CR e Title no. 85-S23 Studies of Disturbed Regions near Discontinuities in Reinforced Concrete Members 99 by William D. Cook and Denis Mitchell Discontinuities caused by abrupt changes in cross-sectional dimen- sions or by concentrated loads result in what is termed “disturbed re- sions’? due 10 the disturbance in the flow of the stresses around the discontinuities. The use of simple strut and tie models to analyze and design disturbed regions such as corbels, dapped end beams, bearas with openings, and deep Beams is explained. Strength predictions of 4 corbel and a dapped end beam using strut and tie models are com- pared with test results. Complete response predictions using a non- linear finite element computer program are used 10 verify the strut ‘and tle models and to provide further insight into the response of disturbed regions. Keywords: beams (supports; brackets; corbels; deep beams: dealing: finite tlement method: openings Feinforced concrete; shear properties: sruetu analysis structural design ‘There are many situations in the design of reinforced concrete members where discontinuities such as abrupt changes in cross-sectional dimensions or the presence of concentrated loads or reactions cause disturbances in TN (a) Smoly supported beam (b) Dapped end beam on compet ak, (@ Wal wih concentrated (C) Deep bean 008 Fig. I—Examples of disturbed regions modeled with concrete compressive struts and reinforcement tension ties 206 the flow of the internal forces. These disturbances in the flow of internal forces around discontinuities result in “disturbed regions,”” as shown in Fig. 1. For exam- ple, the concentrated reaction acting on the beam shown in Fig. 1(a) interrupts the uniform field of com- pressive stresses in the concrete and causes a disturbed region due to the fanning of the compressive stresses into the support. The nib of the dapped end beam shown in Fig. 1(b) causes a disturbance in the flow of the forces resulting in fanning of the compressive stresses in the full-depth portion of the beam and con- centrated compressive stresses or struts in the nib. The disturbed region in the corbel is characterized by high local compressive stresses at the beam bearing area, with these stresses fanning into the column, The con- centrated load acting on the deep beam shown in Fig. 1(o) is transmitted directly to the supports by concen- trated unidirectional compressive stresses in the con: crete. Since the flow of the forces is transmitted by compressive struts, the entire deep beam is considered a disturbed region. The concentrated loads acting on the wall shown in Fig. 1(d) cause a disturbed region due to the fanning of the high compressive stresses into the uniform field of compressive stresses. It is not appro- priate to design these disturbed regions using the usual beam theory that assumes that plane sections remain plane While clastic finite element analysis may be used to determine the stresses in the concrete prior to cracking, this analysis method may not be appropriate for de- sign, since considerable redistribution of stresses may occur after cracking. To provide a tool capable of pre- dicting the complete response of disturbed regions, a microcomputer program was developed’ that combines two-dimensional, nonlinear finite element analysis and ‘Received Mar. 24, 1987, and reviewed under Institute publication polices, Copyright © 1988, American Concrete Institute. All igs reserved, including the! aking of copies unless permission is obtained from the copyright propr rs. Pertinen’ discussion wil be published inthe Januaiy February 1989 ACT ‘Struciural Journal resived by Sept 1 1988 ACI Structural Journal / March-April 1988ACI member Wiliam D. Cook isa research associate in the Department of Gil Engineering and Applied Mechanics at McGill University, Montrel (Quebec, Canada. His reseurch interests include nonlinear analysis of rein forced concrete, shear design, and progressive collapse of reinforced concrete Dens Mitchell. FACI, iso professor in the Department of Civil Engineering and Applied Mechoncs at McGill University, Montreal, Quebec, Canad, He lea member of the joint ACI-ASCE Committe 445, Shear and Torsion; ACI Commities 408, Bond and Development of Reinforcement; and E 901, Shot srships: and the Canadian Concrete Code Committe, the compression field theory"? to account for the strain softening of the cracked concrete. In the design of disturbed regions, the flow of the forces can be idealized by a truss model in which the zones of concentrated compressive stresses are repre- sented by compressive struts and the principal rein- forcement is represented by tension ties. This strut and tie model or truss model was pioneered by Ritter' and Mérsch‘ and was further refined by Thiirlimann et al.,* Marti,’ and Schlaich and Schafer.