52 COLOQUIO S08 COMPORTAMENTO Firof. Kurt Sehdfer PUC/ RU DEPTO. DE ENGENHARIA CIVIL — _PONTIFICIA UNIVERSIDADE CATOLIC RIO DE t SCOO’ ity of Stul tqart University of Stuttgart pu. Kurt Sehitfer; Univ Ing. Dr.-Ing. E-h. Jéry Schlaich Professor Dr Pr CONSISTENT DESIGN OF STRUCTURAL CONCRETE USING STRUT-AND-TIE MODELS 1 INTRODUCTION : Those parts of a structure, to which the Bernoulli hypothesis of Jinear strain distribution applies, are normally designed with an pe ee i a vhile other parts, the more complicated details” of the structure, are mostly désigned by rules of thumb or past experience. Hovever, these "details" are of similar, some- times paramount importance for the structure's behaviour and sofety Therefore a design concept which is consistent for ell types of structures and all their parts is required. To be satisfactory, this concept must be based on realistic physical models. Strut-and-tie models, a generalization of the vell known truss analogy for beams, are proposed as the appropriate approach for designing and detailing cracked structural concrete, which in— cludes both reinforced and prestressed concrete structures. in the: lectures the general Gesigh procecurc 18 oULLinuY and ae 4s shoun, hov suitable.’ strut-and-tie models can be systematically de- veloped. Then, simplified rules for dimensioning the model's elements with uniform criteria for all possible cases are proposed. These me- theds ore applied to a variety of cxamples with sometimes very in- structive results. Concluding, the suitability of the method as a basis of a design code for structural concrete will be discussed. This paper condenses and modifies a more detailed treatment of the subject in the PCI-Journal /1/, adding to it also some new material about nodes and some new examples of application. Without pretending for completeness, some of the authors shall be mentioned here, who prepared the soil on which the method grovs: IL was actually at the turn of the last century, when Ritter /2/ and Mérsch /3/ introduced the truss analogy. This method was leter refined and expanded by Leonhardt /4/, Risch /5/, Kupfer /6/, and others until Thirlimann's Zurich school /7/, with Marti /8/ and Mueller /9/, created its scientific basis for a rational application in tracing the concept back to the theory of plasticity. a Collins and Mitchell further considered the deformations of the truss model-and dérived a rational design method for shear and torsion /10/.i In various applications, Bay, Franz, Leonhardt and Thiirlimann had { shown that strut-and-Lie models could be usefully applied to deep beams and corbels. From that point, the present authors and other members of the Institute for Concrete Structures at the University of Stuttgart began their efforts to systematically expand such mo- dels to entire structures and all structures /11/, /12/. 2 THE STRUCTURE *S B- AND D-REGIONS * Those regions of a structure, in which the Bernoulli hypothesis of linear Sstrain distribution is assumed valid, vill be refered to as B-regions (vhere 8 stands for beam or Bernoulli). Their internal forces or stresses, can be derived from moments, shear and axial forces analysed by means of the statical system of beams, frames, plates etc. If uncracked, the stres- ses are calculated using the bending theory for linear elastic material. For cracked B-regions the truss models or the standard methods of codes apply. : : These standard methods are not applicable to the other regions and details of a structures, where the strain distribution is significantly nonlinear, e.g. near concentrated loads, corners, bends, openings and other disconti- nuities (Fig. 1). Such regions vill be called D-regions, vhere D stands for discontinuity, disturbance or detail. The internal flow of forces in D-regions can bé reasonably well described by strut-and-tie models. Not much accuracy is necessary in determining the dividing sections bet— ween B- and D-regions, These sections can be assumed to lie approximately in a distance h from the geometrical discontinuity or the concentrated joad, where h is equal to the depth of the adjacent B-region (fig. 1). ! Thic’accimntinn de juetified hy Saint—Wenantt] nrinnipte /1/ ay, wh (VE 7 bent bt ; hy thy + Fig. 1a. D-regions (shaded areas) with non-linear strain distribution due to geometrical discontinuities.Ps , c . 4 D n> : D. ‘ >" t i D D a Fig. 1b. D-regions (shaded areas) with non-linear strain distribution " due to statical and/or geometrical discontinuities. > 7 > ; 3 > PRINCIPLES OF STRUT-AND-TIE MODEL DESIGN > > In a strut-and-tie model the struts represent concrete stress fields with A prevailing compression in the direction of the strut, Accordingly the ties > normally represent one or several layers of tensile reinforcement. Hovever, model ties can occasionally also stand for concrete tensile stress fields. 5 i This is evident from models of practically approved details, the structural 4 safety of vhich can be explained only if concrete ties are assumed in pla- oi ). ces “where no reinforcement is provided. Typical exanples ate slabs without Tups or bar anchorages without spiral or transverse reinforcement. If a suitable model of a D-region is known, the forces of the struts and ties will be calculated, thereby satisfying equilibrium between applied loads and inner forces. The struts, ties and their nodes will be dimen- sioned or checked to carry the inner forces as described later. 