residual stresses is highlighted. Both separation and precambering influence stress distribution in steel dowels. Observations from fabrication proved that residual stresses can differ in theoretically the same straight beams as deformations after separation differed sometimes 100%. Moreover, two parts of the same beam deformed different way what can suggest the residual stress can be asymmetric in steel beam section. Hence detailed recommendations cannot be given at this stage and further investigations are needed. 19

22 1.2.4 Fire resistance of composite girders The fire resistance of Preco-Beam is comparable to concrete encased composite sections analogue [EN , 2006]. The concrete encased composite sections can be classified into fire resistance classes. Exposed to fire the materials are changing their properties regarding their bearing behaviour. For higher temperature the E-Modul and the yield strength decrease. Deformation and deformation rate increase progressively until ultimate failure. The critical temperature for steel is at 500. [EN , 2006] gives three alternative methods for determination of the fire resistance of composite members [Hanswille G, 2003]: 1. Tables: Calculation of the load-bearing ratio for cold conditions. The tables give a maximum load-bearing ratio in consideration of cross-section type and dimensions. 2. Simplified calculation method: Decreasing of the cross-section or of the strength and stiffness of parts with regard to their position. 3. General calculation method: Thermal analysis including an analysis of the stresses at the global system taking thermal material properties into account. If the tables do not fit for the considered cross-section, the simplified calculation method is the most common; a general calculation using FEA means high effort and may be not accepted in different countries. For method 2 the dimensions of the cross-section are reduced with regard to the fire resistance class and the ratio Area/Volume. For the residual cross-section the design values for stiffness and for strength under cold conditions are reduced with regard to the fire resistance class and the position of each part. The evaluation of the bearing capacity is done in accordance to cross-sections under cold conditions. Information about general correlation between temperature and bearing behaviour can be taken from investigations of Schaumann and Upmeyer [Schaumann P., et al., 2002]. high load-bearing ratio means lower fire resistance small dimensions of cross-section means lower fire resistance higher steel grade means lower fire resistance higher reinforcement ratio means higher fire resistance higher concrete grade means higher fire resistance Figure 12 Examples of regular composite sections with ensured fire resistance [Merkblatt 117, 1991] 20

23 Figure 13 Preco-Beam sections To increase the resistance to fire the experience for steel sections / encased composite sections can be taken into account again. An embedded steel girder already is protected against fire. Higher bearing capacity can be obtained by additional reinforcement (Figure 12c and Figure 12d, Figure 13d). In case that the steel part is arranged outside the concrete slab (Figure 12a and Figure 12b, Figure 13a) additional protection by plasterboard or coating will be necessary. 1.3 Evaluation of application areas Preco-Beam in buildings For the design of Preco-Beams for buildings the following demands have to be taken into account: cost effectiveness light weight of elements slender construction easy and economic fire resistance Traditional floor slabs are shown in Figure 4. Prefabricated concrete units like type (B) can be used up to 10 m span. For longer spans slab types (C) and (D) could be applied. They can be used up to 20 m span transferring loads about 20 kn/m². Disadvantageous is their sensitive behaviour regarding to high (point) loads. Conventional composite slabs (E) are used for high loads. A: In situ cast concrete slab B: In situ cast concrete slab prefabricated concrete units C: Prefabricated and Pre-stressed concrete slab (Hollow section) D: Prefabricated and pre-stressed concrete slab (TT-element) E: Traditional composite beam with welded studs F: Preco-Beam system Figure 14 Different types of floor slabs 21

24 Table 1 Compare of different methods with the span length 16 m Parameter A B C D F Live load capac. ult. state [kn/m²] not not Ratio live/total load recom. recom % 50 74% 71 92% Live load capac. ser. stat [kn/m²] for for Ratio live/total load 16 m 16 m 30 50% 24 58% 34 77% Height of section [mm] Dead load [kn/m²] Table 2 Compare of different methods with the span length 20 m Parameter A B C D F Live load capac. ult. state [kn/m²] not not Ratio live/total load recom. recom % 49 59% 49 87% Live load capac. ser. stat [kn/m²] for for Ratio live/total load 16 m 16 m 20 24% 20 34% 20 54% Height of section [mm] Dead load [kn/m²] To analyse the application of Preco-Beams in buildings in a frame of a study [Nilsson G., 2007] the different slab types were compared regarding span, bearing capacity and slenderness next to other items. Sections with C30/40 and S355 were compared. For span about 8 m length composite beams and Preco-Beams are not recommended regarding to economical and effort aspect. Results for span of 16 m and 20 m are shown in Table 2 and Table 3. The following figures present the load / span curves for the system C and D and the Preco-Beam. The curves show the load capacities in ultimate limit state (ULS) and in safety class 3. Generally precast HD/F elements are very common. In Figure 15 the different sizes are presented. The width of all products is always 1200 mm. The elements are of pre-stressed concrete. The load / span curve in Figure 17 shows the bearing capacity of HD/F elements. It can be seen the change between shear- and bearing capacity when the curves are bending a little bit; at the left side the shear capacity is decisive criterion and at the right side is the moment capacity is decisive criterion. The curves show that HD/F elements have lower capacities compared to Preco-Beam elements when the span length increases up to 17 m and more. 22

25 Figure 15 HD/F elements / TT/F precast elements TT/F precast elements are also very common. In Figure 15 a sketch of an element is shown. The width of the products is 2400 mm and sometimes 3000 mm. The height varies from 200 to 640 mm. The elements are out of pre-stressed concrete. The thickness of the concrete in the flanges is about mm for the elements with heights of mm and mm for the elements with heights mm. Figure 17 shows the load capacity for both Preco-Beam and precast TT/F element. The curves shows, similar for HD/F elements, that TT/F elements have lower capacities compared to Preco-Beam elements when the span length increases up to 17 m and more. The material in the Preco-Beam elements is C45/50 for the precast concrete and C35/45 for the in situ concrete but for simplicities in the calculations is used C35/45 for concrete. The steel grade is S355. The width of the element is 3 m and the thicknesses are 50 mm for the precast concrete and 200 mm for the in situ concrete. Figure 16 Preco-Beam cross-section used in the frame of the study As conclusion from the study it can be summarised that for short spans (< 8 m) traditionally in situ concrete or prefabricated and pre-stressed concrete elements are used as they are cheapest and strong enough for most types of constructions. Preco-Beams are not beneficial in comparison to traditional slabs in economical regard. For medium spans (8 16 m) usually prefabricated and pre-stressed concrete elements or composite girders are applied. Preco-Beams are advantageous for high (point) loads for these spans and could be used additionally for strengthening. For the application for span length more than 16 m also prefabricated and pre-stressed concrete are used. As these slab types are sensitive to high loads, Preco-Beam will be successful with high (point) loads for this span. It is advantage that it is not necessary to change the construction height regarding to the different load levels because the distance between the steel girders can be varied. 23

26 mm prefab, 200mm in situ, 3m width HD/F precast element HEB 400 HEB 500 HEB 600 Load excl. dead load (KN/m2) HEB 700 HEB 800 HEB 900 HEB 1000 HD/F 120/15 HD/F 120/19 HD/F 120/20 HD/F 120/27 HD/F 120/32 HD/F 120/38 HD/F 120/40 HD/F 120/ Span (m) Load excl. dead load (KN/m2) mm prefab, 200mm in situ, 3m width TT/F precast element HEB 400 HEB 500 HEB 600 HEB 700 HEB 800 HEB 900 HEB 1000 TT/F 240/20 TT/F 240/30 TT/F 240/40 TT/F 240/50 TT/F 240/60 TT/F 240/24 TT/F 240/34 TT/F 240/44 TT/F 240/54 TT/F 240/ Span (m) Figure 17 Load / span curves for Preco-Beam elements (ULS) and precast HD/F elements / Preco- Beam elements and TT/F elements Preco-Beam in bridges To apply Preco-Beams in bridges it is important to ensure a sustainable construction in aggressive conditions as the climate is changing. Furthermore for high loaded structures like bridges which has to last more than 70 years it is indispensable that the structure needs to be inspected easily during its life cycle. Due to the danger of shock the structure should be resistant to impacts of vehicles. The item of fatigue always is particular important for cyclic loads on bridges. So this means an advantage for Preco- Beams as the fatigue resistance of a structure with rolled girders and composite dowels and without welding is high in comparison to other composite and steel structures. Additionally for railway bridges it is important to obtain a stiff structure with small deflections caused by high traffic loads to ensure a fixed position of the rails. After all it should be kept in mind that this system will be used in practice only if the cost effectiveness is more or less in competition with other types of structures. 24

27 Figure 18 shows cross-sections which could be feasible for bridges. Cross-section (a) might be applied in bridges which are not at risk of vehicle impact, for example for bridges over rivers. Cross-section (f) is convenient for railway bridges with high loads. It is similar to the very common construction method of embedded steel girders. This type may be useful for locations where a minimum construction height is desired. Figure 18 Feasible cross-sections for bridges In all cases the high prefabrication ratio and the fact of few effort on the construction site is an advantage over traditional in situ construction methods. The short erection time will be favourable to the economy and efficiency of the Preco-Beam for bridges. 1.4 Economic efficiency and durability One main objective of the developing of new construction methods is to create economic and durable systems to obtain competitive composite bridges and ensure the application in practice. On the other hand the economic efficiency of the Preco-Beam system can be seen in the production of the beams. First experience with the production of Preco-Beam in the frame of the test series and with the construction procedure in the frame of the pilot projects as scope of this research project approved the idea of this construction method to be very fast and easy. The construction procedure of a Preco- Beam can be described in the following steps: 1. Oxygen cutting of the rolled beam section. The composite dowel results of the cutting of a steel girder into two halves (Figure 19). 2. Coating of the girder at the steel plant (Figure 19). This extensive work can be easily done under plant conditions thus high quality and durability can be achieved. Afterwards the girders have to be transported to the pre-casting plant. 3. At the pre-casting plant the reinforcement can be placed directly on the steel girder without any disturbing formwork (Figure 20). First experience in practice (pilot project in Vigaun) show that prove this advantage. 4. Together with the finished reinforcement the steel girders are lifted into the form work. The form work can be used several times and is adapted from prefabricated concrete elements. Afterwards the prefabricated element is concreted (Figure 21). 5. For a certain time the prefabricated element has to be supported in a strainless way (Figure 22). This is necessary to ensure the composite system to act for the dead loads already. Thus the materials can be used in an effective way from the beginning. 25

28 6. The prefabricated girders can be transported to the construction site. Due to the comparatively lightweight construction the elements can be transported as usual (Figure 22). 7. The girders are placed on the abutments. The proper fabrication of the elements at the plant avoids problems on the construction site ( Figure 23). 8. Finally the in-situ concrete can be added without using any additional formwork. The prefabricated concrete is acting as formwork. Figure 19 Oxygen cutting of the rolled beam section / Coated steel beams Figure 20 Installation of reinforcement Figure 21 Girder and reinforcement are lifted into the formwork / Concreting of the prefabricated element 26

29 Figure 22 Strainless support / Figure 23 Detail stiff-joined abutment Due to rapid and easy construction and due to the combination of approved construction systems the costs concerning the production and construction of the Preco-Beams can be minimized. Closing off periods can be reduced to a minimum because of the high prefabrication level of the superstructure at the plant. Thus high quality standards and a reduction of imponderability at the construction site are possible. Usage of standardised girders of rolled steel simplifies the availability and the delivery times. These low-cost rolled steel sections supersede any welding in the workshop. For the production of Preco-Beams standard formwork of pre-stressed concrete precast girders can be used; so no new investment in the pre-casting plant is necessary. Thanks to the low installation weight of the girders usual cranes can be used. Next to the production costs the constructions durability has an important influence to the total lifecycle costs. This asks for structures which are robust, produced with high quality standards and can be inspected easily. The Preco-Beam system satisfies the demand of durability in several aspects: Small corrosion protection surfaces Robust reinforced concrete cross section High standard of quality thanks to the large degree of prefabrication in the rolled girders and the precast parts Open structure for easy structural inspection To evaluate the economy of Preco-Beams in bridges three different construction methods using prefabricated elements has been compared: the Preco-Beam method, VFT girders and pre-stressed girders. For pre-stressed girders usually H-sections are an economic solution if the bridge deck is standardized which is the case especially in the Eastern Europe countries due to their prefabrication facilities. The highlighted already built bridges are in Centre European countries (Germany and Austria) where the T-beam sections are common in use. That is the reason for comparing these cross-sections according to Figure 24. For this methods the production, preparations, the construction process and the effort on the construction site is nearly the same. The Vigaun-Bridge is used as a reference project and stands for the Preco-Beam construction method: frame system three bays each m span composite joints at the columns width between railing = 4.70 m 27

30 bridge area = 370 m² For the VFT and the pre-stressed girder a preliminary design has been done to determine the material consumption. The dimensions can be seen in Figure 24. The costs are listed below separated for the items regarding the costs of the prefabricated girders. Figure 24 Dimensions of Preco-Beam girders for Vigaun-Bridge and of preliminary design for VFT and pre-stressed girders 28

31 Item Unit Preco-Beam amount unit price price p. item in-situ concrete plate, d = 0.25 m m³ reinforcement, in-situ concrete plate t prefabrication concrete m³ reinforcement, prefabrication concrete t steel, incl. corrosion protection t shear connector t 0-0 prestressing steel t 0-0 transport / placing include crane all in total costs cost / m² bridge 503 Item Unit VFT-girder amount unit price price p. item in-situ concrete plate, d = 0.25 m m³ reinforcement, in-situ concrete plate t prefabrication concrete m³ reinforcement, prefabrication concrete t steel, incl. corrosion protection t shear connector t prestressing steel t 0-0 transport / placing include crane all in total costs cost / m² bridge 534 Item Unit Prestressed girder amount unit price price p. item in-situ concrete plate, d = 0.25 m m³ reinforcement, in-situ concrete plate t prefabrication concrete m³ reinforcement, prefabrication concrete t steel, incl. corrosion protection t 0-0 shear connector t 0-0 prestressing steel t transport / placing include crane all in total costs cost / m² bridge 531 In comparison to the VFT girders high reduction of costs can be reached by less amount of steel. The steel of a Preco-Beam is a little bit more expensive due to the cutting but the amount of steel can be reduced to 75 %. Especially the saving due to the shear connector using for the VFT girders has to be highlighted. Finally the transportation and placing of the girders at the construction site is cheaper, as the Preco-Beams do not have to be fixed for the transportation. For comparison of Preco-Beam with pre-stress girders it can be seen that high reduction of costs can be reached regarding the prefabricated concrete. For the pre-stressed elements a width of 60 cm for the concrete web is necessary to have enough space to place the tendons. Thus the amount of concrete is much higher than for Preco-Beams which can be designed with a width of 30 cm. On the other hand this results in heavy pre-stressed girders with about 15 t more weight for each girder. Due to the more lightweight girders of the Preco-Beam method the transport and placing is cheaper here. Finally the high costs for the tendons can be saved. In Annex 1.2 information about final costs is given. 29

32 1.5 Investigations in bearing behaviour To enable clients and designers to choose Preco-Beam with composite dowels it is essential to offer them a comprehensive design guidance including design recommendations and rules. Thus the bearing behaviour of the composite dowels both the shear transmission and the behaviour in girders - has to be analysed. Basis of the research will be the investigations of [Zapfe C., 2001] and [Wurzer, O., 1998]. They pointed out three failure mechanisms of the concrete part exceeding of possible compression in the concrete dowel core, pry-out of concrete above the steel dowel, shear failure of the concrete dowel and shear failure of the steel part of the composite dowel. To extend the knowledge about the internal forces and to verify mechanical models Finite Element Analysis (FEA) should be done. FEA is an economic and efficient method for parametrical studies to optimize the design unlike time-consuming experimental tests. However, numerical models are often complex, the behaviour is high-grade nonlinear and influenced by many parameters. Composite dowels are very complicated task to handle by FEA due to a three-axial stress situation generated in concrete dowels combined with large stress gradients in concrete and nonlinear interactions between steel and concrete. Moreover, different shapes and dimensions are possible, which influences the slip-behaviour and load-bearing capacity of this connection. Hence the first task to handle this is to get a good agreement between experiment and theoretical results in push-out tests. Then some general assumptions concerning numerical models would be the basis for later FE study and shape optimisation. Figure 25 Composite dowel nomenclature An elastic-plastic model for investigations concerning the steel part is elaborated using the FE software ABAQUS (done at TUWRO). To determine geometrical influences and concrete behaviour an additional FE model is established using the FE software ADINA (done at UBWM) Mechanical models The behaviour of the composite dowel can be divided in two elements of the composite action. The force flow interacts between the steel dowel 1 and the concrete dowel 2 (Figure 26). Both components consist of two different materials showing individual properties under load. 30

33 A composite dowel consisting of a very stiff steel dowel which is completely embedded in concrete achieves the highest capacity. That s why no destructive influences caused by local stress peaks act in the contact surface. This maximal capacity is defined as general capacity. Under load the adhesion, the friction and the local compression dowel is activated at the contact surface of the dowel. Figure 26 Composite dowel, a composition of steel dowel (1) and concrete dowel (2) and their influence of load bearing behaviour The flexible steel dowel is deformed by an increasing force on the dowel which mainly depends on its slenderness. High and slender dowels are yielded like a cantilever loaded by a bending moment. Steel strips with only small openings are under shear forces comparable to a deep beam and are much stiffer. Slender dowels generate a high ductility due to a failure of shear. But the slenderness reduces the maximal capacity of the composite dowel caused by the initial distribution of the load introduction to the dowel base. Stress peaks are introduced and it results in destruction of concrete in this area. When the yielding in the steel dowel is obtained, the connection is intensely deformed until the ultimate strain of the steel is reached. Normally the load-slip diagram is on its peak. A stiff dowel introduces smooth and hence favourable local stresses into the concrete. If bending is obtained the stresses move in direction of the dowel root. Concrete is destroyed in the compression area and the concrete dowel is uplifted. The undercut of the steel dowel carries the uplifting forces. The load bearing behaviour of a composite dowel is characterised by means of the typified load-slip diagram in Figure 28. It is divided in three sections of loading. Load range 1: Linear elastic behaviour, mobilisation of friction and local compression on dowel surface Load range 2: Plastified material behaviour accompanied by initial cracks in the concrete or ignition of yielding of the steel dowel until consolidation; the maximum shear force is obtained (Figure 27) Load range 3: Generally the behaviour after the failure shows an apparent crack pattern in the concrete and also initial cracks in the steel dowel An effectively reinforced and adequately embedded concrete dowel shows ductile behaviour beyond ultimate capacity. 31

34 Figure 27: Possible failure modes in load range 2 I) Yielding of steel dowel by bending and shearing II) Vertical crack in the non-reinforced plate III) Shearing of the concrete dowel IV) Horizontal crack in the concrete V) Spalling of concrete cover VI) Pry-out cone in the concrete cover At the beginning of loading the two surfaces are bonded by adhesion which is a very stiff connection. Consequently in the load-slip diagram the slope of the graph is steep. Once the adhesion is surmounted, the load proportion is reduced to the value of friction, which is illustrated as load proportion A in Figure 28. If the load is increasing corresponding to the slip, the front surface of the dowel is partially loaded which introduces the main part of the dowel force into the concrete dowel. If the steel dowel wasn t completely embedded in the concrete, the stresses by the partial loaded surface would crush the concrete to the opposite side of the load introduction. This process is specified in MC90. This wedge cannot crush because of the entirely concrete surrounding. The concrete matrix in front of the dowel is compact pulverised. 32

35 Figure 28: Load proportions of a typified load-slip diagram (w/o bond) to be supposed in ductile concrete failure: A: Adhesion/Friction of contact interfaces B: Compression of steel-concrete surfaces C: Shear in the concrete interfaces D: Dowel action of transversal reinforcement E: Block effect of twin re-bars in the concrete dowel *: Pulverisation of concrete Forced by the proceeding slip this front-sided pulverised wedge penetrates the concrete dowel. The partial loading creates splitting in the concrete dowel. If the tensile force of the concrete is exceeded it a crack in the plain of the concrete dowel and the dowel occurs reinforcement bears the tension force. In this state the force of the steel dowel is introduced into the concrete by skewed compressive struts. The concrete dowel is loaded by shearing the more the steel dowel penetrates the concrete and the wider the crack opens up. The local compression B changes into the force of the skewed compression struts load proportion C. An adequate reinforced concrete specimen may feature two different failure patterns. In plate-shaped concrete specimen the concrete cover pries out like a cone at the maximum load starting in the center of the concrete dowel. If the concrete plate is narrow or the specimen like a beam, a spalling of the cover occurs in the part which is not reinforced. If the shear capacity of the concrete dowel is obtained, the slip between steel and concrete is already that large that the reinforcement bar in the concrete dowel locks the concrete interfaces carrying the value D in Figure 28 which improves the post-failure behaviour. If the concrete matrix continues to crush, the implementation of twin reinforcement bars increases the load bearing capacity of the concrete dowel mainly because of the block action. More details are figured out in [Seidl, G., 2009] 33

36 1.5.2 Actions in the steel part of the dowel Description of the FE model Push-out tests conducted by [Fink, J. et al., 2006] were the basis for the initial FEM study. Three fundamental concrete failure mechanisms and steel failure had to be considered. The proposed FE model, which is partly validated by experimental results, predicts this specific behaviour of the structure. A numerical model is generated with the objective of analysing the structural behaviour of the connection and determining the failure mechanisms. ABAQUS software was used. For numerical simulation of composite dowels complex geometry combined with a multiplicity of nonlinearities has to be taken into account. To establish an elementary model is difficult, especially due to the concrete part which is high-graded nonlinear in whole load range. Therefore the following aspects are focused: material nonlinearities contact interactions complex geometry Push-out tests are an elementary method to determine the bearing capacity and to study the behaviour of a shear connector. Hence a numerical model of push-out test is created first. Investigation of the model s behaviour is possible by comparing it with experimental results. Basically the model consists of three parts: steel, concrete and reinforcement bars. As the push-out specimen has two symmetric planes symmetric boundary conditions are used and only ¼ of specimen is modelled. This model is indicated 3P1, because three teeth of the push-out specimen are taken into account. FE elements in the steel part represent half of web thickness (plane 1-3 view). The size of all FE elements is similar. Nonlinear material laws are applied to the structural steel and to the concrete. A linear material law is applied to reinforcement bars. The Mises criterion of isotropic hardening is implemented for construction steel and Concrete Damaged Plasticity model describes the concrete behaviour. It must be pointed out, that no degradation of stiffness is applied as well as no decreasing part of tension curve appears: the value of f ct after cracking stress is constant. A curve for uniaxial compression is defined according to [Model Code 90, 1993]. This concrete model is justified if an element is generally under compression. The decreasing part of the curve under tension results in local problems inside the concrete dowel and not realistic post-failure behaviour: The force-slip curve is rapidly decreasing after the maximum load level. Concrete Damaged Plasticity material parameters are: eccentricity e = 0.15 dilation angle ψ = 15 K c = The ratio of stresses σ b0 /σ c0 = 1.16 is assumed, f c, f ct, and E c are then determined according to experimental results from the concrete compression tests. Contact between steel and concrete is assumed to be hard and friction coefficient is taken with R = 0.3. The specimen in the FE model is supported at the bottom surface of concrete part (Figure 19). The predetermined vertical displacement is applied to the upper surface of the horizontal steel plate. Therefore a rigid body is implemented. Hence u 2 acts at a single point which defines the reference point. The reference point is situated at the intersection of the symmetry planes of pushout specimen. 34

37 The displacement u 2 is increasing from zero to u 2,max which ranges from 3 mm to -10 mm. In accordance to the symmetry in plane 1-2 and plane 2-3 appropriate symmetric boundary conditions are used. Hence the load-slip relation is defined by the reaction force R 2 versus the displacement u 2 measured in the reference point. In general the steel part and the concrete part are modelled with continuum elements. Different element sizes were studied for one model. Reinforcement bars are modelled with beam elements instead of truss elements because they enable dowel action according to [Model Code 90, 1993]. This difference influences the behaviour and load capacity of the structure. Hence trusses do not seem to be the appropriate choice for the simulation. The beam elements are embedded in the concrete elements with the appropriate boundary conditions in plane 1-2. Figure 29 Geometry of push-out (3P1) model: a) mesh, b) components Figure 30 Push-out model results: a) Mises stress, b) u2 displacements, c) additional plate which enables modelling of contact support, d) Mises stress (yielded steel) for 3P4 model (with high number of elements) Two different methods are common for solving numerical problems: implicit method (ABAQUS/Standard) and explicit method (ABAQUS/Explicit). Both approaches are applied and compared due to the ability to solve the problem and the efficiency of calculation. Finally explicit method is chosen as the favourable method to simulate the behaviour of concrete dowels. In ABAQUS/Explicit approach the time steps are controlled by the stability limit of the central difference operator. This procedure is efficient for large models and for the analysis of extremely discontinuous processes. If the explicit method is used smooth step curve is applied for displacement u 2. Contact interactions with kinematic contact method and default weighting factor is implemented in the solution. The comparison of FEM results with experimental results confirms that the generated push-out model represents the behaviour in the push-out tests quite well. The bearing capacity of the connection is depending on the confining effect of concrete by the reinforcement bars. Also the boundary conditions 35

38 are important as well as the concrete material law. It is stated that using the decreasing part of concrete law, which represents tension, used with Concrete Damaged Plasticity -model results in a local failure mechanism. This numerical problem does not appear in the real structure. This simplification is substituted later by detailed concrete study by UBWM in chapter A stable model of push-out test is defined, some general statements need to be done and different structural elements need to be analysed. It is obvious, that many parameters can influence the behaviour of the model, e.g. time period t tot, variables and coefficients in concrete, Concrete Damaged Plasticity -model, the size of elements, boundary conditions etc. Their influence was discussed in [Lorenc W., et al., 2007]. A simplified model was necessary to derive from which parameters the structural behaviour is influenced. Therefore different material laws are combined and the failure mechanisms are evaluated due to the influence of the parameters. The following combinations were investigated: NLsNLc (nonlinear steel/nonlinear concrete) NLsLc (nonlinear steel/linear concrete) LsNLc (linear steel/nonlinear concrete) LsLc (linear steel/linear concrete) The general approach to calculate the shear transmission of composite dowels is a one-steel-toothmodel embedded in reinforced concrete (1D1-model, Figure 31). Appropriate boundary conditions and interactions are necessary to represent the behaviour of the structure. The geometry and the boundary conditions and interactions are shown in Figure 31. The 1D1-model was a basic tool for initial study of concrete behaviour and extensive parametric study of steel shape. Figure 31 The 1D1-model: a) boundary conditions, b) parts and interactions Conclusions for steel part The work of TUWRO concerning concrete was finished at this stage and detailed concrete behaviour was studied by UBWM TUWRO focused on FE study of steel part under SLS conditions. ULS for steel was handled by Mises criterion and it was not the main problem in design; as ULS concrete resistance is lower than steel resistance in realistic structures, what was confirmed by experimental tests. Hence all models derived by TUWRO later are LsLc models (linear steel/linear concrete). Nonlinear procedures are used due to the contact conditions. As it was confirmed by the study small sliding formulation for contact 36