‘ Schlaich and Schafer have suggested a strut and tie design procedure for dis- turbed regions that involves choosing compressive struts oriented to approximate the flow of stresses obtained from an elastic analysis. The truss design approach suggested by Marti? uses a limiting concrete compres- sive stress in the struts of 0.6f. The strut and tie model developed by Collins and Mitchell”"” and adopted by the Canadian Standards Association Concrete Code (CAN3-A23.3-M84)"" uses the strain softening concept to determine the crushing strength of the struts. RESEARCH SIGNIFICANCE This paper compares the analyses of disturbed re- gions using simple strut and tie models with the predic~ tions of nonlinear finite element analyses and with test results. The purpose of this research program is to pro- vide design guidance in the use of strut and tie models. NONLINEAR FINITE ELEMENT COMPUTER PROGRAM A nonlinear finite element program, FIELDS, was developed’ for use on a personal computer. This pro- gram, which uses two-dimensional plane stress ele- ments, uses the compression field theory in the evalua- tion of the element tangent stiffnesses. In analyzing the complete response of a member, the program starts with the self-weight of the member and increments the applied loading in steps chosen by the user. For a given load step, the program uses an iterative solution tech- nique in which the out-of-balance loads and the tan- gent stiffness from the last displaced configuration are used to compute incremental displacements. Iteration continues until convergence requirements are met, that is, until the norms of out-of-balance loads and incre- mental displacements are within tolerance limits. Triangular and quadrilateral elements having up to nine nodes are used to model the reinforced concrete behavior. The reinforcement and the cracking are as- sumed to be smeared uniformly within the element. To ACI Structural Journal | March-April 1988 account for the significant nontinearities within an ele- ment, which could result in wide variations of stiffness, up to four-by-four Gauss points are permitted in the numerical integration of the tangent stiffness matrix. Fig. 2 illustrates the manner in which the stresses are evaluated at each Gauss point. The principal tensile strain ¢, the principal compressive strain ¢,, the strain in the x direction ¢,, the strain in the y direction ¢,, and the principal compressive strain direction @, are inter- related by the requirements of strain compatibility [see Fig, 2(b)]. It is assumed that the principal compressive stress direction coincides with the principal compres- sive strain direction 6. For a given state of strain in an element, it is a sim- ple task to determine the average steel stresses f,, and f,, corresponding to the strains ¢, and ¢,, from the stress- strain relationships of the reinforcement [see Fig. 2(c)]. It is not as easy to determine the principal stresses f., and f, in the cracked concrete. The principal compressive stress f is not only a function of the principal compressive strain e; but also depends on the principal tensile strain ¢,, if cracks are present. As ¢, increases, fy decreases—an effect that is called strain softening. From the work carried out by Vecchio and Collins’ the compressive stress-strain rela- tionship [Fig. 3(a)] for cracked concrete can be written Sa = BS [: () - (:)] ° where ~ 0.8 + 0.34(6,/e!) <1.00 8 t vate (c) Determining stresses at a Gauss point ‘corresponding to stran state Fig. 2—Evaluating stresses at Gauss points in quadri- lateral element 207(a) Dotarricing average concrete compressive stessilce fiom strains €, and ¢2 (©) Determining average concrete tensile stress fe, ‘rom stain ey Fig. 3—Stress-strain relationships for diagonally cracked concrete and! = strain in concrete at peak compressive stress. After cracking, the principal tensile stress in the con- crete varies from zero at a crack location to a maxi- mum value between cracks. Fig. 3(b) illustrates the av- erage principal tensile stress-strain relationship for the concrete as suggested by Vecchio and Collins.’ The av- erage principal tensile stress is given by fife, «, then f, Si VI -¥ 200¢, @ where E, is the initial tangent concrete modulus, f, is the concrete cracking stress, and ¢,, is the strain in the concrete at cracking. Fig. 4 shows that this average principal tensile stress fa may be limited by yielding of the steel reinforcement across the cracks or by sliding along the crack inter- face, Between the cracks, the concrete and the steel are assumed to have average values of stress [see Fig. 4(c)], while at a crack the tensile stress in the concrete is zero, the stress in the reinforcement is a maximum, and a shear stress v,, may exist at the crack interface [see Fig. 4(d)]. Based on the interface shear transfer tests con- ducted by Walraven,” an approximate expression to limit the shear stress along the crack was developed.” This expression was further simplified as @) cn af (a) Cracked reintorced (0) Transmting shear across crack Tmeriace 2 Vive (e) Average suosses (0) Sonsos at crack betwebn crocs itetace Fig. 4—Investigating stresses at crack interface where f! is in psi units and w and q are in inches (for SI units of MPa and mm, replace the 2.16 by 0.18 and the 0.63 by 16). The average crack width w is assumed to be equal to the average crack spacing times ¢ and a is the maximum aggregate size. Since the states of stress in Fig. 4(c) and (d) are statically equivalent, it is possi- ble to investigate whether yielding of the reinforcement across the crack (i.