5 { This method implies that the structure is designed according to the lower bound theorem of the theory of plasticity. However, sinee structural ma terials, in particular concrete, permit only limited plastic deformations, the internal structural system (the strut-and-tie model) has to be chosen in a vay that the deformation capacity is not exceeded at any point, be- fore the assumed state of stress is reached in the rest of the structure. EE OE eS‘J regions this ductility requirement is fulfilled by adapt- and ties of the model to the direction and size of the intel they would appear from the theory of elasticity. In highly ing, Lhe strut nal forces in normaly or lightly stressed regions the direction of the struts and ties in the model may deviate considerably from the elastic pattern without exceeding the structure's ductility. The ties and hence the reinforcement may be arranged according to practical considerations. The structure adapts itself to the assumed internal structural system, This method of orienting the strut-and-tie model along the force paths in- Sicated by the theory of elasticity obviously neglets some ultimate load Capacity vhich could be utilized by a pure application of the theory of plasticity. On the other hand, it has the major advantage that the same model can be used for both the ultimate load and the serviceability check. If for some reason the purpose of the analysis is to find the actual ulti mate load, the model can easily be adapted to this stage of loading by shifting its struts and ties in order to increase the resistance of the structure. In this éase, however, the rotation capacity of the model has to be considered, Strut-and-tie modeling obviously provides the structural analysist vith some freedom of choice vhich can be used to aim cither at the safest, or at the cheapest or at an otherwise optimized solution. Modeling therefore requires some design experience as does the choice of a representative overall statical system or of a reasonable finite element net. The modeling process also covers much of vhat is normally called detailing and therefore requires considerable knowledge about practicable reinforce- ment layout; on the other hand it is just in this field where strut-and- tie models replace experience and guesswork by a more systematic and under- standable design. 4 5 MODELING OF INDIVIDUAL D-REGIONS , Before modeling of a D-region begins, all the forces and reactions acting on the D-region must be evaluated (fig, 2a). The forces or stresses in sections bound by B-regions are taken from 8-region design. New strut-and-tie models can be systematically developed by tracing the flow of forces through the structure, using the "load path” method: The stress diagrams of all the forces applied to the D-region boundaries are subdivided in such a way that the individual stress resultants on op- posite sides of the D-region correspond in magnitude and can be connected by streamlined "load paths" which do not cross each other (fig. 2b). After sketching the load paths smoothly curved and replacing them by polygons, further struts and ties must be added for transverse equilibrium (fig. 2c). Obviously, in some cases the stress diagrams or the loads are not completely balanced with the load paths described; then the load path for the remaining forces enters the structure and leaves it after a U-turn on the same side (fig. 5). Developing o model of a D-region is much simplified, if the elastic stres- ses and principal stress directions are available from an elastic FEM analy- sis. The direction of struts can then be taken in accordance with the meanand main direction of principal compressive stresses, or the more. impo: . tant struts and ties can be located at the center of gravity of the stre: i diagram of typical sections (fig. 4). When modeling, the angles between struts and ties, in particular those with relatively high fore hould be chosen larger than 45° (better 60°) ft in order to avoid incompatibility problens. i Be oto , loadpath_ Beg Wh amet de al bys Fig. 2. The load path method: (a) The structure and its loads; (b) The Joad paths through the structure; (c) the corresponding strut -and-tie model _ if {fF i TT locdpattr a eC c. permet eT, By = Bz a) b) c) “2 Fig. 3. The load path method, including a "U-turn": (a) the structure ey and its loads; (b) the load paths through the structure; (c) the corresponding strut-and-tie modelFig. 4. acne HALE LR Elastic stress trajectories, distribution of elastic stresses ‘and corresponding strut-and-tie model. Fig. 5. The combination of tuo models for the dapped beam is better than the individual modelsCOVeser gs VSeSeverTuvrvTv Set Ci LYN tt Ating models are quite often kinematic, vhich means that the gedmetry of such a model is strictly related to a particular load con- figuration ond cannot be used for other loads without modification, There- fore, the governing load combinations have to be investigated. This disadvantage is not a peculiarity of the strut-and-tie method but is inherent to the nonlinear material properties of cracked concrete. Accordingly, superposition of two models is only possible if the combined model satisfies the requirements on reasonable angles between struts and ties. By combining tvo simple models it is sometimes possible to develop much better but rather complicated models (fig.5). Instead of investigating such a hyperstatic