39 conditions can be applied at SLS conditions and the problem can be treated as almost linear. Finally it will result in linear formulas for design of steel. At the beginning the stress state in the steel dowels was assumed to be a superposition of three different actions as discussed in [Hechler O., et al., 2008]. The global action (G) is generated by the nominal stresses in the steel girder and the notching effect at the root of the steel dowel. Local action (L) is caused by the shear force in the composite joint. The third action (U) uplift was supposed to influence the superposition of stresses and was assumed to depend on the location of the shear joint in respect of the neutral axis of the cross section [Hechler O., et al., 2008]. State of plane stress conditions are assumed for steel dowels. It should be pointed out that principal and reduced stress in dowel base, represented by edges in the side view, are changing in accordance to the arc length and it can be described against variable s. The correlation depends on the connector s shape, varies for different actions and is nonlinear with regard to s. Hence compared to studs, the steel dowels have to be considered as an integral part of the steel girder. This results in complex superposition of stresses in the steel dowel root (Figure 33). In the frame of Preco-Beam three possible dowel shapes (Figure 32) were studied and their optimal a/h ratios were calculated by FEA for local actions L. PZ (Puzzle shape) SA (Fin shape) CL (Clothoidal shape) chosen ratio for PZ Figure 32 Shapes 37

40 Figure 33 Results of searching of optimal ratio for SN shape and CL shape Local action L was studied at first. On the basis of presented FE results and by some theoretical studies it was stated, that the resistance of steel strip can be established per meter as it does not depend on the dowel s size but it depends on the dowel s shape only. This is a very important remark as the size factor can be excluded from the study. The resistance factor A el can be calculated for particular shapes and for each shape an optimum a/h ratio can be found (Figure 33). FE studies for different shapes were conducted to establish the best a/h ratio. It has to be pointed out, that the represented bearing behaviour of the CL shape and its high resistance is caused by the fixed upper part of steel dowel in concrete. The elastic resistance for SLS can be calculated from (Equation 1): v y A el t w f y (Equation 1) Analogue the influence of the global force G does not depend on the dowel s size but on the dowel s shape. One obtains a notching factor, which is compared to effect of longitudinal stiffeners. Thus the increase of stiffness depending on the length/height ratio of the dowel is influencing the notch effect. For the validation of the loading on a steel dowel and the resulting stresses the FE-analysis to the PZ shape has been consulted. It was noticed during study that steel dowel must be analyzed with large part of steel web. Hence for the local effects due to longitudinal shear, the model of NPOT specimen (see chapter 0 for NPOT) was modified to model (M3). Hereby the shear force is introduced into the steel part via the concrete part by using contact interactions. The stresses due to the notching effect and the nominal stresses in the web and the uplift forces have been calculated considering only the steel part in model (M2). The concrete part is negligible for those actions. The geometric properties of the dowel have been chosen to b 2 = 125 mm, h = 100 mm and t w = 10.2 mm. Hereafter the influence of each loading on the stresses of a single puzzle tooth has been evaluated along the cut. For this purpose, separate calculations have been conducted for P = 50 kn (L), P Up = 50 kn (U) and the global stresses in the web σ N = 50 MPa (G). Uplift was assumed to be the same value as P and applied according to front of steel dowel only. In Figure 23 the resulting principle stresses at the critical region of stress 38

41 concentration are presented, both as stress plots and as diagram in dependency of the location along the edge of the radius. Figure 34 Principal stress distribution in steel dowel along arc length from actions: σ N ( G), P (L) and P up (U) Figure 35 Beam model for WP3 tests (a: steel part, b: concrete part, c: FE model of the beam) and d: appropriate model for determination of G and L actions. The results were transformed to design formulas. The crucial point was to establish the nature of uplift. It was done during beam tests experimentally and confirmed by FE theoretically. Once it was stated that uplift governed by flow of forces in composite beam is theoretically zero (see chapter 0), more precise models for determination of G and L actions were prepared Figure 35. Afterwards design formulas were derived from appropriate mathematical transformations. The formulas are presented in chapter 0. 39

42 1.5.3 Fatigue As mentioned in chapter fatigue is of vital importance for the steel part and can be neglected for the concrete part. Quality defect of the cutting line can compromise fatigue resistance. A notch appears in the picture on the cutting line at the point of a puzzle shape where the curvature of the line has to change. The cutting tool probably stopped at the passage point from a rectilinear zone into a circular zone of the cutting line with reduced radius of curvature. FE studies from above show that the notching factor depends on the increasing of stiffness, too. To enable an evaluation of fatigue resistance some additional studies for the different dowel shapes have been carried out and are presented below. Concerning steel bridges, constructive details are classified with the risk of crack initiation caused by fatigue in the [EN , 2005]. Once the fatigue crack appears, it can lead after a phase of propagation to brittle failure which has to be avoided. The classification of the constructive details is usually elaborate by tests. In the frame of Preco-Beam only few cyclic tests are foreseen so theoretical research should enlarge the knowledge about the fatigue behaviour of the composite dowel by FEA. Figure 36 Manufacturing defect can appear in the detail of the cutting line of a puzzle form Preco- Beam connector General: Fatigue in steel construction state of the art The fatigue strength of the bridge details under the effect of the longitudinal direct stresses caused by flexion compression or traction, are already classified in the [EN , 2005] on the basis of experimental results. The following picture shows most of the current bridge details and their fatigue class. FEM models have been built for various representative details with the aim to establish a correlation between the Eurocode classification, and the fatigue damage caused in the different geometries. The same history of load variation is applied to every detail, and the fatigue damage is computed with the classical Palmgreen-Miner damages summation law that is described in Eurocode 3. Code_Aster (version 9.0), from Eau de France R&D, has been used to study the fatigue behaviour of composite dowels. This is FEA software with advanced tools for the study of structure regarding fatigue. For every detail, a uniform stress of 50 MPa is applied in a first step to an extremity of the plate. The load is applied in 12 steps according to the schema in Figure

43 Figure 37 Bridges fatigue strength details for direct stress ranges Figure 38 Introduced loads for fatigue analysis In the sense of the Eurocode, the total stress variation is 100 MPa. Stresses due to the thermal shrinkage of the welds or of the cuts are constant and are not introduced in the stress history. For all details the 41

44 Woehler curve used by Aster is the 125 MPa curve. It corresponds to the quality of cutting that is available today in France for bridge fabrication. It has been tested that this assumption has no impact on the results which are only computed to compare details. For one cycle with the stress variation of 100 MPa, the damage can be calculated easily to 1 / n 1. It can also be regarded as a probability of crack initiation during this cycle. Figure 39 Woehler curve Investigations in fatigue by FEA As a matter of verification, this first damage value 1/n 1 = E-6 can be obtained in a simple plate without any geometric disruption of the stress flow. It corresponds to the class 150 MPa of the Eurocode because no cutting or welding defect has been described in the Code_Aster model. It can be noted that the damage and the stresses are quasi uniform all over the plate. As the plate is supported by the web in the middle of in Z direction and also blocked at the web in the Y direction to avoid Poisson effects, stress concentrations occur here. Similar calculations have been performed for other details. 42

45 Figure 40 Simple plate with longitudinal stresses (average damage = E-6 ) / Slope 1:4 / Vertical butt weld with short attachment of 50 mm Application of the IIS recommendations In the precedents examples, the narrowness of the mesh could be selected in order to achieve a result independent on the refinement of the mesh. But in some particular cases of 3D hot spots, the result concerning welded details with poor class regarding fatigue, always depends on the refinement of the mesh of the FEM analysis. No convergence appears with the narrowness of the mesh. The IIS/IIW (international institute of welding) published recommendations [Hobbacher A., 2005] proposing a way to handle this convergence problem. These recommendations have been followed by generating a appropriate mesh and by using a smoothing device to post-process the results. With these precautions, the classification of poor class regarding fatigue welded details is coherent with the other details. The concerned details are the long vertical butt weld attachment of 300 mm, and the gusset plate but welded without radius transition at crossing flanges Preco-Beam shapes Following figures show the mesh supporting points and the general view of the mesh for the different shapes proposed. Below the damage evaluation for the different shapes is shown. 43

46 Figure 41 FE models for fatigue analysis for Preco-Beam shapes Conclusions for fatigue The study presents a method to evaluate the robustness of standard assembling details. Models of constructive details from the Eurocode 3 are introduced for Code_Aster to evaluate the fatigue damage under a standard given stress history. The comparison of the result shows a good correlation between evaluated damages and the experimental Eurocode 3 classification. The evaluation of the fatigue class of innovative details like Preco-Beam connection cuttings is made possible regarding longitudinal direct stresses, in other words compression or traction due for instance to the flexion of the beam. These results must be completed with fatigue tests by experience for future. Concerning the longitudinal stresses, at the origin of the main fatigue problem, the area that has to remain in a clothoidal form is a local area. The clothoidal geometry allows doubling the radius of curvature. The cut of a Preco-Beam connection in form of a clothoid with a radius of 80 mm in the sensible re-entrant angle has the objective to improve fatigue resistance by elimination of curvature 44

47 discontinuity in the sensible zone. For this reason, it is pertinent to propose clothoidal geometry for the cutting line of a Preco-Beam type connector to be used for bridges. Figure 42 Woehler for Preco-Beam details Concrete part and geometrical influences Next to the stress evaluation in the steel part and the extensive studies in fatigue behaviour of the steel part the concrete part of the composite dowel as well as the geometric influences has to be investigated to develop a design manual for composite dowels. For the investigations in SLS for the steel part the nonlinearity of the steel behaviour and the concrete material law could be neglected. To analyse the concrete part and thereby the geometrical influences like interaction of several dowel strips and reinforcement the nonlinearities and a realistic load transmission has to be taken into account which results in instabilities and extensive computing time of the calculations. Some assumptions have to be done and thus the FE models for concrete developed in the frame of Preco-Beam were verified by the test results and can be used for further investigations General The shape of the stress distribution under local compression in concrete is the basis of the load carrying behaviour of concrete dowels, as earlier researches [Zapfe C., 2001] and [Wurzer O., 1998] have shown. Therefore a 3D model of the test specimen respectively a patch of the test specimen is important to investigate and understand this subject. Particularly in regard to the aimed test setup, which contains also inquiries about the zone of influence of the dowel, a 3D-model is obligatory. For a preferably realistic model the contact in the concrete dowel is realised by contact elements. The contact elements allow an exact reproduction of the dowel influence zone in the concrete. 45

48 For the correct modelling of the load carrying behaviour and the force-deflection behaviour of the concrete in the array of the dowel, the material model Drucker is suitable. This constitutive equation is based on the theory of plasticity. The compression of the pore volume in the hardened cement paste, as a result of hydrostatic stresses and strains, is considered by an ellipsoidal yield surface. Initial position and peak proportion are appointed by the relation between hydrostatic compression and plastic volume expansion. The commercial software ADINA 8.5 was used for the FE investigations. Figure 43 ADINA model Drucker 46

49 D Modelling Figure 44: FE models in comparison to the POST (A: series C, B: series E, C: series F) For investigating of the concrete part the Push-Out specimens steel girder as well as the concrete slab are modelled by 8-node solid elements. To analyse local effects it is important to simulate the load introduction in a realistic way. Therefore the contact between steel and concrete is established by quadratic contact elements. The reinforcement is represented by beam elements. Due to the symmetry and economical reasons, only a half of a Push-Out specimen has to be designed. The Drucker-Prager model code was used for the concrete. It allows a realistic reproduction of the real concrete behaviour. Further, the Drucker-Prager model gives the opportunity to implement all concrete grades. Therefore, the needed parameters can be taken from [Model Code 1990, 1993] or experimental tests. The material characteristic of the steel (girder and reinforcement) is covered by a multi-linear approach. Therefore, the exact strain-stress curve of the steel can be implemented without any limitation. The described material codes lead to an implementation of normal strength and high strength materials in the FE models. The load is applied by a prescribed displacement. The models show a good accordance to the executed experimental tests as can be seen in Figure 44. Further, the load bearing behaviour of the FE model and experimental tests are shown. In Figure 44A the base model is represented and Figure 44B displays the model for the investigation on the t-beam cross sections. Finally, the interaction of two dowel strips is studied using the model of Figure 44C. 47

50 The stresses in the contact area are covered by the designed models quite well. Additionally, the elaborated FE models allow an indication of the failure criterion. The following Figure 45 displays the 3-axial stress state in the contact area. Figure 45: Stress mapping Finally, it can be stated, that the elaborated FE models are able to reproduce the load-bearing characteristic of the experimental tests adequately Verification to tests / studies The elaborated FE models are used for parameter studies. First, the load bearing behaviour of different T-beam cross sections as exemplarily shown in Figure 44B is studied. Therefore, the width of the T- beam was varied from 25 to 45 cm. The parameter study resulted in the diagram shown in Figure 46. Concrete dowels in T-beam cross sections with a T-beam width of at least 41 cm can reach the same load bearing capacity such as concrete dowels in conventional concrete slabs do. Further, the ultimate load can exceed the reference value of the standard tests if T-beams with a bigger width are used. This is an influence of the height of the T-beam. Load per Dowel 110% 105% 100% 95% 90% 85% 80% I nfluence of B eam width Influence of T-Beam width Standard Test Influence of T-Beam height 75% T-Beam width [cm] 41 Figure 46: Influence of T-Beam width The distance between two or more steel girders with Preco-Beam shear connectors also influences the ultimate load bearing capacity. FE studies using the model shown in Figure 44C are performed. The FEA show an influence up to a distance of about 100 cm. The result of the parameter study was confirmed by the experimental beam tests. The reduction of the maximum load capacity is subjected to the distance. 48

51 Influence of clearance Influence of displacement Distance % Maximum load [%] 99% 97% 95% Distance Clearance [mm] Figure 47: Influence of displacement D Modelling for beams By means of FE analysis, the tested composite beams are also investigated numerically for comparison. Two dimensional models are created in order to grasp the behaviour of composite girders. The detailed description of the composite girder and its FE model are shown below in Figure 48. Due to its geometrical symmetry only a half of the girder is modelled. Figure 48: Beam model All models consist of five different structural components steel web steel flange concrete slab steel reinforcement concrete dowels The concrete slab is modelled with concrete material provided by ADINA and meshed with square 8- noded 2D solid elements. Plastic multi-linear material is assigned to the steel web and flange. They are meshed with square 9-noded 2D solid elements and they are only distinguished by geometrical width inputs. Reinforcement is modelled by truss elements and plastic bilinear material is adopted. Finally, the composite dowels, namely the shear connection between the steel web and the concrete slab, are modelled as springs to save computation time and to ensure high stability of the FE model. This is different from the 3D-one-dowel-model in which the local effects should be investigated. And it is 49

52 assumed to be of no importance, as the beam models should investigate the global bearing behaviour. The force-displacement curve obtained by the push-out test is used to define the spring property. Test sample was built with two different widths of concrete slab 800 mm and 1000 mm and the characteristics values of concrete material C90/105 was selected based on the compression test of the cubic sample. The same composite girder but in upside-down configuration is also designed in order to simulate the behaviour of the composite girder near or on the middle supports where tensile stress dominates in the concrete slab. The results of the tests and the numerical simulations are compared in Figure 49. Load (kn) FE Test Deflection (mm) Load (kn) FE 200 Test links Test rechts Slip (mm) Figure 49: Comparison between experimental Test and FEA Next, the test specimens with two steel girders are simulated. Dimensions are given in Figure 50. The distance between two steel girders is either 350 mm or 450 mm. This composite girder is simplified for 2D FE model by inputting a doubled thickness or width in x direction and doubled spring coefficients as shown in Figure 50. z x z y P C C S C S C S C C C S C S 350 or Figure 50: Two beam model The results of test and FE analysis are also compared. Simulation also captures the behaviour of this composite girder, although it cannot, certainly, distinguish the influence of the different distance between two steel girders. The comparison of results indicate that even this simplified FE analysis can be used to estimate this kind of girder by approximation. 50

53 1.5.5 Behaviour of girders The load bearing behaviour of Preco-Beam as a flexural member is classified in 4 conditions due to Figure 51: 1) Composite dowel in compression zone 2) Composite dowel in tension 3) Composite dowel in cracked concrete under tension 4) Uplift in composite dowels due to bending 1) 2) σ c 3) f ct σ s Figure 51: Composite dowel action in different conditions The composite dowel under condition 1) in the region of negative bending moment and condition 3) in the region of positive bending moment is the classical cross-section with external reinforcement. In condition 1) the combination of compression stresses caused by the hogging moment additional to the local action of the dowel shear transmission force at the backside of the steel dowel has to be calculated for SLS. In the positive bending moment area the tension of global bending and the stresses of local bending action are to calculate for fatigue loads (FLS). Due to fatigue reasons the cross section 2) is favourably. Also for negative bending moments cracks caused by fatigue are not able to propagate in the compression zone of the steel. The uplift force in the composite dowel is usually zero. In the case of indirect loading the uplift derives for constant line loading to: P up J p J s v e x 51

54 P up J s f = deflection p p p Figure 52: Uplift force P up for indirect constant loading For a single load for which the load application point is not at the steel section the maximal uplifting force is calculated by P up 1 J F 8 J s v e c x In special cases e.g. steel sections with a large moment of inertia J s combined with a large point load introduction the uplift forces might influence the load bearing capacity of the composite dowel. h po c o c e x 45 c F F Figure 53: Uplift force P up for indirect point load 52

55 2 STATIC AND CYCLIC LABORATORY TESTS 2.1 General The complete test setup covers 65 Push-Out and Pull-Out tests according to the EC4. Figure 54: POST according to EC 4 The static bearing behaviour regarding shear forces is analysed by 50 tests; 15 tests are loaded with cyclic loads. Additionally to the Push Out and Pull Out tests 27 large scale beam tests are performed to investigate the behaviour of the Preco-Beam system under bending conditions. They are separated in 19 static tests and 9 cyclic tests. The used materials were a C70/85 for the concrete and a S460 for the steel. To ensure concrete failure C40/50 was used for some composite beams. Figure 55: Overview test setup At first, 9 cyclic tests and 9 static tests were executed and on basis of their results one of 3 shapes (CL, SA, PZ) were chosen for further tests. One static test of CL shape was conducted in addition with large number of strain gauges on steel dowels to confirm predictions of FEA concerning stress distribution. 53

56 CL shape, although its results were very promising, was excluded from further tests because of no existence of realistic technology if its production, SA shape was excluded because of the worst fatigue resistance due to notch behind the dowel and PZ shape was chosen to the further examinations as symmetric shape easy to produce by single cutting line and acceptable fatigue resistance. Following tables give an overview of the test series, their design and their objectives: Table 3 Overview POST 54

57 Table 4 Overview Beam tests Best shape Table 5 Criterion for best dowel shape In the frame of Preco-Beam the shapes shown in Table 5 has been investigated. At the beginning of the project it was decided to analyse PZ shape in detail in accordance to the considerations below. 55

58 Due to the experience gained during the project regarding bearing behaviour, fatigue and production methods now it seams to be necessary to advance the investigated dowel shapes. A dowel shape with only one cutting line to avoid notches by stopping the cutter has to be designed. A symmetric shape is favourable due to appearance of shear force in two directions caused by vehicle loads. Furthermore the high bearing capacity of the Puzzle shape should be combined with the good fatigue behaviour of the Clothoidal shape. Thus the following shapes seem to be possible: d = R at A R R A R on A-B B Figure 56 Advanced dowel shape -PZCL This kind of connector will provide a good resistance regarding detachment under the effect of uplift forces. When this resistance is not necessary, the distance d = 0.070m can e.g. be increased to prioritise the resistance to shear forces. To lay down complete rules for fatigue analysis, it is necessary to establish a predictive model of the lifespan of the connection detail that is subjected at the same time to slipping shear forces or uplift forces as well as compression or traction. Criteria of multiaxial fatigue may probably be useful for this purpose and the question will be studied in a next step. The tests and detailed FE investigations are done for the Puzzle shape, as the modified dowel shape was developed in accordance to the Preco-Beam results. 56

59 2.2 Test execution Test setup Push Out tests Figure 57: Specimen for POST The experimental tests are performed according to the test procedure of the EC4. The test procedure is named Standard Push Out Test Procedure (POST). A steel specimen is placed between two concrete slabs and the applied load against the relative displacement between steel and concrete in horizontal and vertical direction is measured. The displacement controlled test results in a dowel characteristic curve as shown in Figure 58 Displacement transducers Load-deformation characteristics Push-Out-Test installation Measurement equipment Figure 58 Test setup POST 57

60 Figure 59: NPOT (New Push-Out Test) Additional to the POSTs New Push Out Tests (NPOT) are performed (Figure 59). The Push Out Standard Test is modified that way that the steel part of the specimen which has to carry the shear load is in tension. The compressive force from the jack was put to the subsidiary steel part which welded to the bottom of the main steel part introduces tensile forces into it. Thanks to this operation main steel part is found in boundary conditions closely corresponding to those in real beams subjected to positive bending moment, where steel is in tension and concrete in compression Pull Out tests Next to the POSTs Pull-Out tests are executed. A steel specimen is placed in a concrete body. At the bottom of the concrete body a second steel plate is embedded. This plate is important for the anchorage. During the test the upper plate is displacement controlled pulled out of the concrete. Next to the load the relative displacement between steel and concrete is measured during the test. Figure 60: Pull Out Test Setup Beam tests Next to the POSTs large scale beam tests were performed. Beams are tested by loaded with two equal concentrated forces (4 point bending) according to the sketch below. Depending on the loading the beams vary from 3.6 up to 4.2 m of theoretical length. Test setup is presented in Figure

61 Figure 61: Test Setup Positive Bending Moment The beams are equipped with strain gauges on the steel girder and the reinforcement. Displacement transducers measure the slip between concrete slab and steel girder in vertical and longitudinal direction. For a few static tests (test series A and C) the load introduction was optimised to ensure concrete failure. Figure 62: Test Setup Positive Bending Moment Amended Figure 63: Test Setup Negative Bending Moment Composite beams with negative bending moments are tested to evaluate the behaviour of concrete dowels in continuous systems. These beams are loaded with one concentrated force (3 point bending) Load application Static loading 59

62 At first a load of 40% of the estimated ultimate load state has to be applied in two steps. Further 25 cycles between 5% and 40% of the estimated ultimate load has to be tested. For the static test the ultimate load has to be applied in at least 15 min. After reaching 100% the load can be measured down to 80%. The next figure shows the three stages: i = P i/p stat stage I stage II stage III slip [mm] Figure 64: Test procedure POST 1.2 P/P max [-] Idealisierte Dübelkennlinie 1.0 P max P max : Ultimate load state max P stat P Rk P stat : Static load state 0,8 0.4 uk 0.2 [mm] P Rk : Characteristic load state The following diagram shows the information which can be derived of a push-out test. Figure 65: Load-slip-behaviour Cyclic loads 60

63 Figure 66: Test procedure POST [Zapfe C., 2001] For the determination of fatigue resistance the load is applied as shown in the following figure in stages 1 and 2. P is the bearing capacity determined by FEA. The residual bearing capacity after cyclic loading may be of interest. If so the specimens are tested as shown in stage Evaluation of tests On basis of test series A (cyclic and static tests) puzzle shape (PZ) was chosen to perform on it more static and cyclic tests, including beam tests. POSTs are good method to conclude about ultimate resistance of connection while beam tests concerns in addition SLS behaviour Push Out Tests / Pull Out Tests Testing of shear transmission strip Each of test series was designed to allow for estimation of influence of one factor on resistance and ductility of shear connection. The experimental tests verify the application of composite dowels made of high strength materials. Therefore, for the steel a S460 and for the concrete a C70/85 is used. Test series A Within Test series A nine experimental static tests were performed. The tests were separated in three groups. The three groups differed in the implemented cutting line. The concrete grade C70/85 and the steel grade S460 (IPE 500) were constant. Further, the thickness of the steel web 10.2 mm and the dowel height 100 mm were constant for all tests. The results were used to identify the most effective cutting line. The specimens failed by a mixed failure mechanism (steel concrete). The following figure shows the cutting line and exemplarily three load characteristics curves of the performed tests. The SA shape shows the highest bearing capacity and the CL shape is the weakest but the most ductile connection. 61

64 700 Dowel characteristic curves 600 Dowel force [kn] SA-A PZ-A CL-A-3 Slip [mm] Figure 67: Force slip curves for A series For the cyclic tests is can be noted that the CL shape shows the best fatigue behaviour. For the PZ shape a crack occurred at the point where the radius of the dowel starts (see also chapter and chapter 2.1.1). The PZ shape was chosen for the further test series as already elucidated. Test series A* Test series A*consists of 3 POSTs, in which the steel part was made of IPE 500 (web thickness 10.2 mm), steel S355. The height and spacing of dowels are 100 mm and 300 mm respectively. The dowel reinforcement is 2 diameters of 12 mm. Such test series make possible to determine the effect of the steel grade on the resistance / ductility of the connection. The results can be compared with the results for series C (t w = 10 mm). The force slip curves are presented in Figure 68. Figure 68: Force slip curves for A*series Test series B Test series B consists of 3 POSTs, in which the steel part was made of IPE 500 (web thickness 10.2 mm), steel S460. The height and spacing of dowels are 50 mm and 150 mm respectively. The dowel reinforcement is 1 diameter of 12 mm. This test series are concerned with the dowel size, hence half-scaled dowels are used. Results can be compared with series C (t w = 10 mm). The force slip curves are presented below. 62