€.5 fixer OF Sonor equals f,) oF sliding of the crack interface (i.¢., Yj, €qUAIS Ving.) Will result in a value of J, lower than that given by Eq. (2). It is important to recognize that the resulting consti- tutive relationships for the cracked concrete are not only nonlinear but also interdependent. For a given strain state, incremental strains are applied to the exist- ing strains to find the corresponding changes in stress, permitting the tangent stiffness to be determined. DESIGN USING STRUT AND TIE MODELS. ‘This section summarizes the strut and tie model ap- proach of the 1984 Canadian Concrete Code" for the design of disturbed regions. The first step in the design is to visualize the flow of the forces and identify regions of constant fields, fans, and concentrated compressive stresses in the concrete [see Fig. 5(a)]. As shown in Fig. 5(b), a truss model consisting of concrete compressive struts and reinforce- ment tension ties is used to realize the flow of the forces in the disturbed regions of the dapped end beam. The intersection of the truss members delineate nodal zones (the nodes of the truss) of multidirectionally stressed concrete. The nodal zone stresses are limited to 0.85¢,f° for nodal zones bounded by compressive struts and bear- ing areas to 0.754,f! for nodal zones crossed by ten- sion tie reinforcement in one direction and to 0.60¢.f for nodal zones crossed by tension ties in two direc- tions. The application of these nodal zone stress limits is illustrated in Fig. 5(a). The Canadian Concrete Code uses load factors of 1.25 for dead load and 1.50 for live load. Material resistance factors ¢, = 0.6 for concrete ACI Structural Journal | March-April 1988| tea € 0.78 6s effective area ‘anchoring tension te compression tal fan Nee € 06 el (a) Strut and tie model strut anchoring horizontal tension tie strut representing fan region strut representing uniform fag. oF comoression (b) Truss idealization Fig. 5—Strut and tie modeling of a dapped end beam and 4, = 0.85 for reinforcing bars are used in lieu of member strength reduction factors. The nodal zones must be chosen large enough to in- sure that the nodal zone stresses are less than the nodal zone stress limits. If it is assumed that on each face of a nodal zone only uniform axial compressive stresses exist, then from equilibrium these stresses must be equal on all the faces. The geometry of the truss is de- termined by locating the nodes of the truss at the points of intersection of the forces meeting at che nodal zones Fig. 5 shows that the compressive stresses fanning into the bottom of the main vertical tension tie, the concen- trated compressive stresses in the nib, and the uniform field of compressive stresses are each represented by compressive struts located to simulate the flow of the compressive forces in the concrete. After determining the forces in the truss members from statics, the required area of tension tie reinforce- ment is chosen and the capacities of the compressive struts are checked. The cross-sectional area of a com- pressive strut is determined by the dimensions of the nodal zones at the ends of the strut. The stress limit in the compressive struts can be derived by setting «’ and € equal to 0,002 in Eq. (1) resulting in ft Fame = 08 + 1706, SOK @) where \ is a factor to account for low-density concrete (A = 1.00 for normal density, 0.85 for structural semi- low density, and 0.75 for structural low-density con- crete). ACI Structural Journal | March-Ap 1988 (d) Beam with single ledge support Fig. 6—Strut and tie models and truss idealizations for brackets and corbels ‘The maximum compressive stress the strut can carry decreases as the principal tensile strain e, increases. The principal tensile strain ¢, is determined from strain compatibility in the regions where a tension tie crosses a compressive strut. From Mohr’s circle for strain in Fig. 2(b), if e, is replaced by the strain ¢, in the tension tie, if @ is replaced by the angle a, between the tension tie and the strut, and if it is assumed that the compres- sive strain in the strut is 0.002, then 4 + 0.002 6 tan? a, The strain in the tension tie may be conservatively as- sumed to be f,/E,. In investigating a region of