model with representative stiffnesses of the struts and ties, it is normally more adequate to immediately allot the loads to the tvo simple models with an eye on the expected stiffness ratio of the individual: models, Doubts could arise as to whether the correct model has been chosen out of several possible ones (fig. 6). In selecting the model, it is helpful to realize that loads try to use the path with the least forces and de- - ~~} formations. Since reinforcement ties are much more deformable than conérete struts, the model with the least and shortest ties is the best. This simple criterion for optimizing a model may be formulated as follows: EF iL, Eqy = Minimum vhere F, = force in strut or tie i 1 = length of member i ao trai if membs E Sak gees gigior tenn a trae com mertly la omelbeal uc the a) good Coo ! ben Fig. 6. The good model (a) has shorter ties than the bad model (b) f i5 OIMENSIONING THE STRUTS, TIES AND NODES Sul Reinforced and not reinforced ties Normally tie forces are carried by reinforcement. Its cross section fol lovs From the tie force in the ultimate limit state and the desigi yield strength of the steel. For crack distribution the reinforcement shall be distributed over the tensile zone. Crack widths cen be analysed if the reinforced tie is con- sidered as a prismatic reinforced bar with an effective concrete area. The tensile strength of concrete should only be utilized for equilibrium forces if no progressive failure must be expected and if local failure | zones are assumed. Thereby restraint forces and microcracks have to be taken into account even in "uncracked" concrete..further, some positive experience with similar details and loading should be available, 5.2 Conerete struts or compression stress fields To cover all cases of compression stress fields three typical configura- tions are sufficient: a) The fan shaped stress field (fig. 7a) is an idealisation of a stress field with negligible éurvature. It does not develop transverse stresses, hidd WWE T T | Il | WM Lost, Ios nto id ‘O- ‘o— a) na b) ¢) Fig. 7. The basic compression fields: (a) the "fan"; (b) the "bottle"; (c) the “prism”,b) The bottle shaped stress field (fig. 7b) ‘vith its Gulginy stress qtrajeclories develops considerable trenverse stresses: Compression in the bottle neck and tension further avay. The transvei tension can cause longitudinal cracks and initiate an : early failure. It is therefore necessary to reinforce the stressfield in the transverse direction or to consider the transverse tension when , determining the failure load of the strut. The transverse Lension can be determined from a strut-and-tie medel of the stress field. Diagrams - simplify its dimensioning (fig. 8). ce) The prismatic or parallel stress field (Fig. 7c) is a frequent special case of the preceding two stress Fields. aie - Or i ea dealing D* = rrbiaxial compression in baile neck. a . és 7c lh Og Lemay aff transverse reiniorcement > cegreewi.in the “ belly region go vacracked, plain concrete 42. th confinement 5 without Oy ~ 28 ot Og 22 Beb/o Fig. 8. Dimensioning plane bottle shaped stress Tields: (a) diagrams giving safe pressure values p_ with regard to cracking and crushing of plain unreinforced conérete stress fields and yielding of trans- verse reinforcement; (b) geometry of the stress field; (c) model and reinforcement layout of stress field with transverse rein— forcement . ae a ow SRY VV ee~ 10 - The fan shaped and the bottle shaped stress fields are frequently found in'D-regions vhere concentrated loads are introduced into a structure and spread out. The prismatic stress field is typical for B-regions. The strength of the concrete in compression stress fields depends to a con- siderable extent on the miltiaxial state of stress and on disturbances From cracks and reinforcement. For practical dimensioning of all kinds of stress- fields, the following simplified design strength values (4 are proposed: f£, = 1,0 fg for an undisturbed and uniaxial state of stress, ey = 0,8 F,4 for compression fields with cracks parallel to the compression stresses, 7 f£4 = 0,6 fy for compression fields with skew cracks, where’ fgg denotes the concrete compressive design strength for uniaxial compression according to the cade of practice. The design values given above for cracked coneréte are meant for structu— ral conerete, whose crack widths are limited in the usual manner. The va- lues for cracked concrete shall also be applied for concrete with trans- verse tension below the expected tensile strength and if tensile reinforce- ment is crossing thé stress field. Skew cracks are not expected if the theo- ry of elasticity is closely folloved during modeling. Hovever, skew cracks ‘fay also be left over from a previous loading case vith a different stress situation, The increase in strength due to two- or three-dimensional states of com- pressive stresses may be taken into account if the simultaneously acting transverse compressive stresses are reliable. fath Aaccolimg om me of the Gren thegih Vatu Barve de Hnfrra danidiay an ane nfuthe aiven etrength values, hath traneuerse din rections must always be considered, 5.3 The nodes The nodes are in reality regions where forces are deviated over a certain length and widths. The "smeared" or "continuous" nodes, where wide concre- te stress fields join each other or with closely distributed reinforcing bars, are not critical; it is sufficient