65 Figure 69: Force slip curves for B series There is the same relationship for both sizes of dowels up to characteristic load, thus dowel size does not affect ultimate capacity. Prk, s vrk, s e n (e spacing of dowels, n number of dowels). The ultimate resistance can be approximated to vult Ault tw f y.the dowel size influences ductility and, moreover, the part of the load-slip curve beyond P Rk. For small dowels an almost linear transformation can be assumed with view to the curve of dowels with reference dimensions. The coefficient is equal to the scale factor. Test series C Test series C was performed to determine the influence of the web thickness of the steel. Tests with tw = 10 / 17.4 / 30 mm were performed at TUWRO. Test series C consists of 9 POSTs subdivided into 3 groups. Each group was made from different steel profile. 3 specimens was made of IPE 500 (web thickness 10.2 mm), another 3 of HP305x126 (web thickness 17.4 mm), and 3 of HD400x382 (web thickness of 29.8 mm). S460 was used. The height and spacing of dowels are 100 mm and 300 mm respectively. The dowel reinforcement is 2 diameters of 12 mm P [kn] C3-tw10 2. C4-tw20 3. C7-tw30 s [mm] Figure 70: Force slip curves for C series (C1 3 tw10, C4 6 tw17, C7 9 tw30) The specimen collapsed because of concrete failure. Force slip curves are presented in Figure 70 (for legibility reasons only one curve per each web thickness is presented). It can be seen that the bearing capacity increases with higher web thickness but show a little less ductile behaviour. The results of test series C-tw10 can be compared with the results of the beam tests directly. Tests with t w = 17.4 mm should be used as reference for the test series E, G, H and J. As the test C-tw17.4 has been performed at TUWRO and series E, G, H and J have been performed at UBWM there are little 63

66 differences between the test setup and the concrete grade. Therefore a comparison between these tests has to be drawn carefully. Test series C contains additional 2 standard test specimens with a web thickness of 20 mm to have reference for series F. The steel parts were made of a HP320x147 (S460). A comparison of the load characteristic curves of the test series C and tests series A (t w = 10 mm) show higher stiffness and higher bearing capacity of the specimen of test series C as displayed in Figure 71. Dowel characteristic curves PZ 900 Dowel force [kn] PZ-C PZ-C PZ-A-3 Slip [mm] Figure 71: Dowel characteristic curves of additional specimen in test series C Test series D These test series concern the reinforcement of the dowel, in detail the number of rebar and its diameter. It consists of 4 POSTs. The steel part is out of HP320x147 (web thickness 20 mm), steel S460. The height and spacing of the dowels are 100 mm and 300 mm respectively. The dowel reinforcing is 2 diameters of 14 mm for specimens D1 2 and 1 diameter of 20 mm for D3 4. Results are compared with C series (t w = 20 mm) shown above. Compared with series C (2d10) a higher and more ductile behaviour can be seen for D1/D2 (2d14). Only one rebar increases the bearing capacity but decreases the ductile behaviour P [kn] D1-2d14 5. D2-2d14 6. D3-1d20 7. D4-1d20 s [mm] Figure 72: Force slip curves for D series Testing of geometrical influences The tests focus on the behavior of the concrete concerning the geometrical conditions. Test series E The test series E is concerned with the boundary conditions of a T-beam section. It consists of 6 test specimens with a T-beam concrete cross section. The specimens vary in the width of the concrete part 64

67 and in the implemented dowel reinforcement (E1 = 30 cm, E2 = 30 cm with add. reinforcement, E3 = 40 cm). As additional dowel curve specimen C-1 (bw = 100 cm, reinforcement like E1) can be considered. The steel parts were made of a HP305x126 (S460). The steel web thickness is 17.4 mm. The test specimen failed by a splitting of the concrete web. The performed tests show that the load bearing capacity is linked to the width of the concrete beam. The ultimate load bearing capacity increased with the concrete web width. The additional reinforcement in the concrete dowel lead to a more ductile behaviour after the ultimate load is reached (Figure 73). It has to be noted, that the concrete compression grade is much lower than for the other specimen. Dowel characteristic curves PZ-E 900 Dowel force [kn] PZ-E PZ-E PZ-E Slip [mm] PZ-C Dowel characteristic curves PZ-E2 Dowel force [kn] without additional reinforcement with additional reinforcement Slip [ mm] PZ-E1-2 PZ-E Figure 73: Dowel characteristic curves test series E Test series F In test series F the distance between the steel girders, if more than one steel girder is used in the composite beam, is analysed. The six test specimens of test series F are divided into three groups. The specimens consist of two steel specimens with different distances. Two specimens were manufactured with a distance of 35 cm; 2 specimens have a distance of 45 cm and 2 specimens were tested with a distance of 55 cm. The steel specimen were made of a HP320x147 (t = 20 mm S460). The used concrete was a C70/85. The dowel characteristic curves are presented in Figure 74. Dowel characteristic curves PZ-F Dowel characteristic curves PZ Dowel force [kn] F-2-35cm F-3-45cm 10 0 Slip F-6-55cm [ mm] Dowel force [kn] PZ-F PZ-F PZ-C Slip [mm] Figure 74: Dowel characteristic curves test series F 65

68 The diagram show that the bearing capacity for these tests is lower if the distance between the steel strips is bigger. Even series C with the same boundary conditions but only one steel strip has a higher bearing capacity than the specimens with two strips. This was not expected and it may be explained by different failure mechanisms. Figure 75 shows that in the case of test F1 and F2 the dowel strips are close together. The compression forces (red) of the dowel are connected and the remaining local tensional forces (blue) are carried by the reinforcement. The specimens fail by spalling similar to the reference tests from series C. In the specimens of tests F5 and F6 the compression forces get not connected as the specimen is too short. The forces are transferred from the specimen to the bearing at the bottom. Thus high tension forces remain in the middle of the concrete plate and the concrete fails, as it is not reinforced at these points. As a result the bearing capacity for the specimens with higher distance between the steel strips is lower than for the specimens with closer steel strips as for the specimens F5/F6 the not reinforced concrete part in the middle of the specimen is failing and for the specimens F1/F2 the reinforced dowels with higher bearing capacity are acting. As the failure may be explained by influences of the boundary conditions and the real failure (as for beams) cannot be activated the result should not be used for the evaluation of a design guide. For further investigations beam tests should be considered. Figure 75: Load process in the specimens of series F Test series G (Pull-out) The three test specimens were performed to evaluate the load behaviour of composite dowels under vertical loads. The steel parts consist of a HP305x126 in S460. The used concrete grade was a C70/85. The tests showed that the shear connection in vertical direction is very stiff. The slip between steel and concrete was only 0.2 mm at the maximum load level. The specimens failed by a splitting of the 66

69 concrete cross-section. The ultimate load in vertical direction is about 20% of the maximum bearing capacity in longitudinal direction. The dowel characteristics curves are presented in Figure 76. The conducted Pull-Out tests cast light on the stiffness of the connection and the load slip behaviour in vertical direction. The achieved ultimate load per dowel in vertical direction is about 20% of the ultimate load in longitudinal direction. So the concrete dowel fulfils the demand of the EC4, where an ultimate load of at least 10% of the longitudinal ultimate load is requested. Dowel characteristic curves PZ-G 250 Dowel force [kn] PZ-G-1 PZ-G-2 50 PZ-G-3 Slip [mm] 0 0 0,2 0,4 0,6 0,8 1 Figure 76: Dowel characteristic curves test series G (Pull out tests) Test series H Test series H is concerned with the influence of cracked concrete. It contains five push-out test specimens. Therefore, three specimens were tested statically and two specimens were tested with cyclic loading. The used steel cross section was a HP305x126. The concrete grade was C70/85. The cracks in the concrete have been generated by thin steel plates (t<0.5mm) which are placed in the concrete part of the dowel (one plate in each dowel). The dowel characteristics curves are presented in Figure 77. H1 and H2 are the static tested specimens. Compared to series C (not cracked concrete) the bearing capacity of the cracked specimens is lower. Nevertheless the compression strength of the concrete in specimen C is much higher. That is why interpretation of the results can be hardly done. It may be assumed that the influence of cracks to the bearing capacity of composite dowels is little. Dowel force [kn] Dowel characteristic curves PZ-H/C PZ-H1 PZ-H2 PZ-C1 PZ-H4 res PZ-H3 res Slip [mm] Slip [mm] Slip - Load cycles PZ-H PZ-H-3 PZ-H-4 PZ-H-5 N [-] Figure 77: Dowel characteristic curves test series H 67

70 The cyclic tests (H3, H4 and H5) regarding the influence of cracked concrete were performed to provoke concrete failure. Therefore, the upper load was set from 55% to 65% of the ultimate load. The stress range was limited to 50 kn and 100 kn in order to avoid steel failure. The load range was 400 kn load cycles are applied. Test H3 was cancelled after cycles and the residual bearing capacity was tested. It can be seen that for the first cyclic loaded specimen the same bearing capacity as for tests H1 and H2 can be reached. The specimens were opened and no steel failure could be identified. The slip progress can be explained by pulverising of the concrete at the contact zone. Test series J In continuation of the test series G pull-out tests in cracked concrete were also performed. As for series H the cracks in the concrete have been generated by thin steel plates (t<0.5mm) which are placed in the concrete part of the dowel (one plate in each dowel).the test series was separated in two parts. The first two tests were tested static tests and the next three tests were tested with cyclic loading. The ultimate bearing capacity of the static tests is ca. 90 % of the ultimate load of the specimens without cracks from series G. The slip between concrete and steel is marginally larger for specimens with cracked concrete than for specimens with not cracked systems. Dowel characteristic curves PZ-J 250 Dowel force [kn] PZ-J-2 50 PZ-J-1 0 Slip [mm] 0 0,2 0,4 0,6 0,8 1 Dowel characteristic curves PZ-G/J 250 Dowel force [kn] PZ-G-1 PZ-G-2 PZ-G-3 PZ-J-2 PZ-J-1 0 Slip [mm] 0 0,2 0,4 0,6 0,8 1 Figure 78: Force slip curves for J series The cyclic Pull Out tests were performed with a load range of +/- 25 kn (J4) respectively +/- 50 kn (J3, J5). The higher load range in test J3 and J5 lead to concrete failure. The concrete parts broke as it could be seen for the static tests and only less than load cycles could be reached. The stress rate has been limited in test J4 and 3.5 mio. load cylces can be applied. As conclusion it can be recommended that for composite dowels which are exposed to tension loads additional reinforcement which covers the concrete part below the steel dowel (like rebar number 6 in Figure 79 ) should be applied. 68

71 Figure 79: additional reinforcement for composite dowels exposed to fire Conclusions Uplift exists in POSTs hence it is prescribed to this kind of shear connection and is governed by kinematics of the moving parts (steel vs. concrete), what is directly connected to the shape of dowel root. The correlation of uplift and slip does not depend on the dowel s size but it depends on the shape only just as it was shown for the bearing capacity. The uplift force can be assumed to be a linear function of the slip: u = s tan (for PZ: tan = 0.37), see Figure 80. Figure 80: Interpretation of uplift C series vs. B series Finally Push-Out tests and Pull-Out tests with cracked concrete slabs were conducted. The tests resulted in the same load bearing capacity of the concrete dowels as the tests with not cracked slabs. This means, that the load bearing capacity is independent from the concrete slab condition. 69

72 2.3.2 Beam tests The tests consider effects on the concrete part (A D) and the behaviour of the steel girder (C1 C3N). The following table shows the complete test range. Table 6 Overview Beam Tests serie acro. no. of tests materials stat. cycl. C70/85 C40/50 S 460 IPE 500 subject Concrete A gi.sl.pos 2 3 x x x positive moment A* gi.sl.pos 1 x x x positive moment B gi.sl.neg 1 2 x x x negative moment C gi.sl.eypos 4 x x x distance / positive moment C* gi.sl.eypos 1 x x x distance / positive moment D gi.sl.eyneg 1 x x x distance / negative moment Steel C1 gi.be.ge 4 2 x x x positive moment C2 gi.be.dd 2 x x x dowel size / positive moment C3 gi.be.pos 2 1 x x x positive moment C3N gi.be.neg 1 x x x negative moment

73 T-Beams steel part 4 test series were designed. The cross sections of the beams are presented in Figure 81. Beams were made out of IPE 500 of steel S460 and concrete C70/85. Four different cross sections are tested series C1, C2, C3 and C3N: They vary by the dowel strip location with reference to neutral axis and height of concrete web (300 or 400 mm including slab thickness) Figure 81: Cross sections of beams, from the left side: series C1, C2, C3 and C3N For these test series the specimens are designed to obtain shear connection failure by steel dowel at ultimate load level. Series C1 and C2 make it possible to compare the influence of the concrete boundary condition to the resistance and stiffness of a Preco-Beam (C1 law concrete web, C2 concrete web up to bottom steel flange). In both cases the dowel strip is located near the neutral axis of the beam. Series C3 differs from the previous series regarding the location of the dowel strip: it is situated in the tensile zone. Series C3N was subjected to negative bending moment with the dowel strip in the girder s compression zone. The test setup is presented more detailed in the Annex. The first 2 of 6 beams series C1 are tested under cyclic loading. As a reference the next 2 of 6 beams are tested statically with a length of 3.6 m and the last 2 of 6 beams are performed to static tests with span of 4.2 m. Beams of series C2 (2 of 2) are tested under static loads. The beam of C3N is tested statically. 1 of 3 beams from series C3 is performed as cyclic test, 1 of 3 as static test with length 3.6 m, and 1 of 3 as static test with length of 4.2 m. During the tests procedure the vertical displacement is measured at 5 points (at supports to control supports deflection and at 3 points along beam length including mid-span). The slip between steel and concrete part is determined at 4 points (at both ends, at ¼ and ¾ of beam length) and the strain is determined by strain gauges (normal strain gauges, rosettes and chains of strain gauges 10 strain gauges within ~20 mm). The strain gauges are put in large number on steel dowels but also on concrete slab, reinforcing bars and steel flange and web. On each statically tested beam about 70 stain gauges are placed. In addition Crack Propagation Gauges are used in places where cracks are expected for beams due to cyclic loading; so at the root of the dowels Conclusions for steel part The cyclic loading is applied with frequency of 2.4 Hz. The load range is assumed to be 120 to 280 kn. That results in range of principal stresses of about 250 N/mm 2 for beams series C1 and 280 N/mm 2 for beam C3. This high level of stress range is chosen to ensure fatigue cracks in the steel part and to confirm the results of NPOT tests concerning the crack propagation. The main point is to prove that fatigue cracks which appear in shear connection can propagate through the web and they can even lead to sudden failure of the beam s bottom flange. In beams from series C1 the cracks propagate through the steel web, but after over cycles total failure has not been observed. In beam series C3 the crack 71

74 propagates through the steel flange causing its sudden failure. Crack patterns on all beams are presented on Figure 82. Figure 82: Pattern of fatigue cracks on steel parts of beams after cycling tests and failure of C3 beam These results of the cyclic beam tests confirm the predictions concerning crack propagation and at the worst sudden failure of the beam due to fatigue cracks in shear connection occurs. Hence because of the Fatigue Limit State (FLS) knowledge about the stress state and methods to predict the fatigue behaviour of the beam is of vital importance. The static beam tests are figured out to give information about the strain state in steel dowels and the strain distribution in the cross section. A large number of strain gauges confirm the predictions of the theoretical calculations (FEA) concerning the stress state and the flow of forces. Figure 83: Interpretation of uplift C series vs. B series 72

75 On the basis of particular strain gauges the comparison of C2 against C3 series it is stated, that uplift is not connected to the flow of forces in Preco-Beams as it was initially assumed. Uplift existing in beams and POSTs is prescribed to this kind of shear connection but it is only governed by kinematics of moving parts (steel vs. concrete), which is linked to the shape of dowel root. Thus the acting uplift force in composite dowels is theoretically zero for SLS conditions. It is the crucial point for the steel design concept. This way the steel design approach based on FEA and confirmed by tests could be applied to design of the steel part of Preco-Beams. This concept is based on geometric stress and it is to be applied to FLS and SLS. Moreover, static beam tests confirm the results of POSTs concerning ultimate resistance and ductility. This way all crucial points concerning ULS, SLS and FLS have been clarified by extensive test programme Slabs geometrical influences Additional to the POSTs 15 beam tests are performed. Therefore, 10 composite beams are loaded statically and 5 beams are tested with cyclic loads. The beams are made of IPE500 (S460) with a web thickness of 10 mm and concrete grade C40/50 to force concrete failure. The aims of the performed tests were: - Comparison of POST s and beam tests - Influence of the distance between the steel girder to the load bearing behaviour - Negative bending moment: Implementation of concrete dowels in continuous beams - Fatigue: Limit for the concrete pressure The beam tests are separated in 4 series (A - D). At first beam tests with only one steel girder are performed. Next composite beams with two steel girders with different distances were tested. An overview of the executed tests is presented in the following Figure 84. More details about the test and results and diagrams can be found in the (Preco-Beam Final Report, Background-Document A: WP3 Test series). Figure 84: Overview Beam Test Series concerning concrete conditions The static tests were designed to obtain concrete failure in the shear connection at ultimate load level. Test specimens were equipped with strain gauges on the reinforcement and on the steel girder. Further 73

76 displacement transducers were placed on the beams to receive information about the slip between steel and concrete Conclusions for geometrical influences The reference specimens of the beam tests series A show a good accordance to the results of the POSTs of test series A. This means that a transfer of the POST results to the beam tests is possible. The performed tests in series A contain also 3 fatigue tests. The influence of the mean stress and the stress range was investigated. The fatigue strength of the concrete part is subjected to the mean stress. The experimental tests show, that the concrete part will be destroyed if the upper load is too high. The stress range of the cyclic loading influences mainly the steel part. Thus, cracks in the dowel bottom occurred and propagated straight to the flange of the steel girder. The tests were performed at a frequency of 2 Hz. The upper load was set to 55 65% of the ultimate bearing capacity. Test series C was performed with 2 steel girders in the concrete slab to investigate the load bearing behaviour of the concrete dowels subjected to the distance between the steel girders. Therefore, tests with distances of 35 cm and 45 cm are performed. The tests show a reduced ultimate bearing capacity for a distance of 35 cm. The tests with a distance of 45 cm resulted nearly in the same ultimate bearing capacity as tests of tests series A. Thus the beam tests regarding the distance between two steel strips show different behaviour than the POST tests regarding the same subject did. In (Preco-Beam Final Report, Background-Document A: WP3 Test series) it can be seen, that pry out is the failure. If the distance between the strips is small the pry out cones interfere with each other which means that the loaded total concrete area is smaller than for the specimens with bigger distance between the dowel strips and thus lower bearing capacity occurs. Next to the tests with positive bending moment test series C and D are performed with negative bending moments. Here the concrete slab is in the tension zone. The tests are necessary to verify a possible application of concrete dowels in continuous beams. Tests with one and two steel girders are performed. Two test beams are loaded with cyclic loads additionally to the tests with static loads only. The behaviour of the 2 cyclic beam tests with negative bending moments is equal to the tests with positive bending moments. The fatigue strength of concrete part is subjected to the mean stress. Finally it can be stated, that a limit of 60% of the ultimate bearing capacity avoids concrete failure during cyclic loading. The fatigue strength of the concrete part is almost independent from the stress variation range. The stress variation range influences the fatigue strength of the steel part in the composite system. Finally, the extensive test programme was completed by FEA which allows a clarification of the behaviour of concrete dowels in systems with more than only one steel girder and the behaviour of concrete dowels if high strength materials are implemented. 74

77 3 PRECO-BEAM EXPOSED TO FIRE 3.1 General To enable the application of Preco-Beam system also for buildings it is necessary to increase the knowledge about the bearing behaviour of the beams under fire load. A design guide regarding the design of Preco-Beam in buildings exposed to fire has to be elaborated. Thus the following aims should be achieved: determination of the temperature distribution in different Preco-Beam typical cross sections by using FEA, taking the steel dowel into account defining simple methods to determine these temperature distributions to allow the designer to approach the fire resistance of Preco-Beams verification of the thermal calculations and the simple methods validity by fire tests investigations in the mechanical behaviour of the Preco-Beam under fire, particularly at the dowels level giving a design guide for the Preco-Beams and simple rules to calculate them under fire 3.2 FEA for fire loads Thermal analysis The thermal analysis was performed with the software SAFIR by using 2D and 3D Finite Elements. To simulate the fire action in our calculations, the following assumptions have been made: The thermal properties of the materials are taken from Eurocodes: EN and EN The parameters of the thermal actions are taken from EN The contact between adjacent materials (in this case: between concrete and steel) is assumed as perfect. The presence of the steel reinforcement is neglected for the thermal response. 2D thermal analyses were made to figure out the influence of different parameters on the thermal response of the Preco-Beams (Figure 85). The following parameters have been considered: the steel flange thickness tf ( mm) the steel web thickness tw ( mm) the embedded length of the steel web in the concrete h (50 to 500 by 50mm step mm) the steel flange width b ( mm 75

78 Figure 85: Examples of 2D thermal analysis 76

79 Diamond 2004 for SAFIR FIL E: S eidl5 NO DES : EL EME NTS : SOLIDS PL OT SILCONC_EN STEELEC3 SILCONC_EN Z X Y Figure 86: Example of 3D thermal analysis The Puzzle shape dowel has been analysed in a 3D model, see Figure 86. This model allows a higher precision locally in the steel and concrete dowels. In general, the results are very close those of the 2D model. It has been observed that the temperatures in the 2D-model are higher than in 3D-model. This result has to be linked to the high conductibility of steel. In 3D-model, there is a supplementary heat flow which goes from the steel web to the steel teeth. These differences between the 2D and 3D-model are very small, not more than 30 C. From this observation, it has been concluded that the 2D-model can be sufficient and on the safe side to evaluate the temperatures in the steel dowel. For the concrete dowel, the same conclusion can be made. The 2D-model gives temperatures very close to temperatures of the 3D-model, a little bit higher and thus on the safe side. Therefore, to obtain a good model to evaluate the temperatures reached in the dowels, it is recommended: for the steel dowel, to consider the 2D situation in the section situated at the middle of a steel tooth (section 1-1 in Figure 86) for the concrete dowel, to consider the 2D situation in the section situated in the middle of the distance between two steel teeth (section 2-2 in Figure 86) Mechanical analysis Some mechanical analyses have been performed with the nonlinear F.E. SAFIR software in order to define the test setup. These simulations are made with Bernoulli beams FE and take into account the increase of the temperature and the subsequent modifications of the material mechanical properties in the composite section. An example is given in Figure 87 and Figure

80 After comparison with the tests, the mechanical model gives very good results for the evaluation of the plastic resistance and the fire resistance under fire. Nevertheless, the model overestimates the stiffness of the Preco-Beams. Two main reasons can explain this result. For one part, the beams F.E. used in SAFIR don t take into account the deformation due to the vertical shear force. Secondly, in SAFIR, the connection between steel and concrete is assumed to be perfect without any slip between the two materials. Considering this result, we can conclude that the Preco-Beams tested in Liège had full shear connection (M Pl can be reached) but an incomplete interaction between steel and concrete (deformation due to steel-concrete slip). - Takes into account cracking of the concrete (f ct = 0) - No deformation due to shear force - Full connection : no slip between steel/concrete Mesh of the beam FE section: F0 F F0 Figure 87: SAFIR model of the beam 1_1 beam Total load (2 x Rk) [kn] SAFIR simulation TEST 1_1 (cold test) Displacement [mm] Figure 88: Vertical deflection calculated and measured for beam 1_1 78

81 beam SAFIR simulation Deflection (mm) TEST 1_2 (Fire test) Time (min) Figure 89: Vertical deflection calculated and measured for beam 1_2 3.3 Laboratory tests General The Table 7 summarises the tests performed in University of Liège. Because the bending resistance of a composite section is relatively well known contrary to the shear resistance, the share-out of the tests was changed at the 4 th meeting of the project in Munich. It has to be underlined that the total number of tests has not been changed. The new partitioning of the tests will allow us to test 2 supplementary systems for the shear resistance which is the most important question in Preco-Beams. The performed tests were: - 4 calibration tests in cold condition, similar to those tested under fire condition (loading until failure): 4 different systems - 8 fire tests under static and constant loads: - 2 fire tests for the bending resistance (2 different sections, tested in cold condition) - 6 fire tests for the shear resistance (6 different sections, 4 of which tested in cold condition and 2 of which test for the bending resistance) 79

82 1 2 IPE 500 Table 7 Sections of the 12 Preco-Beams tested at the University of Liège COLD TESTS (4m) FIRE BENDING TESTS (4m) FIRE SHEAR TESTS (2.4m) PrecoFire 1.1 PrecoFire 1.2 PrecoFire IPE IPE IPE 500 PrecoFire 2.1 PrecoFire IPE IPE 500 PrecoFire 3.1 PrecoFire 3.2 PrecoFire HEM HEM HEM 300 PrecoFire 4.1 PrecoFire IPE IPE 500 PrecoFire 5 PrecoFire HEM 300 For the sections for which a shear failure was desired, we used various systems to increase the bending resistance: additional reinforcement, intumescent paint on the steel profiles or ceramic fibres as thermal protection. 80

83 Ø20 mm Intumescent paint Figure 90: Systems placed to increase the bending resistance Finally 3 different types of Preco-Beams were tested: Type 1 : Type 2: Type 3: Figure 91: Typical types of Preco-Beams These sections are expected to be used in ordinary buildings and to be realistic sections financially viable. The idea of the first type of section is to use the steel profiles as support for prefabricated hollow core slabs on which a concrete slab would be cast. Their spans are supposed to be 6-8m in reality. The section type 2 is expected to be used in case of much longer spans (16-20m) for car parks or large floors for example. However, a fire protection could be necessary on the steel part which is not protected by the concrete slab. The section type 3 is close to the 1 st one if the hollow core slabs were neglected or close the 2 nd one if the steel web is protected by concrete. Nevertheless, this type 3 section is special because the shear connectors are situated under the level of the concrete slab. Table 7 (main text) summarizes the tests performed in University of Liège. The maximum span of the tested beams depended on the dimensions of the furnace i.e. 4 m. This maximum span has been used for the bending tests. For the shear resistance tests, the span has been limited to 2.4 m in order to enhance the probability of a shear failure rather than a failure by bending. 81

84 Cold tests and fire bending tests: Rk Rk 1,25 m 1,5 m 1,25 m 4 m 4,3 m Fire shear tests: Rk Rk 0,8 m 0,8 m 0,8 m 2,4 m 2,65 m Figure 92: Test setups With the 6 different sections, different geometric parameters have been tested. Two steel profiles have been used: IPE 500 and HEM 300, with different positions of the dowels. IPE 500 : HEM 300 : Figure 93: Positions of the steel dowel Therefore, the 5 parameters tested for the geometry are: flange thickness t f : 16 or 39 mm (IPE500 or HEM300) web thickness t w : 10,2 or 21 mm (IPE500 or HEM300) total height of the concrete slab web steel height encased in the concrete position of the dowel in the section (in the concrete slab or not) The measurements made during the tests were: The vertical displacements on the upper surface 82