a com- pressive strut that does not have a tension tie crossing it, the compressive stress limit is taken as Ad, Jf. Care needs to be taken in providing adequate an- chorage for the tension ties, particularly where the ten- 209(@) Strut and ve mode! (2 Fete clement mesh and dellected shape at faite nes amie = (6) Truss ideatzation AS : see wn Na 7s ALN oan Sate Eee om? Re y em Se EE sh unt, APY Ey oe — ‘ifi] [fe () Reintoxcoment detais anc "Siure condticns (6) Precicied principal stresses and strans Fig. 7—Corbel Test Specimen C-1 (1 mm = 0.04 in. 1 KN = 0.225 kips; 1 MPa 0.145 ksi) sion tie is anchored in the nodal zone region. In these regions, adequate anchorage must be provided, either by sufficient embedment length or by mechanical an- chorage. In addition, the tension tie reinforcement must be distributed over an effective area of concrete such that the stress in the nodal zone (taken as the tension tie force divided by the effective area anchoring the tie) is Jess than the nodal zone stress limit. The effective area of concrete is equal to the area of concrete surrounding the tension tie reinforcement and having the same cen- troid as that reinforcement [see Fig. 5(a)]. BRACKETS AND CORBELS Strut and tie models, together with truss idealiza- tions for a variety of brackets and corbels, are given in Fig. 6. The double-sided corbel resists the loads by compressive struts feeding directly into the column, as shown in Fig. 6(a). A tension tie is required to resist the out-of-balance forces at the loading points. Fig. 6(b) il- lustrates the flow of the forces in a single-sided corbel The strength of these connections depends greatly on the anchorage details of the tension tie reinforcement. Recommended details for anchorage of the tension ties are given in References 14 and 15. For example, a structural steel angle welded to the main tension tie re- 210 inforcement to enable the development of the yield strength of the tie will also serve as a bearing surface and serve to armor the outer corner of the bracket or corbel. In the ledger beam shown in Fig. 6(c), the hanger steel transmits the vertical component from the in- clined compressive struts toward the top of the beam. This hanger steel can be spread over a short length of the ledger beam and should be provided in addition to the shear and torsion reinforcement required in the ledger beam. The analysis of the double-sided corbel test specimen C-I! using a strut and tie model is illustrated in Fig. 7. In the test, the horizontal force was 20 percent of the vertical force acting on the corbel. Using the measured material properties (all material resistance factors are taken as 1.0), the main tension tie can supply a force of A.J, = 800 mm? x 444 MPa = 355 kN (79.9 kips). If one assumes that the yield force of the tension tie will govern the corbel failure, then from the statics of the truss [see Fig. 7(b)] the predicted failure loads are 408 KN (91.7 kips) vertically and 81.6 kN (18.3 kips) hori- zontally. The fanning compression strut has its maxi- mum stress in the nodal zone at the top of the corbel. The nodal zone stress under the 50 x 300 (11.8 in.) x ACI Structural Journal / March-April 1988imens 25 mm (1 in.) bearing plate is 408 KN / 50 x 300 mm = 27.2 MPa (3.94 ksi). The nodal zone stress limit is 0.75f; = 0.75 x 40.4 MPa = 30.3 MPa (4.39 ksi). Hence, it is predicted that yielding the main tension tie will initiate the failure of the corbel followed closely by crushing of the concrete in the top nodal zone. The actual reinforcement details for this corbel are shown in Fig. 7(c) and 8(a). Additional closed horizontal ties, having an area of at least 50 percent of A, and distrib- uted within the top two-thirds of the corbel, have been provided in accordance with code-detailing requirements! for brackets and corbels. In the analy- ACI Structural Journal | March-April 1988 T-benm «itn web opening (Specimen H-1) Fig. 8—Photographs showing reinforcing cages and failure conditions of test spec- sis using the simplified strut and tie model, this addi- tional reinforcement was neglected. Failure occurred in Specimen C-1 at a vertical load of 502 KN (113 kips) by concrete crushing under the bearing plate after large strains were recorded in the main tension tie and after the occurrence of severe spalling of the concrete cover surrounding the bearing plate [see Fig. 7(c) and 8(a)]. Fig. 7(d) shows the finite element mesh and the de- formed shape predicted by the program FIELDS close to failure. Fig. 7(e) illustrates the principal stresses and strains in the concrete predicted by the finite element analysis. In the modeling of this corbel, stiff truss ele- att“orem te“ (a) Rectanaar cap Cea if (@) T-beam with opening near 109 Fig. 9—Strut and tie models and truss idealizations for beams with dapped ends and openings ments were used to simulate the presence