to ensure safe anchorage of the reinforcing bars in the smeared node and to catch the oukermost fibres of the deviated compressive stress field with reinforcement (fig. 9). On the other hand, where concentrated forces are applied the deviation of forces is locally concentrated in "singular" or “concentrated” nodes. These have to be carefully designed in order to balance the oncoming for- ces of the struts and ties without excessive deformations resp, cracks. Though numerous cases of different singular nodes exist, in most cases their forces balance each other inthe node region through direct compres- sive stresses. Also bond is essentially a load transfer via compressive stresses which are supported by the ribs of the steel bar and by radial pressure in bent bars. However, in many cases also concrete tensile stres- ses develop transverse to the model plane ("third direction"),b) Fig. 9, "Smeared nodes" 1 and "singular nodes* 2 in a D-region; (a) Model, (b) and (c) stress fields and node regions a) The stress distribution in singular nodes is mostly so complicated, that it cannot be analysed individually with bearable expenditure. But experien- ce shows, that some types of nodes and details are repeated again and again in quite different structures and can be designed safely by simplified rules: a) The geometry of the node has to be tuned with the,applied forces. There- fore reinforcement anchored in the node should be distributed over a certain height u with due regard to the widths of the oncoming stress fields and the magnitude of their forces; further, it should be ade- quately distributed in the transverse direction in otder to keep trans- verse tensile stresses lov. b) The average compressive stresses in the node region boundaries have to be checked to be less than : fag = 1,1 fag in nodes where only compression struts meet, thus creating a two- or threedimensional state of compres~ sive stresses in the node region, f1, = 0,8 f,4 in nodes where tensile bars are anchored and an allov- ance in strength must be made for bond action. fs Suitable node region boundaries and the corresponding compression stres- ses can easily be determined as shown in the typical nodes in figs. 10 - 13. 7 As for all nodes, also the stresses of the oncoming struts have to Y be checked according to section 5.2. : c) Safe anchorage of ties in the node has to be assured: Minimum radii . of bent bars and anchorage lengths of bars are selected following : the code, The anchorage must be located within and "behind" the node < (fig. 11 and 13). The anchorage begins vhere the transverse compres- ‘ion stress trajectories meet the bar and are deviated. The bar must extend to the other end of the node region. If this length is less than required by the code the bar may be extended beyond the node region and introduce some of its forces from behind, > Vv,- 2. Node NL (fig. 10) is typical for a node of compression struts in a cor- ners “Two alternative node region boundaries are shown for the same node, both leading to the same results. The node is safe, if Foy Fp Md yy Usd Foye Node _N2 (fig. 10) is a combination of tuo nodes Nl. It is realistic and oAvEAient_to choose a large enough, a, = a, cosad, sin a, cosal, sinad, in order that the bearing pressure @, = 1,1 fy governs the node's design. Nodes N3 and N4 (fig. 10) are typical’ for loads or support forces applied + fo" the je ST"a structure vith a chord force running parallel to this edge through the node. Normally the concrete compressive stresses Feo and Gy = 1,1 Foy govern the. design. Node N5 (fig. 11) applies to the anchorage of ties far from the edges i.e. Insta the structure in the plane of the model. As for all nodes vith ties, the anchorage length must be checked, Node N6 (fig, 11) is typical for end supports. The height u in deep beams ShOGTa be chosen uw0,15h =£0,2h £0,221 vhere h = height of D-region 1 = span of deep beam. Single layer reinforcement shall be placed near the lover edge, vhere the deviation forces are largest. Checks include Sy and Gy £0,8 Foy. Node N7 (fig. 11) is typical in the tension chord of beams or deep beans. Tif,~"vell distributed bars shall be chosen as reinforcement for the tie T, and they shall embrace the tie T,. Concrete stresses G = 0,8 f_, vill ny 1 c cd rérely be decisive. Node _N8 (fig. 11) is a mixture of ‘the nodes N1 and N6, and therefore maximum GampFEssion stresses between those of both node types are proposed: Sea? Fon = OB Ey Besides, the rules for typical node N6 apply: . Node_N9 (fig. 11) is composed of tuo nodes NS; checks are accordingly. This fiode-IS typically over the support of continuous beams and normally also covered by the code rules (check of the beam's cross-section for M, N and V, bearing pressure, anchorage of chord reinforcement). 0 (fig. 12) i8 checked via the admissible radius of the bent bar. or: i in the third direction must be for re (a, < t, fig. 13), the transverse tension fovered by transverse reinforcement designed Bl eos: T=4 eet,-bB- Fig. 10. Nodes without anchorage of reinforcement4 - > [> fea 4 7" fan Fig. 11, Nodes with anchorage of reinforcement (continued next page)(ts - a — D NS rn sa I I it, www Fig. 12. Node with deviation of reinforcement eVvVvvVwVVYvw Ss> 16 - Fig. 13 Treatment of local pressure in typical nodes N2 and N6, applicable also to other nodes Local pressures G,, may be tolerated up to BO ae eee Con apa, ~ a, fea $33 fed