85 The slip between steel and concrete at the ends of the specimens The failure load for the tests in cold condition The temperatures (in the furnace and in the cross section of the beam) Fire resistance time PZ shape dowel type A : scale 100 % of the recommended size (WP3) PZ shape dowel type B : scale 80 % of the recommended size (WP3) Figure 94: Dimensions of the dowels used in tests performed in ULg Figure 95: Measure of the slip between steel and concrete at the end of the beam (cold test on the left and fire test on the right) Figure 96: Measure of the vertical displacement of the beam (cold test on the left and fire test on the right) 83

86 3.3.2 Test execution The tests have been performed in 2009 in the Fire Testing Laboratory of the University of Liège. Table 8 Timetable of the Preco-Beams tested at the University of Liège COLD TESTS FIRE TESTS bending shear span 4 m span 4 m span 2.4 m series 21/01/ /04/ /05/2009 series 26/01/ /05/2009 series 15/01/ /05/ /06/2009 series 08/01/ /06/2009 series /06/2009 series /05/2009 The results of the tests are summarized in Table 9. The complete description of each test, the materials properties and the results are given in (Preco-Beam Final Report, Background-Document B: Documentation of tests WP4) It has to be noted that the connection load ratio is the ratio between the applied load and the resistant load. The resistant load is the minimum value between the ultimate shear resistance of steel dowels given by the Equation of the Preco-Beam mid-term report and the Design resistance to longitudinal shear calculated according to the method from EN based on potential surfaces of shear failure. 84

87 3.3.3 Evaluation of test results Table 9 Summary of the results of the tests performed in ULg COLD TESTS (span 4m) FIRE BENDING TESTS (span 4m) FIRE SHEAR TESTS (span 2.4m) PrecoFire 1.1 PrecoFire 1.2 PrecoFire 1.3 Test failure load [1] : 1246 kn Applied load [1] : 600 kn Applied load [1] : 800 kn Calculated failure load (bending failure) : 1231 kn ==> calculated load ratio: 49% ==> calculated load ratio: 41% Mode of failure: bending failure ==> tested load ratio: 48% ==> tested load ratio: - ==> connection load ratio: 70% ==> connection load ratio: 98% Method to increase the bending resistance: none Method to increase the bending resistance: intumescent paint Fire Resistance: 55 min Fire Resistance: 165 min Mode of failure: bending failure Mode of failure: bending failure T reached in the steel flange [2]: 760 C T reached in the steel flange [2]: 708 C T reached at the base of the steel dowel [3]: 102 C T reached at the base of the steel dowel [4]: 242 C PrecoFire 2.1 PrecoFire 2.2 Test failure load [1] : 1477 kn Applied load [1] : 1000 kn Calculated failure load (bending failure) : 1528 kn ==> calculated load ratio: 42% Mode of failure: bending failure ==> tested load ratio: - ==> connection load ratio: 120% Method to increase the bending resistance: intumescent paint Fire Resistance: 47 min Mode of failure: bending failure T reached in the steel flange [2]: 705 C T reached at the base of the steel dowel [4]: 240 C PrecoFire 3.1 PrecoFire 3.2 PrecoFire 3.3 Test failure load [1] : 1079 kn Applied load [1] : 360 kn Applied load [1] : 600 kn Calculated failure load (shear failure) : 1180 kn ==> calculated load ratio (shear failure) : 31% ==> calculated load ratio (shear failure) : 45% Calculated failure load (bending failure) : 2400 kn ==> tested load ratio: 33% ==> tested load ratio: - Mode of failure: vertical shear failure ==> connection load ratio: 86% ==> connection load ratio: 192% Method to increase the bending resistance: none Method to increase the bending resistance: 2 steel bars d20 Fire Resistance: 80 min Fire Resistance: 96 min Mode of failure: bending failure Mode of failure: bending failure T reached in the steel flange [2]: 861 C T reached in the steel flange [2]: 920 C T reached at the base of the steel dowel [3]: 556 C T reached at the base of the steel dowel [4]: 385 C PrecoFire 4.1 PrecoFire 4.2 Test failure load [1] : 821 kn Applied load [1] : 800 kn Calculated failure load (bending failure) : 778 kn ==> calculated load ratio: 63% Mode of failure: bending failure ==> tested load ratio: - ==> connection load ratio: 131% Method to increase the bending resistance: intumescent paint + ceramic fibres Fire Resistance: 94 min Mode of failure: bending failure T reached in the steel flange [5]: 109 C T reached at the base of the steel dowel [4]: 257 C PrecoFire 5 Applied load [1] : 1000 kn ==> calculated load ratio: 54% ==> tested load ratio: - ==> connection load ratio: 149% Method to increase the bending resistance: intumescent paint + ceramic fibres on flange Fire Resistance: 86 min Mode of failure: bending failure T reached in the steel flange [2]: 591 C T reached at the base of the steel dowel [4]: 425 C PrecoFire 6 Applied load [1] : 800 kn ==> calculated load ratio: 28% ==> tested load ratio: - ==> connection load ratio: 168% Method to increase the bending resistance: 2 steel bars d20 Fire Resistance: 70 min Mode of failure: bending failure T reached in the steel flange [2]: 770 C T reached at the base of the steel dowel [4]: 448 C Notes : [1] The given loads are the sum of the 2 concentrated loads applied on the beams [2] Temperature of the steel flange : thermocouple situated at mid-span [3] Temperature of the steel dowel : thermocouple situated at the base of the dowel at mid-span [4] Temperature of the steel dowel : thermocouple situated near the support [5] Temperature in protected steel flange at mid span but bending failure occured near supports where the steel was not protected 85

88 For these studied geometries, the results show that: No shear failure induced by fire was observed. If a problem occurs (series 3 cold test), the problem is the design in cold condition. The maximum temperature reached in the steel dowels is equal to 556 C. The slip does not increase significantly during the fire and tends to stabilize at the end of the tests. The thermal stresses, the thermal restraints and other thermal effects have not created unexpected problems. The behaviour of the Preco-Beams in fire condition was good and the failure was always ductile. There is a good correlation between temperatures calculated by SAFIR and temperatures measured during the tests. 86

89 4 DESIGN OF PRECO-BEAM 4.1 Structural design The use of concrete dowels as shear connectors is an upcoming solution in composite beams. They are characterized by a high initial stiffness, bearing capacity and ductility. With their use, new and economic constructions have been invented, e.g. the PreCoBeams. This PreCoBeam method is based on a rolled steel beam cut longitudinally, with a special shape, in two T-sections and a concrete top chord is concreted. The shape of the cut hereby allows for the shear transmission in the shear joint. In general pre-fabricated bridge elements are produced which are finalized on site. They are placed on the abutments and the residual superstructure is supplemented. This design guide introduces the static and fatigue design of this shear connections used for the PreCoBeam construction method. It contains the following information: - definition of the design criteria for steel and concrete failure, - loading of the shear connection, - ULS, SLS and FLS design of the steel dowel of the shear connection, - ULS, SLS and FLS design of the concrete dowel of the shear connection, - recommendations for the fire design, - final conclusions on the design of the PreCoBeam shear connection. To demonstrate the application of the design guide two design examples have been elaborated one for a building with fire design and one for a bridge with stress and fatigue design verifications. 4.2 Fire design The main conclusion relative to the results of the tests and therefore for the studied geometries is that by providing a good cold design, there is no problem of longitudinal shear in fire condition. Moreover, we have to note that for these tests, some systems have been used to artificially increase the bending resistance and thus increase the solicitation in shear. In the critical beam area where the bending moment is maximum, intumescent paint, ceramic fibres or additional reinforcement bars have been put in place. The aim of these protections was to increase the temperatures in the dowels at failure and to activate a shear failure induced by fire before a bending failure. The tests show that it was nevertheless impossible to reach a dowel temperature greater than 556 C and impossible to get a failure by shear. This is mainly due to the geometrical configuration of the Preco-Beam system as the position of the longitudinal shear connectors in the section is that way that the concrete offers a very efficient thermal protection. Therefore, in general and all the more without such special protections, this conclusion can certainly be extended to real Preco-Beam sections with common spans. However, without any other available result of fire tests that lead to failure by shear, a simple design recommendation is given to the designer to 87

90 limit the temperature of the steel dowel to the maximum temperature reached in these tests i.e. 500 C. This requirement will be easily met for most common Preco-Beam section types. In Appendix 1 a design guide has been defined for Preco-Beams exposed to fire with: Recommendations and guide to realise a good design of Preco-Beams Simple rules to calculate Preco-Beams submitted to fire. Concerning the thermal calculations, simple rules have been derived from thermal simulations with SAFIR to determine the temperatures in the Preco-Beam sections. These simple rules are based on tables for typical cross sections of Preco-Beams submitted to ISO fire curve and for different times: 30min, 60min, 90min and 120min. The tables depend on the flange and web thicknesses, the width of the flange and the height of the steel web. The knowledge of the temperature distribution in the Preco-Beam section allow to verify the mechanical behaviour of the dowels if it is necessary and allow to verify the mechanical response of the Preco-Beam to the external loads when it is submitted to fire. The fire tests confirm that Preco-Beams can be designed according to the rules of Eurocode EN

91 5 VALIDATION OF PRECO-BEAM SYSTEM IN PRACTICE To ensure the practical relevance of Preco-System Pilot Projects were planned in the frame of work package WP 6. The tasks were defined as follows: WP 6.1: Building for infrastructure WP 6.2: Bridge design WP 6.3: Noise protection system As mentioned above already at the beginning of the project the noise protection system turned out to be not suitable for Preco-Beam. Therefore WP 6.2 was enlarged by bridges with different boundary conditions. Below one Road Bridge and one Railway Bridge are highlighted as Pilot Projects. Due to the short project time not all pilot projects could be completed. In some projects it was difficult to convince the client to apply the new construction method before the research regarding the bearing behaviour and especially the fatigue behaviour has been finished. Nevertheless several pilot projects were solved out. Figure 97 and Figure 98 give an overview of all pilot projects which could be acquired during the project period. Some of them are already finished or will be finished soon. For some project it cannot be ensured that they will get realised because the clients did not assure the project yet. Although the preliminary studies are elaborated and the design of the projects in practice and thereby the application to the practice is already finished. It can be noted that the dowel shape is modified in most projects in comparison to the three initial shapes. This results from the further investigation of the dowel shapes during the project period regarding bearing behaviour and fatigue behaviour. As the conclusion the dowel shape used for Hilfsbrücken in Germany, Railway Bridge in Poland and Avre Bridge in France may be used as best shape for composite dowels. Figure 97 Overview pilot projects (buildings and railway bridges) 89

92 90

93 91

94 Figure 98 Overview pilot projects (road bridges) Conclusions Production of composite dowels On the base of the production of the steel specimens for the experimental tests and the girders for the first Pilot Project experience could be gained regarding the significant influence of the production process to the fatigue resistance of the steel part of the composite dowel. It is of importance not to introduce too much heat into the steel girder. Therefore stopping of the cutter should be avoided. This may also happen in case of cutting lines with two cuts as it was necessary for the first edition of the clothoidal shape. Cutting lines with one cut only and uniform cutting speed and oxygene introduction are recommended. For the cutting process it should be minded that the dowel bottom and the dowel root should not have notches and rough surfaces have to be avoided. Design of the dowel shape The design of the dowel shape affects the bearing capacity of the composite dowel as well as its fatigue resistance. To obtain optimal bearing capacity the dowel should have a high moment of inertia. The size of the dowel itself has not much point. It can be determined a best ratio dowel height / dowel length which influences the bearing capacity. Nevertheless in case of cyclic loads the fatigue resistance is of vital importance, too. Thus notches as used for the SA shape within the frame of this project should be avoided. Initially it was not expected that the influence of notches is that high. Even for the PZ shape the weak point regarding fatigue at the dowel bottom where the radius for the dowel starts could be determined. To gain wide support it would be beneficial to improve investigations in fatigue behaviour. Symmetric shapes are advantageous as they are independent from the direction of the shear forces. For constructions under variable loads this may be of importance. The bearing capacity of a composite dowel can be determined by using the Design Guide elaborated in the frame of the Preco-Beam project. Fire resistance of Preco-Beams On the base of the fire tests performed in the scope of the Preco-Beam project it can be stated that fire does not cause lower shear transmission of the composite dowel nor does it influence the slip. Ductile 92

95 behaviour of the girder can be noted. Conventional fire protection, as known from composite structures, can be used. Application area The application of Preco-Beam in practice has already started in the frame of some first Pilot Projects. It can be stated that for buildings Preco-Beam is particular beneficial for spans about m and for high loaded buildings, as it is a slender and lightweight construction method. For high-loaded midspan bridges the advantages of Preco-Beam compared to conventional composite structures are the minimisation of production effort and the decreasing of construction time along with high quality standards and high bearing capacity. In a next step the application to multi-span girders could be focused. Exploitation and impact of the research results Thanks to the work in the frame of the Preco-Beam project now it is possible to design composite girders using composite dowels in an efficient way. Not only was the local bearing behaviour of the shear connector investigated but also the global bearing behaviour of the composite girder considering geometrical influences, fire loads and fatigue behaviour. Theoretical research and the elaboration of realistic FEA models now allow studding local and global influences economically. By theoretical and experimental tests knowledge about composite dowels could be extended. Now the interaction of several dowel-strips and the influence of the concrete slab s width can be determined. For application in building design of Preco-Beam exposed to fire is possible and in particular the fatigue resistance of the steel part of the dowel has been evaluated. Profound knowledge about fatigue of composite dowels has been investigated and transfer to requirements for fatigue design. As a result of all the investigations a Design Guide has been elaborated containing not only design rules for the dowel but also recommendations regarding reinforcement and quality assurance. Altogether it can be stated that the work done in the scope of the Preco-Beam project is highly linked to applications in practice. Thus numerous projects for Preco-Beam in buildings and bridges have been elaborated and various clients have been introduced into the Preco-Beam construction method and a strong interest could be aroused. First constructions already use Preco-Beams which is beneficial for the acceptance of a new construction method as these projects act as references. 93

96

97 LIST OF FIGURES AND TABLES Figure 1 Project overview 6 Figure 2 Influences of the cutting process 13 Figure 3 HE1000x438 a) before cutting, b) cutting process, c) after separation 15 Figure 4 FE model and load by strains*10e-3 15 Figure 5 FE study of cutting process deformation under eigenstresses modelised by strains in rolled girder before and after separation. 16 Figure 6 Normal stress σ1 and deformation before and after cutting (blue = tension, red = compr.) 16 Figure 7 Stress values at the end of beam: before separation: a) σx, b) σy and after separation: c) σx, d) σy 16 Figure 8 Beginning of cutting: failure of steel plate due to high σy value 17 Figure 9 Model of the beam: a) general view, b) finite elements at midspan region 17 Figure 10 Reduced stress: a) step 1 beam under dead load, b) step 2 force applied: yielding of web is visible, c) step 3 stage after force removed: new residual stress layout is visible 18 Figure 11 Plastic strains at longitudinal direction: a) step 2, b) step 3 (midspan region) 19 Figure 12 Examples of regular composite sections with ensured fire resistance [Merkblatt 117, 1991] 20 Figure 13 Preco-Beam sections 21 Figure 14 Different types of floor slabs 21 Figure 15 HD/F elements / TT/F precast elements 23 Figure 16 Preco-Beam cross-section used in the frame of the study 23 Figure 17 Load / span curves for Preco-Beam elements (ULS) and precast HD/F elements / Preco-Beam elements and TT/F elements. 24 Figure 18 Feasible cross-sections for bridges 25 95

98 Figure 19 Oxygen cutting of the rolled beam section / Coated steel beams 26 Figure 20 Installation of reinforcement 26 Figure 21 Girder and reinforcement are lifted into the formwork / Concreting of the prefabricated element 26 Figure 22 Strainless support / Figure 23 Detail stiff-joined abutment 27 Figure 24 Dimensions of Preco-Beam girders for Vigaun-Bridge and of preliminary design for VFT and prestressed girders 28 Figure 25 Composite dowel nomenclature 30 Figure 26 Composite dowel, a composition of steel dowel (1) and concrete dowel (2) and their influence of load bearing behaviour 31 Figure 27: Possible failure modes in load range 2 32 Figure 28: Load proportions of a typified load-slip diagram (w/o bond) to be supposed in ductile concrete failure: Figure 29 Geometry of push-out (3P1) model: a) mesh, b) components 35 Figure 30 Push-out model results: a) Mises stress, b) u2 displacements, c) additional plate which enables modelling of contact support, d) Mises stress (yielded steel) for 3P4 model (with high number of elements) 35 Figure 31 The 1D1-model: a) boundary conditions, b) parts and interactions 36 Figure 32 Shapes 37 Figure 33 Results of searching of optimal ratio for SN shape and CL shape 38 Figure 34 Principal stress distribution in steel dowel along arc length from actions: σ N ( G), P (L) and P up (U) Figure 35 Beam model for WP3 tests (a: steel part, b: concrete part, c: FE model of the beam) and d: appropriate model for determination of G and L actions. 39 Figure 36 Manufacturing defect can appear in the detail of the cutting line of a puzzle form Preco-Beam connector Figure 37 Bridges fatigue strength details for direct stress ranges 41 96

99 Figure 38 Introduced loads for fatigue analysis 41 Figure 39 Woehler curve 42 Figure 40 Simple plate with longitudinal stresses (average damage = E-6 ) / Slope 1:4 / Vertical butt weld with short attachment of 50 mm 43 Figure 41 FE models for fatigue analysis for Preco-Beam shapes 44 Figure 42 Woehler for Preco-Beam details 45 Figure 43 ADINA model Drucker 46 Figure 44: FE models in comparison to the POST (A: series C, B: series E, C: series F) 47 Figure 45: Stress mapping 48 Figure 46: Influence of T-Beam width 48 Figure 47: Influence of displacement 49 Figure 48: Beam model 49 Figure 49: Comparison between experimental Test and FEA 50 Figure 50: Two beam model 50 Figure 51: Composite dowel action in different conditions 51 Figure 52: Uplift force P up for indirect constant loading 52 Figure 53: Uplift force P up for indirect point load 52 Figure 54: POST according to EC 4 53 Figure 55: Overview test setup 53 Figure 56 Advanced dowel shape -PZCL 56 Figure 57: Specimen for POST 57 97

100 Figure 58 Test setup POST 57 Figure 59: NPOT (New Push-Out Test) 58 Figure 60: Pull Out Test Setup 58 Figure 61: Test Setup Positive Bending Moment 59 Figure 62: Test Setup Positive Bending Moment Amended 59 Figure 63: Test Setup Negative Bending Moment 59 Figure 64: Test procedure POST 60 Figure 65: Load-slip-behaviour 60 Figure 66: Test procedure POST [Zapfe C., 2001] 61 Figure 67: Force slip curves for A series 62 Figure 68: Force slip curves for A*series 62 Figure 69: Force slip curves for B series 63 Figure 70: Force slip curves for C series (C1 3 tw10, C4 6 tw17, C7 9 tw30) 63 Figure 71: Dowel characteristic curves of additional specimen in test series C 64 Figure 72: Force slip curves for D series 64 Figure 73: Dowel characteristic curves test series E 65 Figure 74: Dowel characteristic curves test series F 65 Figure 75: Load process in the specimens of series F 66 Figure 76: Dowel characteristic curves test series G (Pull out tests) 67 Figure 77: Dowel characteristic curves test series H 67 98

101 Figure 78: Force slip curves for J series 68 Figure 79: additional reinforcement for composite dowels exposed to fire 69 Figure 80: Interpretation of uplift C series vs. B series 69 Figure 81: Cross sections of beams, from the left side: series C1, C2, C3 and C3N 71 Figure 82: Pattern of fatigue cracks on steel parts of beams after cycling tests and failure of C3 beam 72 Figure 83: Interpretation of uplift C series vs. B series 72 Figure 84: Overview Beam Test Series concerning concrete conditions 73 Figure 85: Examples of 2D thermal analysis 76 Figure 86: Example of 3D thermal analysis 77 Figure 87: SAFIR model of the beam 1_1 78 Figure 88: Vertical deflection calculated and measured for beam 1_1 78 Figure 89: Vertical deflection calculated and measured for beam 1_2 79 Figure 90: Systems placed to increase the bending resistance 81 Figure 91: Typical types of Preco-Beams 81 Figure 92: Test setups 82 Figure 93: Positions of the steel dowel 82 Figure 94: Dimensions of the dowels used in tests performed in ULg 83 Figure 95: Measure of the slip between steel and concrete at the end of the beam (cold test on the left and fire test on the right) 83 Figure 96: Measure of the vertical displacement of the beam (cold test on the left and fire test on the right) 83 Figure 97 Overview pilot projects (buildings and railway bridges) 89 99

102 Figure 98 Overview pilot projects (road bridges) 92 Table 1 Compare of different methods with the span length 16 m 22 Table 2 Compare of different methods with the span length 20 m 22 Table 3 Overview POST 54 Table 4 Overview Beam tests 55 Table 5 Criterion for best dowel shape 55 Table 6 Overview Beam Tests 70 Table 7 Sections of the 12 Preco-Beams tested at the University of Liège 80 Table 8 Timetable of the Preco-Beams tested at the University of Liège 84 Table 9 Summary of the results of the tests performed in ULg

103 List of references Andrä H.-P. Economic Shear Connectors with High Fatigue Strength, IABSE Symposium 1990 Collin Peter and Johansson Bert Wettbewerbsfähige Brücken in Verbundbauweise [Article] // Stahlbau Heft pp EN Eurocode 3: Design of steel structures - Part 1-9: Fatigue. - Brussels : European Committee for Standardization (CEN), EN Eurocode 4: Design of composite steel and concrete structures - Part 1-2: General rules - Structural fire design. - Brussels : European Committee for Standardization (CEN), Fink J., Petraschek T. Neubau der Straßenbrücke bei Vigaun. Versuchsbericht TU Wien im Auftrag der Österreichischen Bundesbahnen, Linz und Schmitt Stumpf Frühauf und Partner, München; Wien Oktober 2006 Hanswille G. Die Bemessung von Stahlverbundstützen nach nationalen und EU-Regeln, Der Prüfingenieur 22/2003 Hechler O., Feldmann M., Rauscher S., Hegger J. Untersuchungen zum Trag- und Verformungsverhalten von Verbundmitteln unter ruhender und nicht ruhender Belastung bei Verwendung hochfester Werkstoffe; AiF-Forschungsvorhaben Nr , RWTH Aachen, Forschungsbericht P621 FOSTA, 2007 Hechler O, Lorenc W, Berthellemy J., Seidl G, Viefhues E. Continuous shear connectors in bridge construction. Composite Construction VI Conference, Colorado, USA, Hobbacher A. Recommendations for fatigue design of welded joints and components, International Institute of welding, IIW Joint Working GroupXII-XV, Document XIII / XV , 2005 Lorenc W., Ignatowicz R., Kubica E., Seidl G. Numerical model of shear connection by concrete dowels. Recent Developments in Structural Engineering Mechanics and Computation, Millpress 2007, Rotterdam, Netherlands. Merkblatt 117 Brandschutztechnische Konstruktion und Bemessung von Stahlverbundbauteilen, Stahl- Informations-Zentrum, Düsseldorf 1991 Model Code 1990: Design Code, Publisher: Telford, Ceb-Fip, ISBN: , 1993 Nilsson G. Ultimate resistance of PRECO composite beams, TRITA-BKN Master thesis 256, Division of Steel Structures KTH,

104 Preco-Beam Final Report, Background-Document A: WP 3 Test Series, not published Preco-Beam Final Report, Background-Document B: Documentation of tests WP 4, not published Preco-Beam Final Report, Background-Document C: Detailed project design Hornsberg, not published Preco-Beam Final Report, Background-Document D: Detailed project design Vigaun, not published Schaumann P., Upmeyer J. Neue Nachweistabellen zum Feuerwiderstand kammerbetonierter Verbundstützen, Stahlbau 71, Heft 5, 2002 Schmitt, V., et al. VFT-Bauweise, Entwicklung von Verbundfertigteilträgern im Brückenbau, Betonund Stahlbetonbau 96, 2001, Heft 4 Seidl, G. Behaviour and load bearing capacity of composite dowels in steel-concrete composite girders, PhD thesis, Wroclaw University of Technology, Raport serii PRE nr 4/2009 Sonsino C.M., Kaufmann H., Müller F., Berghöfer U. Schwingfestigkeit von hochfesten Feinkornbaustählen im brenngeschnittenen Zustand, Forschungsbericht P185 FOSTA, 1992 Veríssimo, G.S., Paes, J.L.R., Valente, I., Cruz, P.J.S., Fakury, R.H. Design and experimental analysis of a new shear connector for steel and concrete compositestructures, Third International Conference on Bridge Maintenance, Safety and Management, July, Porto, Portugal (in CD-Rom), Texto completo (2006). Wurzer O. Capacity of concrete dowels Zur Tragfähigkeit von Betondübeln, PhD, Universität der Bundeswehr, München, 1998 Zapfe C. Capacity and deformation of composite girders with concrete dowels for the transmission of shear forces Trag- und Verformungsverhalten von Verbundträgern mit Betondübeln zur Übertragung der Längsschubkräfte, PhD, Universität der Bundeswehr, München 2001 Zulassungsbescheid Z Perfobond-Leiste, Institut für Bautechnik, Berlin 1991 Zulassungsbescheid Z Kombi-Verdübelung, Deutsches Institut für Bautechnik, Berlin