of a steel loading block that was attached to the top of the steel bearing plate. The two elements in the top row sur- rounding the bearing area were given thicknesses of 300 mm to simulate the spalling of the 25 mm of concrete outside of the bearing plate. Fig. 7(d) shows that the mesh also simulated the likely concrete spalling zones in the unarmored region outside of the bearing plate. All of the reinforcement in the corbel and column was modeled in the finite element analysis. Fig. 7(e) shows that the predicted flow of concrete compressive stresses is somewhat different than that assumed in the simple strut and tie model. Due to the presence of the addi- tional horizontal reinforcement and due to the vertical cracking in the corbel, these compressive stresses are more curved toward the outer surface of the corbel and become more concentrated as they funnel into the col- umn. The program FIELDS predicts failure to occur at a vertical load of 450 KN (101 kips), that is, at 90 percent of the actual failure load, by yielding of the main ten- sion tie reinforcement. The large predicted tensile strains are evident in Fig. 7(e. An additional finite element analysis was carried out with the horizontal loading located directly at the top of the steel bearing plate. This resulted in a predicted failure load of 525 kN (118 kips), which demonstrates the sensitivity of these types of connections to small changes in load eccentricities. DAPPED ENDS AND OPENINGS Fig. 9 illustrates the application of strut and tie models to members with dapped ends and members with openings. The change in the flow of the forces from a uniform compressive field to a fan that delivers compression to the bottom of the main vertical tension tie in a dapped end beam is illustrated in Fig. 9(a). The role of this vertical tension tie is to lift the force to the top of the beam, thus permitting the forces to flow into the support by means of a compressive strut. Horizon- 212 tal tension ties are required to balance the outward thrusts of compression at the support reaction and at the bottom of the main vertical tension tie, To anchor the horizontal tension tie at the level of the support re- action, the tie is welded to an angle at one end and the other end must continue into the region of fanning compressive stresses to achieve proper development. The anchorage of this tie by bond stresses is simulated by the two struts radiating from an anchor point as- sumed to be located 0.5¢, beyond the center of the ver- tical tension tie. Note that this reinforcement must continue well beyond this assumed anchor point into the full-depth portion of the beam. Fig. 9(b) shows a strut and tie model for an inclined dapped end. The inclined tension tie creates large local stresses at the bend and must be anchored appropri- ately above the support reaction (e.g., a welded plate) to create a nodal zone. ‘The manner in which an opening affects the flow of, the forces in a uniformly loaded T-beam is illustrated in Fig. 9(c). The uniform field of diagonal compression is interrupted by the opening, causing higher shear stresses in the section beneath the opening and requir- ing an increase in the amount of stirrup reinforcement in this region. To make use of the full depth of the sec- tion beyond the opening, a vertical tension tie is pro- vided to lift the shear force to the top of the beam. This vertical tension tie enables the force to flow into the support reaction area by means of a direct compressive strut. Fig. 8(4) shows the reinforcing cage for T-beam Specimen H-1' with a rectangular web opening just un- der the flange, as well as the appearance of this uni- formly loaded beam near failure. The steeper inclina- tion of the compression field just under the opening due to the higher shear stresses can be seen from the crack pattern. The cracking also delineates the com- pressive strut going from the top of the vertical tension tie to the support reaction area. Fig. 10(a) shows the strut and tie model used to ana- lyze the rectangular dapped end Test Specimen D-1.' The reinforcement details are shown in Fig. 10(c) and 8(b). This specimen spanned 3.2 m (10.5 ft) with a point load applied at midspan. The additional horizon- tal bars, provided in the nib to satisfy the code detail- ing requirements,""" were not included in the simple truss idealization. In predicting the failure load, it is assumed that the material resistance factors are equal to 1.0. If it is assumed that failure will be governed by yielding of the vertical tension tie, then the force in this, tie will be A,f, = 800 mm* x 445 MPa = 356 kN (80 kips). The forces in the other truss members can be de- termined from statics [see Fig. 10(b)] To investigate the capacity of the nodal zone region at the top of the vertical tension tie, first assume that the concrete cover spalls off down to the