105 APPENDICES Appendix 1: Design guide distributed in the frame of Preco-Beam Table of content 1 INTRODUCTION FAILURE CRITERIA OF THE PRECOBEAM SHEAR CONNECTION ANALYSIS OF LOADING IN THE DOWEL CONTACT AREA DESIGN GUIDANCE STEEL ANALYTIC APPROACH TO THE ULS DESIGN ANALYTIC APPROACH TO THE SLS DESIGN ANALYTIC APPROACH TO THE FLS DESIGN Fatigue resistance of the gas cut edge Fatigue design of the steel web General design methode Design for Puzzle shape Design check ALTERNATIV DESIGN METHOD BASED ON FE ANALYSIS DESIGN GUIDANCE - CONCRETE EMPIRICAL ULS BEARING CAPACITY FOR THE CONCRETE DOWELS Shape factor s T-Beam cross section factor t f Interaction of parallel steel dowel strips Comprehension of Design Concept Concrete (UBWM) Limits of design criterium Fatigue behavior FLS ANALYTICAL DESIGN FOR THE SURROUNDING MATRIX OF THE CONCRETE DOWEL Shear in the concrete dowel Pry-out of concrete cover Increase of pry-out capacity by guiding of reinforcement bars Decreasing capacity in tight laying composite dowels Spalling of concrete cover Comprehension of the Analytical Design Concept Concrete FIRE GENERAL PRINCIPLE CONCEPTUAL DESIGN Examples of conceptual design for different types of Preco-Beam section CALCULATION RULES Thermal calculations Type 1 section: Type 2 section: Type 3 section: Mechanical calculations Bending resistance Type 1 section: Type 2 section: Type 3 section: Connection Longitudinal shear resistance Vertical shear resistance

106 Web buckling under fire DESIGN EXAMPLES PRECOBEAM FOR A ROAD BRIDGE Concrete dowel properties SLS Steel dowel Fatigue design DESIGN OF CONCRETE DOWEL Shear of reinforced concrete dowel Pry-out of concrete cover Spalling of concrete cover Decisive criterion Example for reinforcing the composite dowel FIRE Example Example 2: Simple calculation model: Advanced calculation model (FE SAFIR software): Mechanical calculation: LIST OF FIGURES AND TABLES FORMULA SYMBOLS LIST OF REFERENCES

107 1 INTRODUCTION Generally the behaviour of the PreCoBeam shear connection can be divided in two counterparts of the composite action a steel dowel and a reinforced concrete dowel. The force flow interacts between the steel dowel (1) and the concrete dowel (2). The bearing capacity of this shear connection is limited by steel or concrete failure. Both failures modes can influence this other failure mode. In a adequate design the criteria for each failure are balanced up to the design load. General definitions are shown in Fig. 1. 1) Steel dowel 2) Concrete Dowel 3) Reinforcement of concrete dowel 4) Base of dowel 5) Core of dowel 6) Toe of dowel 7) Top of dowel 8) Reinf. of concrete Fig. 1:Terminology of a composite dowel 1.1 Failure criteria of the PreCoBeam shear connection The steel dowel is subjected to longitudinal shear in the composite joint causing bending and shear stresses in the steel. Steel failure is limited in the ultimate limit state by a) the shear resistance, b) yielding due to bending of the connector and in the fatigue limit state by c) fatigue cracks due to dynamic loading, see Fig. 2. Interaction between a) shear and b) bending has further to be taken into account. 105

108 a) b) c) Fig. 2: Failure modes for steel dowel (here puzzle shape) Concrete failure is characterized by several failure modes. Which mode finally occurs depends on the boundary conditions like geometry, concrete grade, reinforcement design, adding of fibers etc. In case the steel dowel is significantly stiffer than the concrete dowel, failure occurs due to cracks and damage of the concrete matrix. A typical evolution of concrete failure is proposed in Fig. 3 with the following steps of failure mode activation: A: Adhesion/Friction of contact interfaces, B: Compression of steel-concrete surfaces, C: Shear in the concrete interfaces, D: Dowel action of transversal reinforcement, E: Block effect of twin rebars in the concrete dowel. 106

109 Fig. 3: Steps of failure mode activation for a typified load-slip diagram to be supposed in ductile concrete failure [i] An effectively reinforced and adequately embedded concrete dowel shows ductile behaviour beyond ultimate capacity. In this state, the crack formation spreads out in the concrete and possibly in steel too. A corresponding redistribution of the dowel forces occurs from a confined compression state into a strut-and-tie member in the concrete. It is decisive for the bearing capacity to carry the resulting tensile forces in the concrete. When a crack occurs the tension forces are carried by the reinforcement bar which penetrates the crack. Without this reinforcement the dowel has a brittle failure. After the adhesion is unbounded the ultimate bearing capacity of the concrete dowel is at first influenced by pulverization of the local compression in the contact area. The concrete in the slot is split in two zones, zone I and II. In Zone I an almost hydrostatic stress state is present. Zone II is the area of the load diffusion. The ultimate bearing capacity is limited by the three axial concrete pressures in Zone I. There vertical to the pressure tensile occur as displayed in the Fig. 4a. Corresponding to an increasing slip failure criterion C was observed for dowels with recesses of large diameters, deeply placed concrete dowels in the concrete slab or if thicker steel plates were used. A double cut shearing off within the concrete slab at the steel slots occurs (Fig. 4b). 107

110 a) Local pressure b) Shearing of concrete dowel Fig. 4: ULS failure mechanism of concrete [ii] However these two failure criteria lead to the ultimate load resistance only if no second failure mechanism appears failures of the surrounding concrete matrix. Therefore the design of the surrounding concrete matrix is essential for the design of the concrete dowel and their resulting verifications are in addition considered as design checks for the shear connection. Depending on the dowel geometry, material qualities and the reinforcement design different failure mechanisms of the surrounding concrete matrix are occurring. 1.2 Analysis of loading in the dowel contact area To define the action in the surface between concrete and steel, a loading test has been carried out with a large number of strain gauges at the steel part of the connector. In addition a numerical model was analysed and calibrated to the test results. In a next step, the numerical model has been modified for various shapes. The result for the PZ shape taken from the numerical model is shown in Fig. 5. From this analysis an analytic load model for the local behaviour of a shear connector has been derived. Here the puzzle geometry has been focused. However it is possible to transfer the approach to any geometry for each inventor of a new shape. The resulting simplified stress distribution for the local approach is given in Fig. 6 (no transversal reinforcement) and Fig. 11 (transversal reinforcement). 108

111 Fig. 5: Stress distribution in the contact surface of the steel profile The local approach is based on the load introduction on a single tooth. Hereby S represents the center of the projection area A p in shearing direction; h s is the distance from the center to the base of the shear connector. In Fig. 6 for example, as projection area only the area constricting a concrete block on front of the shear connector should be considered. As orientation h s has calculated with FEA to be about 0.25 h d for this particular PZ shape in beams. It has been obtained by dividing bending moment at dowel base (normal stress integration) by resulting shear force (shear stress integration). For simplification it can be considered that the position of the resulting horizontal force P is at the height h d /3 from the bottom of the cut. The force on each steel tooth is composed by the shear force in the composite joint P τ and the stress distribution due to the global loading depending on the geometry of the composite cross section. P up for the uplifting forces due to the location of the shear joint in respect to the neutral axis of the cross section (occurs only, if no transverse reinforcement is added) and σ ne for the notching effect from the nominal stress of the steel section (Fig. 7). h d + Fig. 6: Geometry of puzzle tooth (PZ) P and P up σ ne Fig. 7: Forces for the steel design of the connector (PZ shape) with missing stirrup in the concrete dowel 109

112 For the determination of P it is conservatively assumed that the load distribution along the height of the connector is constant and P is located at h s. The uplifting force P up is resulting from the eccentricity h of the shear joint to the centre of the composite compression chord (Fig. 8). The trajectory generates an uplifting force on the steel part of the connector which would be pushed out of the concrete section if the shape of the steel part doesn t contain any undercut. Hence the undercut of the steel part of the connector implements two functions; first, it generates the 3D stress state for the kernel of the concrete dowel and second, it locks the shear connection against uplift in vertical direction. Fig. 8: Uplifting force due to eccentricity of the shear joint to the neutral axis at support Fig. 9: Determination of h i for ULS and SLS The determination of the uplifting force is based on h i depending on the stress distribution of the composite section (Fig. 9). In the following full shear connection is assumed. It is required to differentiate between the ULS and the SLS respectively the fatigue design. Further the construction stages have to be considered. Generalised P up is therefore calculated according to Fig. 8 as: P up h' P [kn] Eq. 1 e x with h < e x else P up = P. For shear connectors with transversal reinforcement the load introduction has however to be modified. The uplift force are mostly taken over by the reinforcement in current locations (general case) (Fig. 10). If an external force (e.g. a pre-stressing anchorage at the end of the beam, or an effect of curvature in 110

113 elevation) is applied the analysis of the uplift forces is required and special anti-uplift devices may be required. Consequently also the determination of P needs to be modified, as the load is first introduced in the reinforcement bar and subsequently transmitted via the concrete into the steel part of the connector (Fig. 11). Fig. 10: Force distribution in the concrete with transversal reinforcement Fig. 11: Geometry of puzzle tooth (PZ) with transverse reinforcement P Fig. 12: Forces on the steel part of the connector (PZ shape) The maximum spreading angle in the concrete is hereby conservatively assumed to be 45 which leads to a decrease in activated projection area of the connector A p with a corrected height h s,eff to account for the P acting in the global longitudinal direction. Thus a simplified resultant force P located at h s is conservatively assumed. The σ ne is the notching effect on the normal stresses in the web of the steel girder due to the shape of connector. The increase in stress hereby depends on the geometry of the connector. A cutting line with a constant geometric change with a defined radius has been scrutinized by FEA (Fig. 13). From the analysis it could be concluded, that σ ne depends directly on the ratio of the connector length to the radius of the cut out (b 1 / R), but not on length and radius separately. Moreover, the height h d of the connector is unimportant for the notching effect (for logical proportions of the cutting shape). σ ne 111

114 h d Fig. 13: FE model for analysis Fig. 14: Stress distribution due to notch effects On the basis of an extensive parametric study using the FE method the notching stresses have been derived to ne kc N [MPa] Eq. 2 with the notch factor k C b 1 b R 1 R 2 [ - ] Eq. 3 The higher the ratio b 1 / R of the tooth, the higher is the notching effect expressed by the factor k C. Thus, not only the sharpness of the notch itself but also the increase in stiffness depending on the length of the connector is influencing the notch effect, which is in accordance to the effect of longitudinal stiffeners. For more complex geometries, e.g. the clothoidal shape with a varying radius, the notch effect has to be directly determined by the use of FEA. Extensive work has been carried out using the program Code Aster [iii] and compared to details classified already in the Eurocode. For the selected geometries the numerical values are given and compared to the analytic values in Tab. 1. The values reflect the stress concentration in the most critical area of the cutting line (along the arc). Geometry Radius R Length b 2 k C Shark fin 2,50 Puzzle 27mm 121mm 1,80 Clothoide variable ~50mm 1,54 Tab. 1: Notch factor β N for selected geometries 112

115 2 DESIGN GUIDANCE STEEL In the following chapters an analytical approach to design the steel dowel in the ULS, SLS and FLS and an approach based on FEA are presented. In Tab. 2 an overview on the required design checks for steel in the limit states is given. Design check ULS SLS (bridges) FLS Load model LM 1 & LM 2 LM 1 & LM 2 LM 3 Resistance plastic elastic elastic Stresses total force Von Mises Principle stresses Check P max P d σ max f yd Δσ = σ max - σ min D(Δσ) 1 Tab. 2: Design of PreCoBeam shear connection Steel design 2.1 Analytic approach to the ULS design Test specimens with steel failure have been opened and cracks in the steel strip have been observed, whereas the concrete matrix has not been significantly damaged (Fig. 15). In reference to this failure mode a steel failure criterion has been derived to limit excessive yielding. The basis of design is, that the resulting maximum equivalent von-mises stresses do not exceed the yield strength. For transversally non-reinforced PreCoBeam bridge cross sections, this formula has therefore to cover the additional uplifting forces from the global geometry of the cross sections (Fig. 16). Consequently the bearing resistance of a single steel tooth P Rd is determined in dependency of the loading specified and in accordance with [iv]. Hereby influence of the increase of the nominal stresses of the steel section due to the connectors geometry has been neglected as it is insignificant in the plastic design. 113

116 Fig. 15: Photo of experimental ULSfailure Fig. 16: Forces and stresses in a critical section at ULS Thus, the following design criterion is derived: P Rk f y t w h' ( b b 2 i i hs, i 3bi e x 2 b ) [kn] Eq. 4 with f y Yield strength steel [MPa], t w Plate thickness of web [mm], h s,i Distance of centre of gravity to critical section = h s ( 1 cos ) R [mm], b i Width at critical section = b 2 sin R 1 [mm], h i h' (1 cos ) R, e x distance between connectors, fig. 13. α angle along cutting edge, fig. 11, For transversally reinforced cross sections, uplift is not present and therefore the formula simplifies to: P Rk 2 f y tw bi [kn] Eq hs, i 3 bi For the puzzle shape the maximum equivalent stresses derived from Eq. 5 has been located to be at α = 70 for the POST which is in accordance to the tests (Fig. 15). 114

117 The characteristic resistance is further to be reduced by statistical evaluation of the formula has not been carried out. However Annexes l'en1993, usually M =1,10, is proposed to be applied in design. M for design. Due to the limited test number, a M given by the National 2.2 Analytic approach to the SLS design The SLS design is based on the stress verification. Hereby the stresses in the steel dowel due to longitudinal shear should not exceed the yield strength of the steel. The stresses due to global bending are neglected as they are assumed to act only below the dowel line (notch effect of dowel is neglected). Small local plastification is therefore accepted. The design is verified with the following equation: [kn] Eq. 6 max f y / M, sls with max t w b 3 h' ( b 2 i i hs, i 3 bi 2 e x 2 b ) [kn] Eq. 7 f y, t w, h s,i, b i, h i, e x, α see Eq. 4. M, sls is the partial safety factor for SLS design. For transversally reinforced cross sections, uplift is not present and therefore the formula simplifies to: 2 f y t w bi max [kn] Eq h 3bi 2 s, i 2.3 Analytic approach to the FLS design The fatigue design of the steel part is divided into two parts. One part is dedicated to the fatigue design of the web taking the increase in stress due to the notching effect of the steel tooth into account. The second part treats the estimation of the fatigue stresses along the gas cut edge of the connector itself, considering the effect of the shear stresses as well as the nominal stresses in the steel section and their verification Fatigue resistance of the gas cut edge At first, the fatigue resistance of a gas cut edge has to be specified. According to the Eurocode EC3-1-9 the fatigue category of a gas cut edge is 140 when subsequent dressing is applied. Hereby all visible signs of edge discontinuities have to be removed. The cut areas are to be machined or ground and all 115

118 burrs to be removed. Any machinery scratches, for example from grinding operations, can only be parallel to the stresses. If the cut has shallow and regular drag lines with cut quality II according to EN 1090 (for railway bridges cut quality I [DIN FB 103]) the fatigue category is reduced to 125. For both categories repair by weld refill is not allowed. Re-entrant corners are to be improved by grinding appropriate stress concentration factors. However it has been noticed that short stopping of the flame cutter decreases the fatigue strength to 60%. This results from the change of the failure initiation to the cut edge. Further it has been observed that hammering (an effect which may occur due to hammering of the concrete dowel connector in gaps from which plastified concrete may have disappeared), cutting speed, warming before cutting and material strength have hardly an influence on the fatigue strength. Therefore crack initiation occurs in the HAZ along the cut. The design value is therefore conservatively derived σ C to = 125 MPa. If stopping of the flame cutter can not be avoided it should take place at an irrelevant location in terms of fatigue Fatigue design of the steel web General design methode Due to the notch effect of the steel tooth the stresses of the web along the cut edge are increased. The reduction of fatigue resistance due to the geometry is comparable to the effect by longitudinal stiffeners, however only the geometrical effect has to be considered as the material notch due to welding is inexistent. Therefore the fatigue verification has to be performed with the fatigue category of gas cut edges Δσ C = 125 MPa according to [EC3-1-9] and taking the hot spot stress into account to derive the modified nominal stress as follows: E, 2 N, w k [MPa] Eq. 9 C with N, w relevant longitudinal stresses in the web along the bottom line of the connector [MPa], k C stress concentration factor according to Eq. 3 for a constant geometric change. In Fig. 17 the comparison is shown indicating the EC3-1-9 fatigue class on the x-axis and the damage due to the geometry (damage per load cycle for a reference curve of 125 MPa). The notch factors are hereby taken from the Code Aster [iii] model - the results show clearly, that the clothoidal shape is the most appropriate shape for structures under fatigue loading. 116

119 Fig. 17: Comparison of details investigated with details classified according to the EC Design for Puzzle shape Fatigue crack initiation and propagation depend on the principle stresses along the cutting edge. To derive an analytic design model for fatigue verification the principle stresses have consequently to be considered, which are supposed perpendicular to the radius (Fig. 18). Fig. 18: Analytic model for principle stresses (in dependency of the angular α) Fig. 19: Principal stress trajectories in steel part of connector (P Rd and P up 0 and σ N influence) With the loading defined previously and in dependency of α, the following fatigue load resistance has been derived: 117

120 P FAT C k C N, w 2 t w br 3h' i ( b2 b3 ) b cos 6 sin 2 r hsi e x [kn] Eq. 10 with Δσ C Fatigue strength of gas cut edge (125) [MPa] b r 1 ( 90 ) b1 sin R 2 90cos h s,i, t w, h i, e x, α see Eq. 4, N,w, k see Eq. 9, C For transversally reinforced cross sections, uplift is not present and therefore the formula simplifies to: P FAT C k C N, w b r t w b 2 r cos 6h si sin [kn] Eq. 11 For the puzzle geometry investigated the maximum principle stresses along the cut edge have been derive to be located at an angle α = Design check The overall design approach is based on the fatigue verification of the EC3-1-9 and results as follows: P P Eq. 12 Ff Mf F FAT whereas is derived with P and PF 4 i1 i max Eq. 13 and Mf as defined in Tab

121 Assessment method Consequence of failure Low consequence High consequence Damage tolerant Safe life Tab. 3: Partial safety factor for fatigue strength Mf The choice of Mf depends in the consequences. It should hereby be noted, that the web of the beam is directly concerned by the risk of fatigue and tension forces may be present. Therefore a PreCoBeam connector may not be damage tolerant with high consequences of failure. For the fatigue design of the steel web, it has to be noted, that the number of axles of each vehicle is significant. For a detail to be design in mid span, the load histogram for a two axles vehicle is derived according to EC. Bending moment histogram at midspan Shear force histogram at midspan Fig. 20: As example, loading histograms due to a vehicle where obviously n = 2 The results of the rainflow analyses for the Palmgren-Miner rule are derived from Fig

122 Fig. 21: Rainflow analysis of the load histogram of a vehicle crossing for location in midspan Consequently the formula has is extended to account for several axles as follows: 5 F 5 5 ( n 1) [MPa] Eq. 14 It is noted, that the damage accumulation has hereby been carried out according to the Palmgren-Miner in the safe domain of the Wöhler curve with m = 5. The final design check is then carried out with F ( 1 k, ) [MPa] Eq. 15 C w C with Δσ C Fatigue strength of gas cut edge (125) [MPa], k C, w stress concentration factor according to equation 3 for a constant geometric change. according to Eq Alternativ design method based on FE analysis On basis of advanced FEA the maximum design stresses d in the dowel can be derived. Hereby it has been distinguished between: stresses caused by global actions (bending moment and axial force acting on beam) and stresses caused by local action (longitudinal shear from composite action in the shear joint of the beam, which dowel is subjected to resist). Parameters for the stress calculation are the shape factors, which depend on the shape of the connector. Factor k c is the stress concentration factor, as introduced before, to taking into account increasing of normal stresses in regions, where surface of the steel part of the beam is disturbed (here dowel root). This factor is determined by means of FEA of model consisting of only steel part of the composite beam and submitted to axial tensile stresses. The A el factor takes into account effects of local actions between steel and concrete on dowel connector and was derived using FE models consisting of steel and concrete part. The stresses from local actions result 120

123 from the force reaction from the concrete chord into the steel connector this reaction causes local bending stresses and shear stress in the connector. On the basis of this methodology a straight forward design method based on FEA has been elaborated. Hereby the stresses relevant in design (maximum von Mises stresses along the cutting line for SLS and ULS and the principle stresses for fatigue design) are calculated as follows: d 1 L kc [MPa] Eq. 16 N, w Ael tw with t w web thickness, L Longitudinal shear, A see Tab. 4, el k notch factor, see Tab. 4. c Just to note : [Eq. 17] N N A M. y I max [Eq. 18] V L I A' y. d 2 a The factors A el,l and k c have already been derived for the clothoide shape and are listed in Tab. 4 for all design cases. 121

124 Design check ULS SLS FLS A el A ult = k c not to be applied Tab. 4: Loading on a steel part - Steel design for the CL-shape Further the SLS stress verification required for bridges has been elaborated. Stresses along the cut were evaluated by FEA. It has been established that the generalized stress Equation 15 for the CL shape is safe on every point to evaluate the stresses and can be used for the design. / Eq. 19 S f y M, sls and S derived with Eq. 16 M,sls is given by national Annexes l'en bridges. Usually M, ser =1,0. 122

125 3 DESIGN GUIDANCE - CONCRETE In the following chapters an empirical design of the bearing resistance of concrete dowels and an analytic approach for the design of the surrounding concrete matrix is presented. Further recommendations to the fatigue design of concrete are given. In Tab. 5 an overview on the design checks for concrete in the limit states is given. Design check Bearing resistance Concrete dowel FLS Load model LM 1 & LM 2 LM 1 & LM 2 LM 3 - shear Check - local compression - shearing - splitting vertically - pry-out of concrete cover with interaction of tight laying dowels (long./transv.) Upper load limitation - spalling of concrete cover Tab. 5: Design of PreCoBeam shear connection Concrete design 3.1 Empirical ULS bearing capacity for the concrete dowels Concrete failure is characterized by several failure modes. During the conducted experimental test series, three major concrete failure criteria have however been observed. To this failure modes design criteria have been established [v] and statistically assessed. The ultimate bearing capacity of the concrete dowel is at first limited by the local pressure in the contact area between concrete and steel. Therefore, the concrete in the slot is split in two sections. In Zone A an almost hydrostatic stress state is present. Zone B is the area of the load diffusion. The ultimate bearing capacity is limited by the three axial concrete pressures in Zone A. There, vertical to the pressure, tensile stresses occur (Fig. 4a). A second failure criterion was observed by using a perforation with bigger diameters, deeply placed concrete dowels in the concrete slab or if thicker steel plates were used. A double cut shearing off within the concrete slab at the steel slots occurs. The concrete surfaces and the reinforcement surfaces are converted into an ideal concrete dowel section for the calculation. The shearing surfaces are not 123

126 located in parallel to the steel web surfaces (Fig. 4b). Thus, the form factor f h for bigger slots was defined to cover this effect. A third failure criterion could be observed during the tests. The topology of the concrete dowel and the lateral reinforcement is mainly responsible for this failure criterion because concrete dowels based close to the edge of the concrete slab collapsed by a punching cone fracture at the maximum load level. This cone is caused by the transversal tensile stresses which exceeded the concrete tensile strength. A possible transformation to the mechanical calculation model is shown in Fig

127 Fig. 22: Additional failure mechanism of concrete identified in testing (Pry out cone) 0 Fig. 23: Reinforcement T-beam cross section The tests as well as a statistical evaluation of the test results are based on an existing design guide [v]. However for the design of PreCoBeams the failure mechanisms need to be revised with the introduction of the following factors: Shape factor s The load bearing behaviour is influenced by the cut of the PreCoBeam connector. Thus, a shape factor for the PZ-shape was elaborated. The implementation of the shape factor s results in a further improvement of the presented design rules for the PreCoBeam connector. The shape factor s is influenced by the thickness of the steel web. Thin steel webs are more ductile than thick steel webs. The shape factor s covers this effect. The displayed shape factor s was statistically proved by the conducted tests T-Beam cross section factor t f Based on the experimental tests and the FE simulations, the load bearing capacity of PreCoBeam shear connectors implemented in T-beam cross sections was ascertained. The t-beam cross section leads to spalling of the concrete part. Thus, a reduction of the bearing capacity is necessary. The reduction is reliant on the thickness of the T-beam cross section. The bearing capacity is influenced up to a width of the T-beam of about 41 cm. Further, the experimental tests showed an advantageous influence of an additional reinforcement in the concrete section to the post fracture behaviour. The additional reinforcement leads to a better stress distribution in the concrete Interaction of parallel steel dowel strips The distance between two or more steel girders with PreCoBeam shear connectors (Fig. 24), also influences the ultimate load bearing capacity. The experimental tests and the FE simulations showed an 125

128 influence up to a distance of about 100 cm. The reduction of the maximum load capacity is subjected to the distance. Therefore, two limits were defined. Distance Distance factor l cm cm 0.97 > 100 cm 1.00 Fig. 24: Distance e y between steel girders Fig. 25: Notations for the design concept Comprehension of Design Concept Concrete (UBWM) The tests as well as a statistical evaluation of the test results are the basis of the established design guide [v]. The elaborated design procedure differentiates between the three mechanisms of failure are shown above. Therefore, during the dimensioning the proper failure mechanism with the related load bearing capacity has to be determined. Fig. 25 displays the notations of for the design guidance are introduced. The partial safety factor is assessed as γ v =1.25 in correspondence with the Eurocode 4. The rules were statistically risked proved acc. to EC 0. In the following the design guidance is given in reference to the three failure mechanism of concrete: Local compression failure: P 1 [kn] Eq. 20 Rd 72.7 l t f s f 1/ 2 1 ck hd t w v Shearing failure: 126

129 P 1 [kn] Eq. 21 Rd l Ad t f s f ctk i f h v Pry-out failure: P [kn] Eq Rd 2 l h po t f s f ctk i whereas: v i E S / Ecm 1 Asq / Ad [-] Eq f h 1.2 h /180 1 d [-] Eq s 0,0012t w 0,0537 t w 1,29 1,0 [-] Eq. 25 t f h tc 1 75 sp w 0,46 u 1,0 [-] Eq. 26 h c [mm] Eq. 27 with: t w Steel web thickness [mm] f ck Concrete grade [N/mm²] h d Height of the steel dowel [mm] h po Height of the alternative cone [mm] f ctk Concrete grade [N/mm²] A d Concrete dowel area [mm²] f h Form factor [-] w Width of the t-beam cross section [cm] ρ Grade of reinforcement [-] 127

130 s Shape factor [-] t f Reduction factor t-beam cross section [-] γ Partial safety factor (1.25) [-] l Distance factor l [-] The displayed equations are based on partial safety factor γ v of It has to be noted, that e.g. the new German national application document (NAD) contains an amended partial safety factor γ v of 1.5 for the concrete part in composite constructions. Therefore, a correction of the presented design concept for countries with diverging partial safety factors would be necessary Limits of design criterium The application of the presented design concept for the concrete part is limited. The limitation is based on the range of the conducted experimental tests are given in Tab.6. Tab. 6: Limits to empirical design approach Steel part: - Steel S235 S460 - Web thickness mm - Shape Puzzle Concrete part: - Concrete C20/25 C70/85 - Concrete cover 50 mm - Width T-beam cm 128

131 3.1.6 Fatigue behavior FLS The fatigue behaviour of the concrete slab was also scrutinised. Cyclic loads lead to a degradation of the concrete in the contact zone. The tests showed that the fatigue behaviour of the concrete slab mainly depends on the upper load level. Thereby it is possible to design against concrete shear failure by limiting the upper load level to: Pcu Eq. 28 P stat with: λ 0 Working load [-] P cu Characteristic load per dowel [kn] P stat Static bearing capacity per dowel [kn] 3.2 Analytical design for the surrounding matrix of the concrete dowel Due to the test results and existing tests an analytical design model is established in comparison to chapter 3.1. The aim is to get design rules for changing boundary conditions in the surrounding concrete matrix. The basis of the concept are the failure modes given in Tab. 7. Based on former investigations and the test series analytic design models have been derived. Vertical crack in the nonreinforced plate Shearing of the concrete dowel Horizontal crack in the concrete plate Spalling of concrete cover Pry-out cone in the concrete cover Tab. 7: Possible failure modes of the concrete dowel 129

132 3.2.1 Shear in the concrete dowel The reinforcement in the concrete dowel are two bars recommended lying on the dowel s base. For concrete plates as well as for girders out of concrete the concrete dowel stays in a well confined environment. The twin reinforcement bars increase the capacity because of the block action in the composite dowel. If reinforcement is placed in the concrete dowel or in its sphere of influence, however, the composite dowel carries a rising load with disproportionate increasing slip. The wedgeshaped high loaded zone consisting of crushed concrete enlarges and penetrates into the core of the concrete dowel. A progression of the crack width is prevented by the crossing reinforcement and the interface shear starts by interlocking the aggregate. This interlock is supported by the dowel action of the crossing reinforcement bar of the concrete dowel. For rough interfaces of concrete-to-concrete surfaces friction for the ultimate limit state is derived from equation u f 1/ 3 2/3 0.40fc s1 y ; A 1 A d, c Eq. 29 P sh f 1/ 3 2A 2/3 d, c u fc s y Eq. 30 Fig. 26: Shear interface A d, c and vertical tension area d c A, of concrete The ultimate shear resistance is determined by the concrete compression strength and the ratio of the reinforcement in the embedding concrete. Lateral compression and a pre-stressing influence positively the bearing capacity because of a higher friction and the improvement of the 3-axial compression state. The dowel action of the reinforcement bar is activated by the large slip between the concrete s interfaces. The minimal slip between steel dowel and concrete dowel is 0.1 d s. In all test series the maximum load for a slip value is larger than 3mm. Hence reinforcement bars with a diameter up to 28mm are acting as a dowel in the shear interfaces. If there are twin reinforcement bars applied in the concrete dowel P re can also be doubled too. A preloading s of the reinforcement is negligible. It derives to P re 1.30 d f f 1 / f 2 As f e c y s y with 3 f c / f y 3 d 2 y s 1 s Eq

133 Both parts of shearing and dowel action can be added and together they result in the shear resistance of the reinforced shear connection P P P sh, re sh re Pry-out of concrete cover The initial point for the collapse of a concrete dowel is the transgression of the shear capacity. A steel dowel which is well embedded in a concrete specimen is considered. Loaded by local compression, spalling forces are introduced into the concrete dowel. If the concrete is reinforced adequately, the local compression * cc f may increase entirely. The dowel is acting with shearing until the concrete dowel is cracking because its shear capacity is not sufficient. The progressing failure mode of the concrete is schematically described in Fig

134 I) I) Dowel force P introduces stresses by local compression into the uncracked concrete dowel which is confined by a rebar with the area A s II) II) The splitting tensile forces introduced by the dowel exceed the concrete tensile strength. It creates a vertical crack and the tensile stresses are transferred into the transversal reinforcement bar. In front of the steel dowel the concrete matrix crushes and it occurs a pulverized concrete wedge. III) III) The concrete wedge penetrates the concrete dowel by increasing slip values. Shear interfaces arise in this area. IV) IV) These shear interfaces are fully developed. Based on the mutual displacement the dowel action of the reinforcement bar is mobilized and prevents a shear progress in the concrete dowel. Fig. 27: Typified progression of shear behaviour in a concrete dowel 132

135 The load and the slip between the steel dowel and concrete are increasing. The surface in front of the dowel starts to be compressed (I). The pores of the concrete in the introduction zone are contracted and the wedge-shaped pulverized matrix penetrates the dowel core (II). Reaching the shearing capacity of the concrete dowel a shear failure in the dowel occurs. The wedge develops further and the concrete struts of load introduction become steeper (III). If the load is more increasing and in case of a limited concrete cover, the inclined strut looses its reaction force in the concrete. The tensile stresses in the envelop of the strut are exceeded and a concrete cone pries out at the bottom of the concrete (IV). Fig. 28: Schematic failure behavior of pry-out criterion Fig. 29: Parameter of the elliptic cone with an elliptic basis and comparison of the test result with the computed results on the basis of a cut image 133

136 The computed results are sketched in the cut image from [vi] of a push out test of a closed composite dowel. The inclination and the apex angle of the elliptic cone are congruous to the failure pattern in the test specimen (Fig. 29). These geometric boundary conditions precisely describe the pry-out cone. Based on the now established geometry of the pry-out cone it is possible to determine the interaction of tight laying concrete dowels and to adjust the theoretical bearing capacity of the connection. The failure criterium of pry-put cone P po is only decisive if the shear capacity of the concrete dowel is exceeded: P po P sh, re P max P sh, re Eq Increase of pry-out capacity by guiding of reinforcement bars By opening the test specimens a beam action of the twin stirrups can be observed in the concrete dowel. This transversal beam represents no demolition of the concrete matrix after the ultimate load tests. The block effect of the twin reinforcement bars is taken into account for the bearing capacity of the connection. The compression is spreading into the concrete dowel. Caused by the increasing slip, cracks are induced in the matrix of concrete and the compression transfers to the twin reinforcement bars. The horizontal laying reinforcement pattern acts as a single supported beam over the span. The cross-section of this beam is composed of the concrete block with the internal lever arm of the distance of the reinforcement bars ( 134

137 Fig. 30). The load introduction zone spreads out to the rearmost reinforcement bar. The angle for the ultimate state is 45 in concrete. Fig. 30: Load introduction into the twin rein-forcement bars from the composite dowel in ultimate limite state of the concrete By using the equilibrium M M R, the increase of the ultimate resistance for the composite dowel results in adding twin stirrups in the cases of a concrete cone pry-out in P 2re 12,5As 2 f y e e e 2t x re re w Eq Decreasing capacity in tight laying composite dowels The bearing capacity of one dowel depends on the distance e x between the dowels in longitudinal direction and the distance e y between the dowel strips in transversal direction because the shapes of the pry-out cones are overlapping. 135

138 a) b) Fig. 31: Interaction of dowels with a tight distance in a) longitudinal and b) transversal direction Generally it is assumed that the bearing capacity of the pry-out cone is linearly decreasing depending on the value of its elliptic base area. Caused by overlapping of tight laying pry-out cones the elliptic base area is abated by the two segments S PBN. In the case of the pry-out cones overlapping in longitudinal direction parallel to the dowel force, the reduction factor is defined for an infinite number of dowels in a row. l 1 ex ex l 1 2 arccos 2a a ex 1 2a 2 Eq. 34 According to the interaction of dowel capacity in longitudinal direction the pry-out cones are overlapping also in transversal direction if several dowel strips are located in a cross-section. The reduction factor for the capacity in transversal direction is defined by. The geometric parameters are set to to e e and a b analogously. For numerous dowel strips it derives the interaction corresponding x y q q, 1 ey ey 1 2arccos 2b b ey 1 2b 2 Eq. 35 In practice, however, only few dowel strips are applied in transversal direction. Hence the reduction of the load bearing capacity is changing to 136

139 q, n n 1 ey ey 1 2arccos n 2b b 2 ey 1 2b Eq Spalling of concrete cover Actually, the spalling of concrete cover is a failure mode which should be assigned to the failure mechanism horizontal cracking. This failure mode happens frequently because of the typical geometry of a beam cross-section and is therefore highlighted in this chapter. The test results show that the spalling of the concrete cover initiates the ULS state. But the concrete cover isn t secured by reinforcement bars which can absorbe the tension forces. As a result, the bearing behavior skips the SLS stage and it transfers immediately into ULS showing the typical post failure behavior which is mainly influenced by the transversal reinforcement bars in the concrete dowel. The spalling of the concrete cover is a progressing development. In the plane a-a below the reinforcement bars of the concrete dowel the tension stresses exceed the maximum level in a local area. An incipient crack occurs and the tension stresses transfer into the still uncracked region. By increasing tension stresses the crack is spreading a progressing collapse and spalling of the whole concrete cover. The stresses in longitudinal direction introducing the steel dowel force into the concrete by local compression are spreading symmetrically to the dowel axis into the plane A0 2h po b w t w / 2. Considering this fact, the pocket in the concrete caused by the steel dowel is neglected because of their minor influence. The initial crack in the plane below the stirrup occurs if the existing tension stresses exceed the resistance tension stresses of the concrete: f. For a composite dowel located in the centre line of the concrete cord the ultimate dowel resistance is z ct P cov hpobw hd h po f ct Eq

140 Fig. 32: Scheme of the force flow introduced by the uplifting force The tension stresses are calculated from half the dowel force. It has an impact on the horizontal part of the stirrup. But the stirrup is not able to carry loads perpendicular to its longitudinal axis. A crack in the dowel core comes into existence stretching along the bottom side of the stirrup in direction to the concrete web surface until the stirrup turns into the radius point y. At this point the stirrup is able to carry these additional forces in z-direction due to its change in bending. But the basis for carrying the increasing load is the supportive action of the non-cracked concrete zone at each side which has the dimension. The uplifting force is now carried in the plain. The inclination of the concrete strut is explained by the ratio but is limited to the angle of 45 because of geometrical reasons. This equation calculates by the stresses caused by the spreading of the local load introduction in the midplane of the concrete dowel. Caused by the eccentricity of the upper load part of the uplifting force, F t,2 is acting disadvantageously on the plain stress. F t, 2 struts on the horizontally laying stirrup and puts a strain on the outer area of the concrete part one half each. This force can be carried until in point y the resistance of the tension stresses is exceeded. An initial crack arises in the plane a-a which will propagate immediately in the whole plain. Decisive for the initial cracking, are the stresses loaded by uplifting force which corresponds to the tension stress. Furthermore, the tension part caused by the uplifting force is added to the stresses in that plane. f ct, cr P h 2 ' 4 l e 2 cr x P h po b w h 1 2h d po Eq. 38 The dowel force results by the critical composite dowel force P cov caused by spalling of the concrete cover P cov 4l h ' 2 2 crex f ct 0.15 h d 1 h pobw 2h po Eq

141 3.2.4 Comprehension of the Analytical Design Concept Concrete Fig. 33: Assignments for the concept due to Tab

142 Concrete d minmax sh, k ; po, k ; cov, k / v c, k ; V 1,25 ULS UShear Eq. 40 w/o reinf. concrete dowel sh, 0, k d, c 2/ 3 ck 1/ 3 yk A f f / e x within reinf. in concr. 3t w fck / f yk dowel 2d s1 d s2 Eq. 41 1/ f yk 1.5d s1 ds fck f yk ex 2 / 3 sh, re, k 0.9 A d, c fck / As A d, c Eq. 42 USplitting vertically (for non-reinf. concrete) 4.25 ) ( f c e A / 1 t / b cr, v, k ctk o x d, c ex w w UPry-out of concrete cover Eq w/o or with 1 reinf. bar in the concrete dowel 2. h f po k, 15 e x po ctk Eq. 44 With two reinf. bars in the concrete dowel As 2 f yk e 2.15 hpo fctk ex ex ere 2t po, re, k re w Interaction of tight lying dowels Reduction factor: l q e x 2a ; a hpo e y 2b ; b hpo 1 ex ex l 1 2 arccos 2a a ex 1 2a 2 q, n 1 n 1 ey ey 1 2 arccos n 2b b 2 ey 1 2b 140

143 Eq. 45 bw If h po USpalling of concrete cover cov, k If h ' e 1 e x h ' 4l h ' e 2 2 crex 2 x : 2 x f ctk 0.15 h d 1 h pobw 2h po Tab. 8: Decisive capacity of concrete dowel 141

144

145 4 FIRE The main conclusion relative to the results of the tests and therefore for the studied geometries is that by providing a good cold design, there is no problem of longitudinal shear in fire condition. Moreover, we have to note that for these tests, some systems have been used to artificially increase the bending resistance and thus increase the solicitation in shear. In the critical beam area where the bending moment is maximum, intumescent paint, ceramic fibres or additional reinforcement bars have been put in place. The aim of these protections was to increase the temperatures in the dowels at failure and to activate a shear failure induced by fire before a bending failure. The tests show that it was nevertheless impossible to reach a dowel temperature greater than 556 C and impossible to get a failure by shear. This is mainly due to the geometrical configuration of the PreCoBeam system where the position of the longitudinal shear connectors in the section which is such that the concrete offers a very efficient thermal protection. Therefore, in general and all the more without such special protections, this conclusion can certainly be extended to real PreCoBeam sections with common spans. However, without any other available result of fire tests that lead to failure by shear, a simple design recommendation is given to the designer to limit the temperature of the steel dowel to the maximum temperature reached in these tests i.e. 500 C. This requirement will be easily met for most common PreCoBeam section types. Concerning the thermal calculations, simple rules have been derived from thermal simulations with SAFIR to determine the temperatures in the PreCoBeam sections. These simple rules are based on tables for typical cross sections of PreCoBeams submitted to ISO fire curve and for different times: 30min, 60min, 90min and 120min. The tables depend on the flange and web thicknesses, the width of the flange and the height of the steel web. The knowledge of the temperature distribution in the PreCoBeam section allow to verify the mechanical behaviour of the dowels if it is necessary and allow to verify the mechanical response of the PreCoBeam to the external loads when it is submitted to fire. The fire tests confirm that PreCoBeams can be designed according to the rules of Eurocode EN General principle The fire tests showed that the longitudinal shear continuous connectors, i.e. the dowels of the Preco- Beam system, lead to a complete connection and therefore allow reaching the plastic moment in the critical section. The tests also showed that the interaction between steel and concrete is not complete. That leads to an increase of the deformations compare to the deformations of a theoretical case with a perfect contact and a perfect interaction between the two materials. The necessary condition of these conclusions is a good cold design of the Preco-Beam. Consequently, the resistant bending moment value of a Preco-Beam section in fire condition can be calculated according to the Eurocode EN

146 However, considering that the maximum temperature reached in the dowels was 500 C during our fire tests, it is not safe to extend this conclusion to higher temperatures in the connection without any other experimental results. Therefore, the verification of the limitation of the temperature at the basis of the dowel to 500 C has to be done to validate the calculation of the resistant plastic bending moment. In the same way, for the calculation of the deflection in case of fire, the methods proposed in EN based on Bernoulli theory i.e. the conservation of plane cross sections can be applied and gives a sufficiently good estimation of the deformation due to fire. 4.2 Conceptual design The Preco-Beam section has to be designed to ensure resistance to bending. Different possibilities are: - either by delaying the elevation of temperature in the steel flange with a thermal protection; - and/or by protecting thermally the lower part of the steel web. For example by adding a concrete cover on the steel web (this solution is also favourable for the heating of the steel flange because only the lower side of the flange will be heated); - and/or by adding supplementary reinforcement bars in the concrete. The Preco-Beam section has to be designed to ensure resistance to shear resistance by limiting the elevation of temperatures in the dowels: - either by delaying with thermal protection the elevation of temperature in the steel flange or in the steel web because the heating of the dowels comes from the transmission of the heat from the steel flange to the steel web to the steel dowel by thermal conduction; - and/or by placing the dowels at a sufficiently high level in the Preco-Beam section and if it is possible, to place the dowel in the concrete slab; - and/or if the dowel is not positioned in the concrete slab, by ensuring a sufficient width of concrete on both sides of the steel web (anyway this condition is necessary for a good design in cold condition to ensure the longitudinal shear transmission) Examples of conceptual design for different types of Preco-Beam section Type of Preco-Beam section Actions to improve the bending resistance Actions to improve the shear resistance Type 1 - if RF 30min : to over-size the steel section to limit the stress nothing because : - to protect the steel flange - if protected steel flange, the connectors remains cold 144

147 - to add supplementary reinforcement bars in the lower part of the Preco- Beam section - if supplementary bars, the reinforced concrete section has its own shear resistance. It has to be verified whether this resistance is sufficient Type 2 - if RF 30min : to over-size the steel section to limit the stress - to ensure that h 2 is sufficient to have T dowel 500 C h 2 - to protect the steel profile with paint, sprayed material or box (insulation plates) if thermal protection : - by box : there is no probleme for the shear connectors - by painting or sprayed material : to ensure that h 2 is sufficient to have T dowel 500 C Type 3 - if RF 30min : to over-size the steel section to limit the stress - to ensure that b 3 is sufficient or - to place the dowel in the concrete slab h 3 b 3 - to add supplementary reinforcement bars in the lower part of the Preco- Beam section - if supplementary bars, the reinforced concrete section has its own shear resistance. It has to be verified whether this resistance is sufficient - to increase the height h 3 to increase the cold part of the steel web - to protect the steel flange 145

148 Type 4 h 4 - if RF 30min : to over-size the steel section to limit the stress - to protect the steel flanges only - to ensure that h 4 is sufficient to have T dowel 500 C if thermal protection : - to protect the entire steel profiles with paint, sprayed material or box (insulation plates) - by box : there is no probleme for the shear connectors - by painting : to ensure that h 4 is sufficient to have T dowel 500 C - to add supplementary reinforcement bars in the lower part of the Preco- Beam section between the 2 steel webs - if supplementary bars, the reinforced concrete section has its own shear resistance. It has to be verified whether this resistance is sufficient Type 4 bis - idem type 4 - idem type 4 this example shows one possibility to increase h 4 h 4 146

149 Tab. 9: Decisive capacity of concrete dowel 4.3 Calculation rules Thermal calculations The most accurate method to obtain the temperature distributions in the Preco-Beam sections is to use an advanced calculation model according to Eurocode EN The advanced calculation models for thermal response have to be based on the acknowledged principles and assumptions of the theory of heat transfer. The thermal response model considers - the relevant thermal actions specified in EN the variation of the thermal properties of the materials according to EN In this research project, the advanced model used was the Finite Elements Sofware SAFIR, developed in University of Liege. The comparisons between 2D and 3D models showed that the two models give similar results and therefore that the 2D model is sufficient. However, according to the type of the Preco-Beam section considered simple calculation rules exist or have been developed in this research to obtain the temperature distribution in the section without doing numerical simulations Type 1 section: The aim of this method is to give the temperatures in the Preco-Beam section taking into account the specific configuration of this type of cross section. The temperatures are given for the standard temperature-time curve temperature. The Preco-Beam section is divided in 3 zones (Fig. 34). The zone I is situated in the concrete part where the steel profile has no influence. The zone II goes across the steel flange and the concrete slab. The zone III is situated above the steel flange and along the steel web. Type 1 : b eff T I h c,h h d T II T III z h III t f b t w h c h c,craked zone III zone II zone I Fig. 34: Temperature distribution according to Preco-Beam method for type 1 Preco-Beam section 147

150 Zone I: Depth Temperature of the concrete T I [ C] after a fire duration in min of z (mm) R15 R30 R60 R90 R120 R180 R Tab. 10: Temperature distribution in a solid slab of 200 mm thickness composed of normal weight concrete and not insulated. Tab. 10 gives the temperature as a function of the concrete depth for different times and may be used to obtain the location of the isotherm as a conservative approximation. Because the concrete resistance starts to decrease for temperatures higher than 100 C, the properties of the concrete situated at a depth greater than 200mm can be considered as properties of room temperature concrete. 148

151 R240 R180 R120 R90 R60 R30 R Temperature ( C) R15 R30 R60 R90 R120 R180 R z (mm) Fig. 35: Temperatures as a function of concrete depth z for different fire durations Note: This kind of table already exists in Eurocode for the evaluation of the temperatures in a concrete slab (EN Annex D) but for a maximum concrete height of 100mm. Zone II - Steel flange: The temperature in the steel flange can be calculated with the Tab. 11 which gives the temperatures at different fire durations for five flange thicknesses. For a flange thickness different from the five given values, a linear interpolation can be made. Temperature T II ( C) in the steel flange Time (min) t f = 5 mm t f = 10 mm t f = 20 mm t f = 30 mm t f = 40 mm R R R R R R R

152 R R Tab. 11: Temperatures after different fire durations for different flange thicknesses Note: A method exists in Eurocode EN to calculate the flange temperature by a step by step method. The section factor A b 2t is equal to V i i b. t f f. This method gives too high temperatures because the heat conduction to the concrete slab and to the steel web is not considered. Nevertheless, this method gives safe results and presents the advantage that it can be applied with any fire curve. EN equation (4.6) Temperature ( C) R240 R180 R120 R90 R60 R30 R tf (mm) Fig. 36: Temperatures as a function of flange thickness for different fire durations Zone II - Concrete part: The simple evaluation of the temperatures in this zone is very difficult but can be useful to calculate the temperatures in the eventual reinforcement bars placed into it. At the end of the steel flange (limit between zones I and II), the temperatures can be calculated as in the zone I for different fire durations. At the interface between the steel web and the concrete (limit between zones III and II), the temperature can be calculated as in the zone III for different fire durations. Therefore, the method consists to interpolate linearly the temperatures between the 2 zones I and III to find the temperatures in the zone II - concrete part. 150

153 Zone III: The temperature in zone III depends on the steel flange thickness, the steel web thickness, the steel flange width and the encased steel height h III. The method is based on three equations and on tables depending on these geometric parameters. The method is the following: T(z) 20 (t).z T 20 e surf T Kt.Kt.Kb surf,20,10,200 surf T 0.T f w T K K K surf,20,10,200 tf t w b T surf tables K tables K tables K (t) t t f t w b, b f, t, t w t tf 20mm tw 10mm b200mm t (t) 1.k t f.k tw.k. (t) b 20,10,200 (t) k k k h tf tw b (t) tables k tables k tables k t t f t w b, 0,9 2,446.e III 20,10,200,t w,t and z are given in m b tf 20mm tw 10mm b200mm f t t 16h III Time T surf20,10,200 t K f (mm) tf min min min min min min min min min min min min min min Time Time t K w (mm) tw min min min min min min min Time b (mm) K b min min min min min min min Tab. 12: Tables to calculate T surf for different fire durations 151

154 t 20,10,200 k f (mm) tf Time 15 min min min min min min min 5.83 Time 15 min min min min min min min Time t k w (mm) tw min min min min min min min Time k b b (mm) min min min min min min min Tab. 13: Tables to calculate β(t) for different fire durations The temperature T surf is the temperature at the surface between steel flange and steel web. It depends on the 3 parameters t f, t w and b. The temperature along the steel web is given by the T z which is an exponential function of the height z (in m). It also depends on the 3 parameters t f, t w and b. The influence of the height h III is given by the coefficient α. Dowel: The temperature in the steel and concrete dowels (height h d on Fig. 34) can be approximated in a conservative way by the temperature at the level z = h III i.e. at the base of the steel dowel Type 2 section: Type 2 : b eff T I h c,h h d h c T IIIb z h IIIb h c,fi t w h IIIa T IIIa T II t f b zone III zone II zone I Fig. 37: Temperature distribution for type 2 Preco-Beam section The type 2 Preco-Beam section is characterised by a steel part directly exposed to fire and is also divided in 3 zones (Fig. 37). The zone I is situated in the concrete part where the steel profile has no 152

155 influence. The zone II goes across the steel flange and the concrete slab. The zone III is situated above the steel flange and along the steel web. Zone I: The temperature in the concrete slab can be calculated as in the type 1 Preco-Beam section. Another simple method is given in EN Annex F. The section of the concrete slab can be reduced as shown in Fig. 37. The temperature of the upper concrete layer h c,h may be assumed to be 20 C. The values of the thickness reduction h c,fi of a flat concrete slab are given in Tab. 14 for the different fire classes. Standard Fire Resistance Slab Reduction h c,fi [mm] R R R R R Tab. 14: Thickness reduction h c,fi of the concrete slab Zone II and Zone IIIa - Exposed steel profile: The Eurocode EN gives a step by step method to calculate the temperature in the exposed steel section. It is assumed that no heat transfer takes place between the steel flange and the steel web and with the concrete slab. This method is based on equation (4.6) from EN in case of unprotected beam and on equation (4.8) from EN in case of protected beam. The section factor IIIa). A 2. is equal to b tf V i i bt. f for the steel flange (zone II) and 2 t w for the steel web (zone (EN equation (4.6)) 153

156 (EN equation (4.8)) Zone II Concrete part: The temperature in the concrete slab can be calculated as in zone I. In case of a box-protection, this assumption is conservative. Zone IIIb - Steel encased web part: The temperature in the zone IIIb can be calculated as in zone III of the type 1 Preco-Beam section. T Z and (t) can be applied but the value of T surf is here given by the step by step calculation described in EN and applied for the steel web (zone IIIa). Dowel: The temperature in the steel and concrete dowels (height h d on Fig. 37) can be approximated in a conservative way by the temperature at the level z = h IIIb i.e. at the base of the steel dowel Type 3 section: Annex F of Eurocode gives a model for encased steel entire profiles which can be extended to Preco-Beams but it is a mechanical model and doesn t give directly the temperatures. It has to be noted that a field of application is given in EN to apply the Annex F. This verification requests a minimum value of concrete width encasing the steel web (b c ) and a minimum area of concrete (h.b c ) to constitute a sufficient protection for the steel web (see Tab. 15). Therefore, this field of application can also be used for Preco-Beam sections and constitute the limit between the type 2 and type 3 Preco- Beam sections. Standard Fire Resistance min. profile height h / min. width bc [mm] Minimum Area h bc [mm²] R