center line of the stirrups in this highly stressed region. Hence, the stress on the top face of the nodal zone anchoring this tension tie is 356 KN / 110 x 210 mm = 15.4 MPa (2.24 ksi). This is less than the nodal zone stress limit of 0.75f! = 0.75 x 29.8 = 22.4 MPa. ACI Structural Journal | March-April 1988() Strat and te model (6) Finite erent mesh ard detected shape at fate (6) Truss ideatzation and member forces mn kN (6) Predicted pricipal stresses (©) Rebveccement cetais and latxe corstions Fig. 10—Rectangular dapped end Test Specimen D-1 (1 mm 0.225 kips; | MPa = 0.145 ksi) The most critical section of the compressive strut going from the support to the top of the vertical ten- sion tie is located at the interface with the top nodal zone. Since it is assumed that the faces of the nodal zone are equally stressed, then the compressive stress fia in the strut at the face of the nodal zone is equal to 15.4 MPa. The maximum compressive stress fianae that this strut can carry can be determined from Eq. (5) if it is assumed that the strain ¢, in the vertical tension tie crossing the strut is at the yield strain (2.23 x 10°). Hence from Eq. (5) «+ 0.002 * Tan a, 0.00223 + 0.002 tan? 47.3 deg = 0.00223 + = 0.00583 and from Eq. (4) a a Fons = O84 Tie, ~ 0.8 + 170 x 0.00883 16.6 MPa (2.41 ksi) As can be seen, the presence of the tension tie signiti- cantly reduces the compressive strength of the strut, Since the compressive stress f. is slightly less than f.smns ACI Structural Journal / March-April 1988 Chtepatetst = . -| TANS DE TS RRISTS tS ty som RRR TS TS TST S| (1 Predicted principal stains 0.04 ins TKN = the strut and tie model predicts that the tension tie will just yield before the strut crushes An investigation of the bond characteristics of the longitudinal tension tie at the bottom of the main ver- tical tension tie shows that the available tension tie force is just sufficient. It can also be shown that all of the other truss members are adequate; thus, it is pre- dicted that failure will occur when the vertical reaction is 260 KN (58.4 kips) by yielding of the main tension tie. Failure occurred in Specimen D-1 when the reaction reached 307 KN (69.1 kips). The conditions of the beam. at failure are shown in Fig. 10(c) and 8(b). The results of the finite element analysis are shown in Fig. 10(4),(e), and (f). All reinforcement was included in the finite element modeling. The spalling of the con- crete cover outside of the stirrups shown in Fig. 10(c) and 8(b) was simulated in the finite element model by reducing the thickness of elements near the vertical ten- sion tie and by adjusting the mesh geometry [see Fig. 10(d)]. The failure load predicted by FIELDS is 270 KN (60.7 kips), that is, 88 percent of the actual failure load. ‘The finite element analysis predicts a curving of the compressive stresses that flow from the top of the ver- tical tension tie to the support [see Fig. 10(e)]. The principal tensile strains in Fig. 10(f) show that the ver- tical tension tie yields before the concrete compressive strut crushes, as was predicted by the simple strut and tie model. Since the finite element analysis correctly ac- 213(€1 Cominvous beam with compressive strut and tan action Fig. 11—Strut and tie models and truss idealizations for deep beams counts for the changing stiffness of elements, as well as the presence of all reinforcement, it more accurately models the stress flow. It therefore correctly predicts that the shear is transferred by both compressive struts and fields into the nib. DEEP BEAMS Fig. 11 illustrates some strut and tie models and truss idealizations for some deep beams. The simply sup- ported deep beam is easily modeled with a statically de- terminate truss, as shown in Fig. 11(a). The design must insure that premature failure due to an inadequate amount of tension tie reinforcement, insufficient an- chorage of tension ties, crushing of bearing areas, and crushing of compressive struts does not occur. In addi- tion to the main tension tie reinforcement, the Cana- dian Concrete Code!’ requires a minimum reinforce- ‘ment ratio in the transverse and longitudinal directions of 0.002 to control cracking and to increase the ductil- ity of the member. This small amount of uniformly distributed reinforcement is usually neglected in de- signing deep beams using the strut and tie model. Ex- amples of the design of simply supported deep beams are given in References 9 and 10. Strut and tie models and the resulting statically in- determinate truss idealizations for continuous deep beams are shown in Fig. 11(b) and (c). To solve for the truss member forces, it is necessary to account for the relative stiffness of the members. If a significant amount of transverse reinforcement is provided in a deep beam, then the load will be carried by both com- pressive struts and fans. The truss idealization