157 R R R R Tab. 15: Minimum cross-section dimensions In the model given in annex F of EN , the temperature distribution is assumed to be as shown on Fig. 38. This model allows the calculation of the sagging and hogging moment resistances of a partially encased steel beam connected to a concrete slab and exposed to fire beneath the concrete slab according to the standard temperature-time curve. This model gives only the reduction factors of the yield strength in the steel parts of the profile. Nevertheless, these reduction factors can be linked to corresponding temperatures. Type 3 : b eff 20 C 20 C h d h h t w h c,h h c,fi h c h l x h T a t f b c b Fig. 38: Temperature distribution according to EN for type 3 Preco-Beam section Dowel: Because of the influence of a large number of geometric parameters (t f, t w, b, b c, h and h total steel ) and the interactions existing between them, there is no simple calculation method to evaluate the temperatures at each level in the steel web as in the section type 1. Therefore, the only way to calculate a precise temperature in the dowel is to use advanced method. Nevertheless, if the height h h (see Fig. 38) is greater than 0, it indicates that the temperature in the dowel is equal to 20 C Mechanical calculations First, it has to be noted that only sagging moment resistance M fi,rd+ and connection under sagging moment were considered in the fire tests and therefore in this design guide. 155

158 Bending resistance The bending resistance can be calculated according to the Eurocode EN assuming that the connection is verified (see point Shear resistance ). The steel profile is considered without the dowel height and in complete connection (as noticed in fire tests). The width b eff of the concrete slab should be equal to the effective width chosen according to of EN Two methods are described in Eurocode: - Advanced method (Bernoulli assumptions) In this project, we used the F.E. software SAFIR which allows advanced calculation models used in association with any time-temperature heating curve. The advanced calculation model for mechanical response is based on the acknowledged principles and assumptions of the theory of structural mechanics, taking into account the effects of temperature. The mechanical response model takes account of: - the combined effects of mechanical actions, geometrical imperfections and thermal actions; - the temperature dependent mechanical properties of the materials; - geometrical non linear effects; - the effects of non-linear material properties, including the effects of unloading on the structural stiffness. Any potential failure modes not covered by the advanced calculation model (including local buckling, insufficient rotation capacity, spalling and failure in shear), shall be eliminated by appropriate means which may be constructional detailing. For the Preco-Beam sections, it can be considered that the experimental tests have shown that failure by shear is eliminated by the detailing of the dowels. Advanced calculation models may be used when information concerning stress and strain evolution, deformations and / or temperature fields are required. It has to be noted that the deformations at ultimate limit state, given by the calculation model, has to be limited as necessary to ensure that compatibility is maintained between all parts of the structure. - Simple method EN Simple methods are developed in Eurocode to calculate the bending resistant moment for a fixed fire resistance time. The width b eff of the concrete slab should be equal to the effective width chosen according to of EN In order to calculate the sagging moment resistance, the concrete of the slab in compression, the web of the profile, the lower flange of the profile and the reinforcing bars should be considered. For each of 156

159 these parts of the cross section, a corresponding rule may define the effect of the temperature. The concrete in tension of the slab and the concrete between the flanges of the profile should be ignored. On the basis of the essential equilibrium conditions and on the basis of the plastic theory, the neutral bending axis may be defined and the sagging moment resistance may be calculated. The Annex E of the Eurocode gives more details on the way to calculate the resistant bending moment. This method can be applied for the Preco-Beam sections considering the type of the Preco- Beam, their temperature distributions and their corresponding stress distributions (see Fig. 39 to Fig. 41) Type 1 section: Considering the temperature distribution calculated for the type 1 Preco-Beam section, it can be assumed the stress distribution shown on Fig. 40. Type 1 : b eff f c,θi /g M,fi,c T I h c,h f y,θiii-4 /g M,fi,a h d 1/3 h III h c T II T III 1/3 h III 1/3 h III f y,θiii-3 /g M,fi,a f y,θiii-2 /g M,fi,a f y,θiii-1 /g M,fi,a f y,θii /g M,fi,a z h III t f b t w h c,craked zone III zone II zone I Fig. 39: Elements of the type 1 cross-section for the calculation of the sagging moment resistance Zone I (compressive zone): if h c,h < (h c - h cr ) : The temperature in the compressive concrete zone h c,h can be assumed to be 20 C with h cr the depth z according to Tab. 10 corresponding to a concrete temperature below 250 C. if h c,h > (h c - h cr ): Some layers of the compressive zone of concrete are at a temperature higher than 250 C. The method of the Annex E can be applied. Zone II (tensile zone): The yield strength of the steel flange is reduced by the reduction factor corresponding to the temperature T II. f y,θii = k y,θii. f y,20 C 157

160 Zone III (tensile zone): The yield strength of the steel at each level z of the web is reduced by the reduction factor corresponding to the temperature T III (z) calculated according to the method described above (Thermal calculation Type 1 section Zone III). f y,θiii (z) = k y,θiii (z). f y,20 C The steel web can be separated in three parts where the temperature is calculated. The temperatures in the zone III and the corresponding reduction factors can be calculated by linear interpolation between the four temperatures T III-1 to T III-4. The steel dowel is not considered for the bending resistance Type 2 section: Considering the temperature distribution calculated for the type 2 Preco-Beam section, it can be assumed the stress distribution shown on Fig. 40. Type 2 : b eff f c,θi /g M,fi,c T IIIb T I f y,θiiib /g M,fi,a f y,θiiib,z /g M,fi,a z h d h IIIb h c,h h c,fi h c t w h IIIa T IIIa f y,θiiia /g M,fi,a T II f y,θii /g M,fi,a t f b zone III zone II zone I Fig. 40: Elements of the type 2 cross-section for the calculation of the sagging moment resistance Zone I (compressive zone): if h c,h < (h c - h cr ) : The temperature in the compressive concrete zone h c,h can be assumed to be 20 C with h cr the depth z according to Tab. 10 corresponding to a concrete temperature below 250 C. if h c,h > (h c - h cr ): Some layers of the compressive zone of concrete are at a temperature higher than 250 C. The method of the Annex E can be applied. Zone II (tensile zone): 158

161 The yield strength of the steel flange is reduced by the reduction factor corresponding to the temperature T II. f y,θii = k y,θii. f y,20 C Zone IIIa (tensile zone): The yield strength of the steel web is reduced by the reduction factor corresponding to the temperature T IIIa. f y,θiiia = k y,θiiia. f y,20 C Zone IIIb (tensile zone): The yield strength of the steel at each level z of the web is reduced by the reduction factor corresponding to the temperature T IIIb (z) calculated according to the method described above (Thermal calculation Type 2 section Zone IIIb). f y,θiiib (z) = k y,θiiib (z). f y,20 C A linear interpolation between the two temperatures T IIIa to T IIIb can be made. The steel dowel is not considered for the bending resistance Type 3 section: Considering the temperature distribution calculated for the type 3 Preco-Beam section, it can be assumed the stress distribution shown on Fig. 41. Type 3 : b eff 20 C f c /g M,fi,c 20 C f ay /g M,fi,a h d h h t w h c,h h c,fi h c f ay,x /g M,fi,a h l x h T a k a f ay /g M,fi,a t f b c b Fig. 41: Elements of the type 3 cross-section for the calculation of the sagging moment resistance according to the Eurocode EN Annex F for the temperatures calculation The method and the tables used to calculate the reduction factors of this section are described in Eurocode EN Annex F. The steel dowel is not considered for the bending resistance. 159

162 Connection Longitudinal shear resistance According to the fire tests performed in the field of this project, the following recommendations can be given: - The mechanical verification of the connection is not requested as long as the temperature in the dowel is lower than 500 C - If the temperature in the dowel is θ dowel > 500 C, the design method at room temperature can be adapted by taking into account these reductions factors: Steel: f y,20 C f y,θdowel = k y,θdowel. f y,20 C Concrete: f c,20 C f c,θdowel = k c,θdowel. f c,20 C - The temperature in the dowel can be calculated by: - an advanced method with a 2D calculation model (e.g. SAFIR FE software) - simple calculation method: see Thermal calculation part Vertical shear resistance The design value of the shear plastic resistance in the fire situation V fi,pl,rd can be calculated by the method described in Eurocode EN Web buckling under fire The web buckling resistance under fire for the type 2 Preco-Beam section can be calculated as in Eurocode EN

163 5 DESIGN EXAMPLES 5.1 PrecoBeam for a Road Bridge Concrete dowel properties Material Concrete C50/60: f ck 50MN / m² f ctk 4,1MN / m² Reinforcement BSt 500S f yk 500MN / m² Rolled steel section: HD400x463 S460ML f yk 460MN / m² Geometry A d c, 0,0145m² b0 0, 105m b i 0, 175m h d 0, 115m e x 0, 25m d s 1 0, 014m d s 2 0, 014m A s 4 2 1,54 10 m² A s 3, m² 161

164 e re 0, 075m n 1 h 2 1, 16m h ' ex 0, 25m 2 l cr 0, 087m b w 0, 30m t w 0, 0358m t f 0, 0574m a 3,263 h po 3,263 0,160 0, 522m hhd400x463 hd 43,5 hpo t f 5,74 16cm

165 5.1.2 SLS Steel dowel Decisive point: sagging moment Section properties for sagging moment: I y = 0,0385m 4 S y = 0,0294m 3 z i.0 -z.dow = 0,95m Internal forces in the beam: LF Vz,k [kn] My,k [kn] Dead loads LM LM DL+LM DL+LM Design stress in dowel toe due to Chapter d 1 A el L t w kc N, w 1 0,147 0,0294 9,774 s, d 1,54 0,95 396MN / m² 0,125 0,0385 0,0358 0,

166 f y 460 s, d 396MN / m² 1,0 1,0 460MN / m² 1,0 M, ser Fatigue design Limit of maximal stresses in the dowel toe / s, fat f y Mf due to Tab. 3 Load combination for s, fat : Dead Load+LM3 (fatigue load model) 6, ,318 0,0294 s, fat 1,54 0,95 324,7MN / m² 0,0385 0,15 0,0385 0,0358 Assessment safe life with high consequence due to Tab. 3, 324,7MN / m² f / 460 /1,35 340,7MN / m² s fat y Mf Limitation of stress amplitude 1, 5 f s, fat y Loading for s, fat : LM3 (fatigue model) 1,66 1 0,194 0,0294 s, fat 1,54 0,95 90,7MN / m² 0,0385 0,15 0,0385 0,0358 s, fat 90,7MN / m² 1,5 f y 1, MN / m² / s, d c Mf Loading for s, d : LM3 (fatigue model); cut yuality II acc. to EC s, d 90,7MN / m² c / Mf 125 /1,35 92,6MN / m² 5.2 Design of concrete dowel Decisive section is at the abutment of the framing bridge. Section properties 164

167 I y = 0,0353m 4 ;S y = 0,0145m 3 ; z i.0 -z dow = 0,4m Internal forces LC Vz,k [kn] Dead load 847 LM DL+LM Vz S y 1,928 0,0145 Ed 0,79MN / m I 0,0353 y. i P e 0,79 0,25 0, MN for one dowel Ed Ed x 20 and steeldowel in CL-shape limits for concept in chaper 3.1 are exceeded. t w 35,8mm 30mm Use of design concept in chapter Shear of reinforced concrete dowel According to Eq. 41 1/ f yk 1,5 d s1 d s2 1 1,3 1,3 f ck f yk ex 2 / 3 sh, re, k 0,9 A d, c f ck / 2 3t / f 3 0,0358 ck yk d d 20,014 0,014 s1 w s2 f 50/ 500 0,606 ; As A d, c 3,08 0, / 3 sh, re, k 0,9 0, ,73MN m sh, re, k / 1/ , ,5 0,014 0, ,3 0,606 1,3 0, / 0, Pry-out of concrete cover Twin dowel reinforcement acc. to Eq

168 w re x re yk s ctk po x k re po t e e e f A f h e ,, l q arccos a e a e a e x x x l 0,30 0, ,25 1 0,522 0,25 0, ,25 arccos l arccos , b e b e b e n n y y y n q 1,0,30 0 m MN k re po / 4,176 0, ,075 0,25 0, ,54 4,1 0,16 0, , ,, Spalling of concrete cover m MN h h b h e l h f e po d w po x cr ctk x k / 1,207 0,16 2 0, ,30 0,16 0,15 0,25 0, ,25 4,1 0, ' cov, Decisive criterion m MN m MN Rd v v k k po k sh Rd / 0,79 / 0,966 1,207 /1,25 / ;1,207 1,73;4,176 max min / ; ; max min cov,,, 166

169 5.2.5 Example for reinforcing the composite dowel Reinforced Cross-section at hugging moment area Reinforced cross-section at sagging moment area Fig. 42: Design example for a PrecoBeam with concrete web and single steel girder 167

170 5.3 Fire Example 1 The simply supported Preco-Beam with the section shown below is used to support a floor in a building. The steel grade is S460 and the concrete grade is C30/37. The span of the Preco-Beam is 8m and the self weight is 7.96 kn/m. The effective width of the concrete slab is 2m. 1 2 IPE 550 Fig. 43: Example 1 Preco-Beam section The bending resistant moment of this section in cold condition is 693 knm (i.e. a maximum distributed total load = kn/m). The fire resistance requested is 90 min (Standard fire resistance i.e. ISO fire curve). The Preco-Beam section is a type 3 section for which the Annex F of EN can be used (see Fig. 41). Requested fire resistance is R90: Verification that the method is applicable (F.3: Field of application): - h = 200 mm 170 mm : OK - b c = 210 mm 170 mm : OK - h.b c = mm² mm² : OK Geometric parameters calculation: - Thickness reduction of the concrete slab : h c,fi = 30 mm - Bottom part of the steel web h l : - a 1 = 14000mm - a 2 = mm² - h = 200 mm 168

171 - b c = 210 mm a1 a 2.t w h l 108,95mm b b.h c c - h l > h l,min = 40 mm - Top part of the steel web : h h = 207,8mm - h l = 98,85mm - Steel flange reduction factor calculation: - a 0 = 0,018.t f + 0,7 = 0,018.17,2 + 0,7 = 1,0096 k 17 0,12 a - a 0 bc 38.bc h = 0,065 0,06;0,12 This reduction factor corresponds to a temperature of the steel = 890 C A a mm² = b.t f f yk a N/mm² = k a.f ay A a mm² = h l. t w f yk a N/mm² = (k a.f ay + f ay ) / 2 A a mm² = h h.t w f yk a3 460 N/mm² = f ay N amax kn Maximum normal force in the structural steel section = A a1.f yk a1 + A a2.f yk a2 + A a3.f yk a3 h c mm h c2 0 mm b c mm b c2 210 mm f cfi 30 N/mm² N cmax kn Maximum compressive normal force in the concrete flange = h c1. b c1. f cfi + h c2. b c2. f cfi Nc,max > Na,max N cf kn Design value of the compressive normal force in the h compressed concrete mm x = compressed concrete height M pl,rk knm EN kn/m (total distributed load) The maximum fire load which can be applied is thus 24kN/m to have a fire resistance of 90 min. Assuming that the cold design of the connection is good, the connection has not to be verified because h h 0 T dowel 400 C 169

172 If it is assumed that the vertical shear resistance is only due to the steel web, the plastic resistant shear force is equal to: V fi,pl,rd = (A a2.f yk,a2 + A a3.f yk,a3 ) / 3 = 462,4 kn > V Sd,fi = 96,5 kn A comparison with advanced calculation model (SAFIR software) was made. The FE models (thermal and mechanical models) give the following results. The temperature distribution after 90 min is shown at Fig the temperature at the dowel base is 270 C - the steel flange temperature is comprised between 890 C and 981 C - the compressive concrete height is well in the zone where T < 250 C The mechanical model constituted by beam FE is shown at Fig. 45. The deflection as a function of time is shown at Fig. 46 for a fire load = 24kN/m. The fire resistance calculated by the F.E. model is nearly exactly the same as the one calculated by the simple model. FILE: ex NODES: 2160 ELEMENTS: 2048 SOLIDS PLOT CONTOUR PLOT TEMPERATURE PLOT TIME: 5400 sec >Tmax <Tmin Y X Z Fig. 44: Example 1: temperature distribution in the section after 90min DISTRIBUTED LOADS PLOT DISPLACEMENT PLOT ( x 1) TIME: sec ex.tem F0 F0 F0 Fig. 45: Example 1: mechanical SAFIR model of the beam Deflection at failure 170

173 1200 SAFIR defl (mm) Deflection (mm) Time (min) Fig. 46: Example 1: deflection as a function of time at mid span of the beam Example 2: In this example, the type of the Preco-Beam section is different from the example 1. Therefore, the thermal response of this section is different and the temperature distribution must be calculated with the method developed for type 1 Preco-Beam sections. It is showed in the next pages. For the mechanical point of view, the general calculation method is the same for all the Preco-Beam sections and the fire resistance can be calculated as in the example 1, by taking into account the different temperature distribution in the section Simple calculation model: The temperatures calculated in the web are given in Tab. 16. Fig. 47 and Temperature ( C) T at z = 0.00 mm T at z = mm T at z = mm T at z = mm T at z = mm T at z = mm T at z = mm Tim e (m in) show two different graphic presentations of the results of Tab. 16. t f = 17.2 t w = 11.1 b = 210 h steel =

174 mm mm mm mm T at z = 0.00 mm T at z = mm T at z = mm T at z = mm T at z = mm T at z = mm T at z = mm 0 min min min min min min min min Tab. 16: Temperatures calculated for example 2 Preco-Beam section Temperature ( C) min 180 min 120 min 90 min 60 min 30 min 15 min z (mm) 172

175 Fig. 47: Example 2: thermal response with simple calculation model: T according to height z Temperature ( C) T at z = 0.00 mm T at z = mm T at z = mm T at z = mm T at z = mm T at z = mm T at z = mm Tim e (m in) Fig. 48: Example 2: thermal response with simple calculation model: T according to time Time K tf t f (mm) t f (mm) k tf t f (mm) tf (mm) min min min min min min min min min min min min min min Time Time K tw t w (mm) t w (mm) k tw t w (mm) t w (mm) min min min min min min min min min min min min min min Time Time K b b (mm) b (mm) b (mm) b (mm) k b min min min min min min min min min min min min min min Time Time T surf20,10, min min min min min min min min min min min min min min Time 20,10,200 (t) 173

176 Advanced calculation model (FE SAFIR software): The model and the results of the advanced calculation are shown on Fig. 49 to Fig. 51. DIAMOND 2009 for SAFIR FILE: t1_ NODES: 1910 ELEMENTS: 1818 SOLIDS PLOT CONTOUR PLOT TEMPERATURE PLOT TIME: 5400 sec >Tmax <Tmin Y X Z Fig. 49: Example 2: temperature distribution in the section after 90min Temperature ( C) min 180 min 120 min 90 min 60 min 30 min 15 min z (mm) Fig. 50: Example 2: thermal response with the advanced calculation model: T according to height z (compare to Fig. 47 ) 174

177 Temperature ( C) T at z = 0.00 mm T at z = mm T at z = mm T at z = mm T at z = mm T at z = mm T at z = mm Tim e (m in) Fig. 51: Example 2: thermal response with the advanced calculation model: T according to time (compare to Fig.48) Temperature ( C) min (simple model) 240 min (SAFIR) 120 min (simple model) 120 min (SAFIR) 60 min (simple model) 60 min (SAFIR) 30 min (simple model) 30 min (SAFIR) z (mm) Fig. 52: Example 2: comparison between thermal response with simple and advanced calculation models: T according to time These results show a very good correlation between the 2 models. A mechanical calculation can be made, based on these temperatures. The fire resistance requested is 90 min (Standard fire resistance i.e. ISO fire curve) For this example, we can take the same assumptions as in example 1: - The steel grade is S460 and the concrete grade is C30/37. - The span of the Preco-Beam is 8m. - The effective width of the concrete slab is 2m. 175

178 Geometric parameters calculation Preco-Beam method: T ( C) z (mm) ky (T ) Tflange TIII, TIII, TIII, TIII, The temperature in the flange is given by linear interpolation between values of the Tab. 11 (Temperatures after different fire durations for different flange thicknesses) Mechanical calculation: A a mm² = b.t f f yk a N/m = k y flange.f yflange A a mm² = h w /3. t w f yk a N/m = (k a1.f ay 1 + k a2.f ay 2 ) / 2 A a mm² = h w /3. t w f yk a N/m = (k a2.f ay 2 + k a3.f ay 3 ) / 2 A a mm² = h w /3. t w f yk a N/m = (k a3.f ay 3 + k a4.f ay 4 ) / 2 N amax kn Maximum normal force in the structural steel section = A a1.f yk a1 + A a2.f yk a2 + A a3.f yk a3 + A a4.f yk a4 h c mm b c mm f cfi 30 N/m N cmax kn Maximum compressive normal force in the concrete flange = h c1. b c1. f cfi + h c2. b c2. f cfi N cmax > N amax N cf kn Design value of the compressive normal force in the concrete flange with h compressed mm x = compressed concrete height h tot mm y anp mm M pl,rk knm EN ==> kn/m (total distributed load) 176

179 The maximum fire load which can be applied is thus 27.3 kn/m to have a fire resistance of 90 min. Assuming that the cold design of the connection is good, the connection has not to be verified because T dowel 400 C. A comparison with advanced calculation model (SAFIR software) was made. DISPLACEMENT PLOT ( x 1) TIME: sec ex2.tem F0 F0 F0 Fig. 53: Example 2: mechanical SAFIR model of the beam Deflection at failure SAFIR defl (mm) Deflection (mm) Time (min) Fig. 54: Example 2: deflection as a function of time at mid span of the beam The FE models (thermal and mechanical models) give the following results: - The temperature distribution after 90 min is shown at Fig The mechanical model constituted by beam FE is shown at Fig The deflection as a function of time is shown at Fig. 54 for a fire load = 27.3 kn/m. The fire resistance calculated by the F.E. model is nearly exactly the same as the one calculated by the simple model. 177

180

181 LIST OF FIGURES AND TABLES Figures Fig. 1:Terminology of a composite dowel 105 Fig. 2: Failure modes for steel dowel (here puzzle shape) 106 Fig. 3: Steps of failure mode activation for a typified load-slip diagram to be supposed in ductile concrete failure [] 107 Fig. 4: ULS failure mechanism of concrete [] 108 Fig. 5: Stress distribution in the contact surface of the steel profile 109 Fig. 6: Geometry of puzzle tooth (PZ) 109 Fig. 7: Forces for the steel design of the connector (PZ shape) with missing stirrup in the concrete dowel 109 Fig. 8: Uplifting force due to eccentricity of the shear joint to the neutral axis at support 110 Fig. 9: Determination of h i for ULS and SLS 110 Fig. 10: Force distribution in the concrete with transversal reinforcement 111 Fig. 11: Geometry of puzzle tooth (PZ) with transverse reinforcement 111 Fig. 12: Forces on the steel part of the connector (PZ shape) 111 Fig. 13: FE model for analysis 112 Fig. 14: Stress distribution due to notch effects 112 Fig. 15: Photo of experimental ULS-failure 114 Fig. 16: Forces and stresses in a critical section at ULS 114 Fig. 17: Comparison of details investigated with details classified according to the EC Fig. 18: Analytic model for principle stresses (in dependency of the angular α)

182 Fig. 19: Principal stress trajectories in steel part of connector (P Rd and P up 0 and σ N influence) 117 Fig. 20: As example, loading histograms due to a vehicle where obviously n = Fig. 21: Rainflow analysis of the load histogram of a vehicle crossing for location in midspan 120 Fig. 22: Additional failure mechanism of concrete identified in testing (Pry out cone) Fig. 23: Reinforcement T-beam cross section 125 Fig. 24: Distance e y between steel girders 126 Fig. 25: Notations for the design concept 126 Fig. 26: Shear interface A d, c and vertical tension area d c A, of concrete 130 Fig. 27: Typified progression of shear behaviour in a concrete dowel 132 Fig. 28: Schematic failure behavior of pry-out criterion 133 Fig. 29: Parameter of the elliptic cone with an elliptic basis and comparison of the test result with the computed results on the basis of a cut image 133 Fig. 30: Load introduction into the twin rein-forcement bars from the composite dowel in ultimate limite state of the concrete 135 Fig. 31: Interaction of dowels with a tight distance in a) longitudinal and b) transversal direction 136 Fig. 32: Scheme of the force flow introduced by the uplifting force 138 Fig. 33: Assignments for the concept due to Tab Fig. 34: Temperature distribution according to Preco-Beam method for type 1 Preco-Beam section 147 Fig. 35: Temperatures as a function of concrete depth z for different fire durations 149 Fig. 36: Temperatures as a function of flange thickness for different fire durations 150 Fig. 37: Temperature distribution for type 2 Preco-Beam section

183 Fig. 38: Temperature distribution according to EN for type 3 Preco-Beam section 155 Fig. 39: Elements of the type 1 cross-section for the calculation of the sagging moment resistance 157 Fig. 40: Elements of the type 2 cross-section for the calculation of the sagging moment resistance 158 Fig. 41: Elements of the type 3 cross-section for the calculation of the sagging moment resistance according to the Eurocode EN Annex F for the temperatures calculation 159 Fig. 42: Design example for a PrecoBeam with concrete web and single steel girder 167 Fig. 43: Example 1 Preco-Beam section 168 Fig. 44: Example 1: temperature distribution in the section after 90min 170 Fig. 45: Example 1: mechanical SAFIR model of the beam Deflection at failure 170 Fig. 46: Example 1: deflection as a function of time at mid span of the beam 171 Fig. 47: Example 2: thermal response with simple calculation model: T according to height z 173 Fig. 48: Example 2: thermal response with simple calculation model: T according to time 173 Fig. 49: Example 2: temperature distribution in the section after 90min 174 Fig. 50: Example 2: thermal response with the advanced calculation model: T according to height z (compare to Fig. 47 ) 174 Fig. 51: Example 2: thermal response with the advanced calculation model: T according to time (compare to Fig.48) 175 Fig. 52: Example 2: comparison between thermal response with simple and advanced calculation models: T according to time 175 Fig. 53: Example 2: mechanical SAFIR model of the beam Deflection at failure 177 Fig. 54: Example 2: deflection as a function of time at mid span of the beam 177 Tables Tab. 1: Notch factor β N for selected geometries