for such a case is given in Fig. 11(c) where it has been assumed that the transverse reinforcement between the loading points and the reactions is modeled as a single tension tie. Beams containing transverse reinforcement that are in the transition region between deep beam action and shallow beam action can also be modeled in this man- ner. As the beam becomes more shallow, a uniform compressive field could develop between the fanning 214 regions near the supports and loading points. The de- sign of these more complex transfer girders can be car- ried out by assuming a simplified truss idealization in which the relative stiffness of the members is assumed. Guidance on the use of more sophisticated truss models is given by Rogowsky and MacGregor."* Two of the continuous deep beams tested by Ro- gowsky, MacGregor, and Ong” demonstrate the influ- ence of shear span-to-depth ratio on the response. Specimens $/1.5 and $/2.0 have shear span-to-depth ratios of 1.5 and 2.0 respectively. The reinforcement details are shown in Fig. 12. Both specimens have sim- ilar concrete strengths 39.6 and 41.1 MPa) and are 200 mm thick. Fig. 12 also shows the deflected shapes, the principal stresses, and the principal strains predicted by the program FIELDS at ultimate load. Specimen 5/1.5 failed in the north shear span when the applied load in this span reached 858 kN (193 kips). After strengthening the north side, the south shear span failed at a load of 879 KN (198 kips). The predicted ca- pacity is 890 kN (200 kips). Specimen $/2.0 failed at corresponding loads of 677 kN (152 kips) and 693 kN (156 kips), which compare well with the predicted ca- pacity of 695 KN (156 kips). Since both specimens had the same total amount of transverse reinforcement in each shear span, the 21 percent drop in load-carrying capacity between these two specimens is due to the dif- ference in their geometries. This geometric difference resulted in a larger direct compressive strut action in Specimen $/1.5. As Fig. 12 shows, the compressive struts in both specimens become wider between the loading point and the reaction areas. In both cases, failure is predicted to occur in the interior shear span by yielding of the transverse reinforcement, followed by crushing of the concrete, The negative moment near the central support results in larger principal tensile strains [see Fig. 12(d)], thus softening the concrete and reduc- ing its compressive strength. These large tensile strains also reduce the principal tensile stress to zero in some elements [see Fig. 12(0)]. CONCLUSIONS The simple strut and tie models presented in this pa- per provide conservative estimates of the ultimate ca- pacities of the disturbed regions investigated, Nonlin- ear finite element analysis provides a means of predict- ing the complete response of reinforced concrete ‘members containing disturbed regions. ‘The test results and the predictions using both strut and tie models and nonlinear finite element analysis, demonstrate the need to account for the following im- portant features of disturbed regions: a) Since the ultimate capacities of corbels and dapped ended beams are sensitive to small changes in load ec- centricities, it is necessary to model carefully the details of the bearing and loading areas. b) Unrestrained concrete outside of bearing areas and plain concrete cover has the potential to spall and therefore should be neglected in the analysis of the ul- timate failure conditions ACI Structural Journal | March-April 1988ee Lf own [| Setmrvome SPLOMEN 5/20 | | Ee fli PAA nse = Es I wun LY i 1 (@) Pradieted pineal suains Fig. 12—Continuous deep beams tested by Rogowsky, MacGregor, and Ong” (mm = 0.04 in 1 MPa ©) The termination of reinforcing bars results in sig- nificant bond stresses and variation of force along the length of the bar. In the finite element analysis, this was accounted for by reducing the area of steel along the development length. In the design using strut and tie models, it is necessary to insure that the reinforce- ment is detailed such that the required forces in the tension ties can be achieved. It is noted that although in the strut and tie analysis it is assumed that anchor- age takes place at nodes, it is essential to continue the reinforcement beyond these nodes to achieve the devel- ‘opment of this reinforcement 4) The ultimate capacity is sensitive to the manner in which nodal zones are detailed. Elements modeling the region around main tension ties should be chosen so that the elements are approximately the same width as the effective zone around the tie. Nodal zone dimen- sions and elements modeling nodal zone regions should ACI Structural Journal / March-April 1988 0.145 ksi) have a thickness consistent with the available anchor- age details (e.g., the difference between open and closed stirrups can be substantial). The nonlinear finite element analysis provides a more accurate prediction of the response of disturbed re- ns than simple strut and tie models. This more ac- curate analysis tool, together with experimental results, has enabled the development of simple strut and tie models, suitable for design, for a variety of disturbed regions. ACKNOWLEDGMENTS. ‘The authors gratefully acknowledge the Natural Sciences and En- incering Research Council of Canada for funding this research, The experimental research reported in this paper was carried out in the Jamieson Structures Laboratory of the Department of Civil Engi- neering and Applied Mechanics at McGill University. The authors would like to thank Lionel Lemay for testing the beam with web openings. 