184 Tab. 2: Design of PreCoBeam shear connection Steel design 113 Tab. 3: Partial safety factor for fatigue strength Mf 119 Tab. 4: Loading on a steel part - Steel design for the CL-shape 122 Tab. 5: Design of PreCoBeam shear connection Concrete design 123 Tab. 6: Limits to empirical design approach 128 Tab. 7: Possible failure modes of the concrete dowel 129 Tab. 8: Decisive capacity of concrete dowel 141 Tab. 9: Decisive capacity of concrete dowel 147 Tab. 10: Temperature distribution in a solid slab of 200 mm thickness composed of normal weight concrete and not insulated. 148 Tab. 11: Temperatures after different fire durations for different flange thicknesses 150 Tab. 12: Tables to calculate T surf for different fire durations 151 Tab. 13: Tables to calculate β(t) for different fire durations 152 Tab. 14: Thickness reduction h c,fi of the concrete slab 153 Tab. 15: Minimum cross-section dimensions 155 Tab. 16: Temperatures calculated for example 2 Preco-Beam section

185 FORMULA SYMBOLS A Area A c Area of concrete A cr, v Vertical crack area of a concrete dowel A d, c Area of shear interface A s1 Area of the load side reinforcement in the concrete dowel A s2 Area of the load averted reinforcement in the concrete dowel b w Width of concrete web b r arc length at critical section c y Lateral concrete cover c u Concrete cover beneath the concrete dowel d s Diameter of a reinforcement bar d br Diameter of bending block of reinforcement E s Young s modulus of steel E cm Young s modulus of concrete (average value) e x Longitudinal distance of composite dowels 183

186 e y Distance of dowel strips in transversal direction e re Distance between twin stirrups f ct, m Mean values of tensile strength of concrete f ct Tensile strength of concrete f * cc Compressive bearing capacity of concrete f c Compression strength of concrete f cm, cube Cube s mean value of concrete compression strength f cm Mean value of concrete compression strength f ck Characteristic value of concrete compressive strength f y Yielding strength of steel h d Depth of dowel h 0 Effective depth of steel dowel h po Height of pry-out cone h 2 Internal lever arm of the compression strut corresponding to the centre line of the steel dowel h ' 2 Corrected lever arm h 2 h s, i Distance of centre of gravity to critical section 184

187 n Number of dowel strips in transversal direction P Force in composite dowel P cov Dowel capacity when spalling the concrete cover P cr Dowel capacity at first crack pattern P cr, h Dowel capacity when occurring a horiz. crack P cr, v Dowel capacity when occurring a vertical crack P d Design value of composite dowel P max Maximal load of composite dowel P po Dowel resistance of pry-out cone failure P Rk Characteristic resistance of composite dowel P Rd Design value of the composite dowel s resistance P re Load ratio covered by reinforcement P sh Dowel resistance when shearing of concrete dowel P sh,0 Dowel resistance when shearing of concrete dowel without crossing reinforcement P sh, re Dowel resistance when shearing of concrete dowel with crossing reinforcement P 2 re Portion of dowel resistance if two reinforcement bars are crossing the concrete dowel 185

188 T Shear flow T Shear flow in one composite dowel t w Steel web thickness V pl Vertical shear resistance x 0 Length of dowel base a Angle of inclination Deformation / slip Reinf. ratio in the concrete dowel As A d, c u Shear stresses in ultimate resistance state d c Thickness of plate 186

189 LIST OF REFERENCES [i] Seidl G.: Behaviour and load bearing capacity of composite dowels in steel-concrete composite girders, PhD thesis, Wrocław University of Technology, Raport serii PRE nr 4/2009 [ii] Wurzer, O.: Zur Tragfähigkeit von Betonduebeln. PhD-Thesis, University of the Federal Armed Forces Munich, Germany, 1997 [iii] Code Aster, Fatigue softwar of EDF (Electricité de France) [iv] Feldmann M., Hegger J., Hechler O., Rauscher S.: Untersuchungen zum Trag- und Verformungsverhalten von Verbundmitteln unter ruhender und nicht ruhender Belastung bei Verwendung hochfester Werkstoffe. AiF-Forschungsvorhaben P621, AiF-Nr , Arbeitsgemeinschaft industrieller Forschungsvereinigungen e.v., FOSTA, 2007 [v] Zapfe, C.: Verbundträger mit Betonduebeln. PhD-Thesis, University of the Federal Armed Forces Munich, Germany, 2001 [vi] Mangerig, I.; Burger, S.: Ermüdungsverhalten von Betondübeln. Final report of AiF research project, University of the Federal Armed Forces Munich, Germany,

190 1 Appendix 2: Validation of Preco-Systems in practice 1.1 Pilot Project 1: Busterminal Hornsberg The bus depot at Hornsberg has been run as a fictive project since none of the proposed pilot projects has started in time for the Preco-Beam evaluation and design project. However, lately the Hornsberg project has begun to run. The project is one of many sub projects in a complex project for the entire area. On top of the depot there will be several office and real estate buildings together with the football field which will be re-established on top of the building. The landowner is Skanska Fastigheter AB. Skanska will also construct the depot through another of their sub companies, Skanska Sverige AB. Our client, SL (Stockholms Lokaltrafik), will be running the bus depot when it is ready for use. SL is a community owned company responsible for the public transport system in Stockholm city and the surrounding area. Ramböll is the responsible construction design company for the depot. The old bus terminal at Hornsberg is located under the black roof and the plan is to locate the terminal under the adjacent football pitch and future buildings. Figure A3. 1Aerial photo, showing the old bus terminal under the black roof and the football pitch where the new terminal will be built The new Hornsberg bus terminal will have parking places for about 120 buses and for 120 cars. In the terminal there will also be shops and bus washing areas. On the roof is a football pitch. The area of the 188

191 bus terminal is about 41,450 m 2. The span of slabs between the walls is 17m and the total load (dead and live) is about 18 kn/m 2. The figure below shows the floor plan for the new bus terminal. The floor plan with walls at the same distance (17 m) is perfect for use of prefabricated elements. It is not possible to use traditional prefabricated concrete elements in this project due to the large spans and high loads. The traditional prefabricated elements used in Sweden do not have the shear capacity for the point loads from lorries. The deck of the football pitch must be designed for traffic of lorries, both in the construction phase and also in the future, if the use of the football pitch should be changed. The construction depth of the deck must kept down because the ground water level is very high and it is not possible to increase the overall height of the structure because the surface of the football pitch must be the same as today. The project has been stopped and restarted a number of times for various reasons but now is moving ahead with Ramböll as the Structural Engineers. It is hoped that Ramboll can convince the developer SKANSKA to use the Preco-Beam system but unfortunately it is too late to use it as a complete pilot project. The project is currently in the early stages of design and Ramböll has incorporated the Preco- Beam solution into the current proposals Design The drawings from the pre design phase were used, with the aims and goals to find the areas within which the Preco-Beam system could be used at Hornsbergs bus terminal and also compare the system with traditional building methods. Loads The total weight of the built up can then be calculated to about 11 kn/m 2. The football pitch can also be used as a concert arena and must be therefore be calculated for crowd of peoples (4 kn/m 2 ) and also traffic loads from lorries. In the Swedish building codes the axle load from smaller lorries is 40% of 210 kn = 84 kn. To avoid discomfort with swaying and oscillations the bearing construction of the pitch must be sufficiently stiff, therefore should the deflection in SLS not be higher than L/200 and the natural frequency not lower than 3.5 Hz. Geometries When the layout of the Preco-Beam element was decided it was important to think about geometry of the element. The Preco-Beam element is built up with half steel beams and precast concrete and on that in situ concrete. 189

192 Figure A3.2 Preco-Beam element, with three phases, the top figure is the prefabricated element, the middle with reinforcement and the bottom with in situ concrete. There is lot of restrictions with wide loads while transporting. To avoid transport delays of the elements to the site it is good to keep the width under 310 cm, thus there are no restrictions on time of day for transportation. In the northern part of Europe there is a lack of daylight for 6 months of the year. Therefore the element width was chosen to be less than 310 cm. The thickness of the precast concrete was chosen very thin (50 mm) to minimise the weight of the element. The dead load of the Preco-Beam element above is kg. In the Hornsberg project with two crane tracks, one at both sides, the jib length needs to be about 40 m if it should be possible to launch the elements on the walls. With slightly more than 12 tons on 40m s jib length it is possible to find cranes Fire For fire reasons it was chosen to have two steel girders together and fill the space between the webs with concrete, see Figure A3.2. To keep down the dead load this space should be filled after the element is installed. The construction was then calculated in the fire program SAFIR with an ISO fire, the temperature in the steel beam reached after 90 minutes about 300 C in the case with 15 mm fire insulation, without the fire insulation the temperature reached 900 C. The conclusions of the fire results are that the construction needs to have fire insulation (15mm Fire Board) to satisfy 90 minutes ISO fire 190

193 Figure A3.3 The built up construction with 15 mm fire insulation (15 mm Gyproc Fire Line) / Temperature in the built up construction with 15 mm fire insulation after 90 minutes Figure A3.4 The built up construction without fire insulation / Temperature in the built up construction without fire insulation after 90 minutes Pre design drawings of Hornsbergs bus terminal There are almost 350 elements to be installed in the Hornsberg project, with only two different types, one with span length 17m and the other with 9m. See the floor plan and section Figure A3.5. Figure A3.5 Plan of Hornsbergs bus terminal, Preco-Beam element The Preco-Beam elements could be supported on the walls without any scaffold and also during concreting. Due to the thin precast concrete deck, it is necessary to strengthen up when concreting, also 191

194 when the elements are transporting. It could be done with the lattice of reinforcement, see following figures. Figure A3.6 Section of Preco-Beam element The lattices transport even the shear forces between and connect the precast and in situ concrete so they act as a single unit. The slab of Preco-Beam elements act as a single span deck when it is loaded of dead load, when the in situ concrete has hardened it could act continuously over the walls. So when the element carrying the live load it could act as a continuous beam, which will reduce the deflections. Figure A3.7 shows detail over supporting walls. Figure A3.7 Section of element supported on walls Calculations for Preco-Beam elements A calculation (design) program was developed in MathCAD to design the Preco-Beam elements. This design programme is located within (Preco-Beam Final Report, Background-Document C: Detailed project design Hornsberg). 192

195 Results from calculations Three different slabs were studied with the program mentioned above, all slabs have a span of 17 m, but different loads. The slabs which were studied are: Hornsberg slab with the football arena above. Dead load 18.9 kn/m2 and live load 4 kn/m2. Fredriksdal slab with car parking above. Dead load 6.7 kn/m2 and live load 2 kn/m2. Fredriksdal slab with a road with bus traffic above. Dead load 12.6 kn/m2 and traffic load of bus and lorry (axis load 84/103 kn). In the first one the Preco-Beam element needs to be 2.7 m wide with two halves of HEB 800 beams. In the second one the Preco-Beam element needs to be 3.0 m wide with two halves of HEB 450 beams. In the third one the Preco-Beam element needs to be 2.8 m wide with two halves of HEB 700 beams Conclusions Economic and production issues It seems that on a purely cost front the Preco-Beam system is not favourable compared to the traditional building methods, this conclusion is not a fact but comes from cost estimation from an experienced building contractor (Skanska). For large projects such as Hornsberg and Fredriksdals bus terminals the economic issue could be better because the construction time can be shortened as it is possible to build without formwork and falsework. Issues which could count against the Preco-Beam system are the transportation of the large elements and storage in the building site. Also the assembling could be negative due to relatively heavy elements which need too significant craneage. The lack of experience of this new building product could be an initial problem. It could be difficult to acquire competitive prices from contractors. In general fire resistance is better when a construction is of concrete but it should not be harder or more complicated to fire protect Preco-Beam elements compared with steel. In conclusion, the Preco-Beam is economically a big competitor on large spans or high load projects. For short and medium spans with normal loads Preco-Beam can not compete on the Scandinavian market. For long spans with medium loads it might compete, as the matter of fact we would have had a project with Preco-Beam built in if the Preco Beam project would have started just a few months earlier. In stead modified precast slabs with the TT-section was used. The contractor did not dare to wait for the design guide to get ready since they had a tight time schedule to catch up with. 193

196 Load capacities A significant benefit to the Preco-Beam system is that the elements have a very high load capacity. The ability to vary the distance between the steel beams and also the size of the beams allows the system to be tailored to the exact requirements. Contractors Ramboll has had contact with the building contractor Skanska, one of the leading construction groups in Sweden with expertise in construction and also with Strängbetong which are one of the leading European manufacturers of prefabricated concrete products. Strängbetong follows the Preco-Beam project with interest but it has been hard to get their contribution to the evaluation. Skanska has been a better partner to work with, and we will probably use the Preco beam in one of their projects soon Outlook Work for the future is to convince contractors to use the Preco-Beam system and work with them to fully evaluate the system. This will allow potential site and manufacturing difficulties to be identified and sorted out. 1.2 Pilot Project 2: Road Bridge Vigaun The Preco-Beam construction method has been used for the first time for the road bridge in Vigaun/Austria. With costs about 1030 /m² for 370 m² this bridge is very cost efficient. The following text contains details about the planning process, the design and the construction process of the bridge General The new bridge is located in Bad Vigaun, Upper Austria. The railway line of Österreichische Bundesbahnen (ÖBB) from Salzburg to Vienna is being upgraded for higher speeds. This is why the level crossing of the farm track over the railway line consisting of crossing gates and traffic lights has been replaced by a viaduct. The bridge is to be planned as a three-span structure, each section with a clear width of m, to avoid too much of a barrier effect from the high embankment. The spans have been defined with a width of m. The detailed project design for the Vigaun bridge can be found in (Preco-Beam Final Report, Background-Document D: Detailed project design Vigaun) Design of the viaduct The bridge in Vigaun was planned with a total length of m over three spans, in a straight line over the whole length. The approaching radius measures 25 m at the west abutment. Behind the east abutment the structure opens into a higher grade road where it widens (Figure A3.8). The railway line 194

197 crosses at an angle of 74 gon and requires a clearance gauge of 7.50 m. To minimize the development length of the viaduct, a longitudinal gradient of 10% is projected on the west side and 9% on the east side (Figure A3.9). The road is 4 m wide between the curb stones, with a gradient of 2.5% to the structural axis, on account of the slight width. The reversed roof pitch is favourable for motor vehicles, pedestrians and cyclists alike. The vehicle wheels are always close to the curb stones even when in an unfavourable position. Pedestrians and cyclists can also be guided by the edge of the road surface. The inwards gradient means that there will not be any excessive water accumulation under torrential rainfall or any icing problems here in winter (Figure A3.10). Figure A3.8 Ground view of Vigaun bridge Figure A3.9 Longitudinal section of Vigaun bridge The bridge is designed as a frame as this is the least costly system in terms of production and maintenance. The draft therefore designs the superstructure with fixed mounting of the pillars and abutments. A structural height of 1.10 m in the bridge axis is chosen with spans measuring m wide. The bridge with a slenderness ratio of 23.5 is insensitive to coercive strains from temperature and settlement. The structural calculations show that the fixed mounting of the pillars in the superstructure results in a high concentration of reinforcements in the head area. This is because the hogging moments are transferred in the various traffic load cases. As shown by the traffic load illustrated in position 1 of Figure A3.11, the support moment of the superstructure is transferred almost completely via pillar 1. This also applies to load case 3, but on the opposite pillar side. Load case 2 results in the highest support moment which is transferred only in the transom. Transferring these three different moment progressions would result in a high concentration of reinforcements: this is prevented by arranging concrete joints over the pillars so that the moments are transferred simply by a continuous reinforcement in the superstructure transom. 195

198 Figure A3.10 Cross section of Vigaun bridge Figure A3.11 Relevant load cases for the design of reinforcement in the cross-girder connection Superstructure The cross section is 5.26 m wide, consisting of a structural width of 4.76 m and cornice width of 0.25 m. The structure consists of two Preco-Beams with a width of 2.37 m and a continuous in-situ concrete slab. The precast girders have a mean height of 0.90 m, the in-situ concrete slab is 0.25 m thick. The precast girder consists of a halved rolled girder HE 600x399 grade S460ML. It is embedded in the 0.30 m wide concrete web with the composite dowels. The web is reinforced with a double stirrup per composite dowel. There are two possibilities for connecting the external reinforcement to the concrete web. The concrete can be continued right down to the steel flange. Shrinkage of the concrete causes 196

199 separation of steel and concrete. If the girder is over a road, the flanks of the girder will usually become coated with salty moisture that penetrates in the horizontal crack between steel flange and concrete as a result of the capillary effect. This particular solution should therefore not be used in the interests of durability. Here the concrete web is clearly set back by 10 cm so that the joint can dry out after damp periods. But in the area of the support, the concrete is continued through to the steel flange, as the concrete is under pressure here and no horizontal cracks are formed. The precast part is produced with grade C70/85, as the test specimens planned in C50/60 that were concreted in advance to confirm the rating criteria showed this corresponding strength. In the area of the support cross girders, reinforcement sockets are positioned on the longitudinal reinforcement of the precast parts to guarantee that the girders can be lifted easily into position over the railway section. Front plates are positioned at the end of the steel girders to transfer the pressure. The front plates are cast with a high-strength free-flowing mortar, over a support cross girder. The gap is planned to be 3 cm in size to absorb tolerances. An additional screw connection guarantees absorption of minor traction forces in this area that can result from unfavourable interaction of temperature and traffic loads. Soil conditions, foundations The geological conditions in the Salzach basin are very uniform. At maximum 2 m under ground level, gravel with very high load-bearing capacity is to be found which permits the pressing stresses greater than 400 kn/m² necessary for spread foundations. Spread foundations are provided for all substructures. Under the abutments, the soil has to be replaced down to a depth of 1 metre. Substructures High pillars are necessary on account of the railway's large clearance profile. As the pillars are secured by concrete joints in the superstructure, the pillars are slender in dimensions with a narrow spread foundation. The shape of the 1.00 m thick pillars is 3.20 m wide with a profile 0.20 m deep in the middle section. The pillar width results from the embedding of the Preco-beams in the cross girders. The abutments differ in height. The land rises sharply at abutment 4. The abutment can be set high in the embankment thanks to the favourable ground conditions. Abutment 1 has its foundations in the horizon of the Salzach basin, with a correspondingly higher abutment wall. Both wings are short to obtain uniform visible abutment surfaces. Features The bridge is given a road structure 14 cm in height consisting of 1 cm sealing layer, 3 cm protection layer, 6 cm levelling course and 4 cm top layer. The cornice caps 0.63 cm wide accommodate the controlling system and are connected by dowel anchors to the in-situ concrete slab. The controlling system integrates railings and polycarbonate protection panels. As this is a narrow bridge, a drainage intake is only provided at the ends of the bridge. These are positioned in the road axis because of the reverse roof pitch. The intakes go through the slab straight down and drain in a rough bed hollow in front of the abutments. 197

200 Test setup for rating the composite dowel connection As the design concept for composite dowels were not already finished tests were required. Given the geometry of the Preco-Beam featured here, it must be presumed that the position of the composite dowel close to the edge will reduce the load-bearing ability of the composite joint. Two test series were planned to obtain confirmed values for proof of stability. The first series with three test specimens entails embedding in a concrete web 30 cm wide. As it cannot be ruled out that the rating values may be too low, the trials continued with two additional test specimens with a concrete web measuring 40 cm in width. Figure A3.12 Dimensions of the test specimens of PO.be.1.3 with a concrete web (30 and 40 cm in width) / Evaluation of test results for 30 and 40 cm of concrete web thickness from The steel specimens in grade S460 for the test specimens were made by ArcelorMittal, Luxembourg. The precasting plant Röss near Ingolstadt added the concrete part in the planned grade C50/60. The mean cube strength was 89.8 MPa, corresponding to concrete grade C70/85. The tests were carried out with distance control at the Institute for Structural Engineering at Vienna University of Technology. Figure A3.12 shows the relative displacement line for dowel force. Just 198

201 before reaching the maximum load, a horizontal crack is visible in the concrete web, indicating that the concrete ceiling is coming away below the stirrup. This failure mechanism occurred in all tests, regardless of the concrete web width. The composite joint shows a highly pronounced deformation capability. The smallest characteristic deformation takes place at more than 6 mm. Deviation of the maximum load from the mean value for the series is calculated at maximum 8.2% and is thus within the required 10% criterion. The distinction characteristic curve remains below the 50% value for relative displacement. The rated dowel force for a web 30 cm wide is calculated as P R,d = 1711 kn/m. The test-based rated shearing load coincides with the shearing load in the composite joint from the structural calculation. For constructional reasons, the undercut at the back of the dowel is rounded out with a radius of 10 mm to improve the flow of forces in the steel. In the project, the concrete ceiling is also given a small reinforcement diameter Production Steel girders The steel girders are cut in the steel plant according to the dowel cutting line in Figure A3.13. The necessary camber of the tension-free workshop form is achieved with the three-point bending method. In the workshop, the front plates are welded to the ends of the rolled girders with DHY welds. In the neighbouring coating plant, the steel girders are protected from corrosion with the coating system for protection class S5 according to - 70µm EP zinc dust - 80µm EP zinc phosphate as edge protection - 3x EP + PUR + PUR, 80µm each Figure A3.13 Modified cutting line for the steel dowel / Detail of front plate connection Pre-casted composite element The steel girders were transported by rail from Esch in Luxembourg to Ingolstadt in Germany. Minor damage to the corrosion protection was reworked correctly by the steel contractor. The steel girders were stacked outside the concreting hall with the tension-free pre-cambering. The steel girder acts as 199

202 template for most convenient braiding of the concrete flange reinforcement. Subsequently the steel girder with reinforcement cage is brought into the building where the formwork for the prefabricated element is ready and waiting. After casting the concrete, the prefabricated part undergoes after-treatment to prevent shrinkage cracks. Here the web is liable to cracking as it scarcely has any scope to reduce its length against the very rigid steel girder. The finished Preco-beams are left stacked for 2-3 weeks on precisely dimensioned supports with the pre-cambering of the tension-free workshop form in the pre-casting plant and are then brought to the building site in a convoy of low loaders. Bridge construction The foundations, rising pillars and abutments have been constructed parallel to the Preco-beams. Short steel girders have been concreted into the working joints to the cross girders to take the prefabricated girders. The Salzburg-Wörgl railway line is closed for 4 hours for lifting the prefabricated girders into position over the track. The girders in span 3 are fitted on the day before the line is closed and the truckmounted crane is then moved into position over the track for fitting the girders of the middle span and span 1. After the girders have been fitted, the gap between the head plates of the girders is cast over the pillars, and the cross girders are reinforced and concreted in. Then the in-situ concrete slab is added. The construction of the bridge started in November 2007 and the works were finished, including a break due to the winter period, in September Following the final jobs of work with caps, surface and safety barriers, the bridge were opened for traffic in October Conclusion The rolled girders including cutting, prefabrication and corrosion protection cost 2,100 per tonne free pre-casting plant, in spite of the extremely high steel prices prevailing in spring 2008, thus making them far less expensive than a comparable welded girder. Production of the entire bridge including finishing and without road construction work cost 1,030 net per m² of bridge. These costs include: - Superstructure of the brigde incl. railings, fascia, asphalt and drainage - Piers and abutments - Foundations It does not include: - Embankment work - Road work and drainage of the raod These are low production costs for a narrow bridge structure with high substructures. Together with economic production, the narrow bridge also fits in well with the surrounding hilly landscape. It can be presumed that scarcely any maintenance work will be required for a bridge structure consisting of a framework system without bearings, external reinforcement with small corrosion protection surfaces. 200

203 Figure A3.14 Viguan bridge in service Figure A3.15 Bottom view of Vigaun bridge Work on this pilot project stood out on account of the very open, constructive cooperation among everyone involved. A special word of thanks at this point goes to them all, particularly the principal. Owner: Construction company: Österreichische Bundesbahnen, track engineering division, DL Büro Linz/Austria Angerlehner, Hoch- und Tiefbaugesellschaft mbh, Pucking/Austria Preco-beams: Steel construction: Röss Bau GmbH Fertigteilwerk, Ingolstadt/ Germany Arcelor Mittal Comercial Deutschland GmbH, Cologne/Germany 201

204 Tests: Approval engineer: Institute for Structural Engineering at Vienna University of Technology, Vienna/Austria SBV Ziviltechniker GmbH, Salzburg/Austria Design: SSF Ingenieure, Munich/Germany These new developments have been possible in an environment characterized by an open approach and great desire to optimize building procedures Outlook Thanks to the positive experience gained with Vigaun bridge, the next bridge over the Salzburg-Wörgl railway line is currently being planned using this same method and will be implemented in Project 2: Railway bridge in Lupino/Corse Some more pilot projects were figured out but could not be realised within the Project time as the clients want more information especially about the fatigue behaviour of the composite dowel. Nevertheless one more project will be introduced with its details figured out. Publication in the "Ouvrages d'art" periodic publication LES POUTRES PRÉCO : Une solution économique pour les petites portées, objet d'un programme de recherche européen. This publication presents the Preco-Beam project for the bridge owners in France. It can be downloaded at: Description of the project A first possible project for the application of Preco-Beam is the rebuilding of the Lupino bridge. It was allotted by the local Authority of Corsica to the Engineering Agency SNCF of Marseilles. Sétra intervenes as consultant to advise the bridge owner. Under its mission, Sétra proposed to consider at the stage of the preliminary study an integral bridge, embedded on its foundations by using the Preco-Beam technique. 202

205 Figure A3.16 Computer images of the Preco-Beam integral bridge The existing bridge is a downtown railway bridge in the south of Bastia. It carries the way of railroad which crosses Bastia and towards the south. It crosses an important artery of road traffic. The skew of crossing is important: skew of 52 grades (where 200 grades = pi radians ). The work of replacement of the existing bridge should minimize the blocking of the road traffic under the bridge as well as the railway traffic. The awaited rail traffic in the future is important. On the other hand it is at present tinier during the low tourism season. The clearance under the existing bridge obliges certain exceptional height convoys to use the other way along the sea. It is desirable that the new bridge improves this situation. Counted perpendicular to the axis of the carriageway, the opening of the current bridge is reduced. It is of 7,96m only. So the narrow pavement represents a danger for the pedestrians. It is desirable that the new bridge improves also this situation. The boxes an overall length of 27,40m could be transported whole in two elements of 13,70m joined up on an open surface of building site near to the site. 203

206 1.250 rail (16.10) Ballast Tôle martyre Chaise Maçonnerie existant Figure A3.17 Elevation Figure A3.18 Cross section of the integral bridge proposition with large use of the Preco-Beam connection Figure A3.19 Cross section at the key of the steel structure of the integral bridge proposition 204