215NOTATION ‘maximum aggregate size area of tension tie reinforceme initial tangent modulus of elast of concrete modulus of elascity of concrete compressive strength of conerete (from a standard cylinder test) stress in concrete at cracking average stress in x reinforcement stress in x reinforcement at a crack average stress in y reinforcement stress in y reinforcement at a crack yield stress of reinforcement average principal stress in concrete principal compressive stress in concrete compressive strength of diagonally cracked concrete development length of reinforcement shear stress on cracked surfaces ‘maximum shear stress permitted on a crack due to aggregate interlock shear stress in reinforced concrete element w= crack width a, = angle between tension tie and compressive strut 8 = factor accounting for the reduction in compressive strength of cracked concrete ~ strain in concrete at peak stress f° (from a standard cylinder test) &, tensile strain in concrete at cracking = tensile strain in tension tie ¢ = strain in the x direction = strain inthe y direction 6, = principal tensile strain = principal compressive strain 4 = angle of inclination of principal compressive strain to the x A= factor to account for low-density concrete ( = 1.00 for normal-density, 0.85 for structural semilow-density, and 0.75 for structural low-density concretes) = fesistance factor for concrete (6, = 0.60) 4 = resistance factor for reinforcement (9, = 0.85) REFERENCES 1. Cook, W. D., "Studies of Reinforced Concrete Regions Near Discontinuties," PAD thesis, McGill University, Montreal, July 1987, 153 pp. 2. Collins, Michael P., and Mitchell, Denis, “'Shear and Torsion, Design of Prestressed and Non-Prestressed Concrete Beams," Jour 216 eS ral, Prestressed Concrete tnstitute, V. 25, No. 5, Sept-Oct. 1980, pp. 32-100 3. Vecchio, Frank J., and Collins, Michael P., “The Modified Compression-Field Theory for Reinforced Concrete Elements Sub- jected to Shear," ACI JouRNAL, Proceedings V. 83, No. 2, Mar Apr. 1986, pp. 219-231 4. Ritter, W., "The Hennebique Design Method (Die Bauweise Hennebique),"” Schweizerische Bauzeitung (Zirich), V. 33, No. 7, Feb. 1899, pp. 59-61 5. Marsch, E., Concrete-Steel Construction (Der Eisenbetonbau), ‘Translation of the 3rd German Edition by E. P. Goodrich, MeG Hill Book Co., New York, 1909, 368 pp. 6. Thirlimann, B.; Marti, P.; Pralong, J.; Rit, P.; and Zimmetli, B., “Application of the Theory of Plasticity to Reinforced Concrete (Anwendung der Plastizitaetstheorie auf Stahibeton),” Institute for Structural Engineering, ETH Zurich, 1983, 252 pp. 7. Marti, Peter, “Basic Tools of Reinforced Concrete Beam De- sign,” ACI JOURNAL, Proceedings V. 82, No. 1, Jan-Feb. 1985, pp. 46-56, 8. Schlaich, J., and Schafer, K., ‘*Konstruieren im Stahlbeton- bau,"" Beton-Kalender 1984, Wilhelm Ernst und Sohn, Berlin, 1984, p. 787-1004, 9. Collins, M. P., and Mitchell, D., “Chapter 4—Shear and Tor- sion,”” CPCA Concrete Design Handbook, Canadian Portland Ce- ment Association, Ottawa, 1985, pp. 4-1—451 10. Collins, Michael P., and Mitchell, Denis, “A Rational Ap- proach to Shear Design—The 1984 Canadian Code Provisions,” ACL JouRNAt, Proceedings V. 83, No. 6, Nov.-Dec. 1986, pp. 925-933. 11, “Design of Concrete Structures for Buildings,"* (CAN3-A23.3- M84), Canadian Standards Association, Rexdale, 1984, 281 pp. 12, Walraven, Joost C., “Fundamental Analysis of Aggregate in- terlock,"” Proceedings, ASCE, V. 107, STI1, Nov. 1981, pp. 2245: 270, 13, Collins, Michael P., and Mitchell, Denis, “Evaluating Exist Bridge Structures Using the Modified Compression Field Theory Strength Evaluation of Existing Conerete Bridges, SP-88, American Concrete Institute, Detroit, 1985, 268 pp. 14. ACI Committee 318, “Building Code Requirements for Rein- forced Concrete (ACI 318-83),"" American Concrete Institute, De- troit, 1983, 111 pp. 15, Mattock, Alan H.: Chen, K. C.; and Soongswang, K., “The Behavior of Reinforced Concrete Corbels,” Journal, Prestressed Concrete Institute, V. 21, No. 2, Mar-Apr. 1976, pp. 52-77 16, Rogowsky, D. M., and MacGregor, J. G., “*Design of Rein forced Concrete Deep Beams,"” Concrete International: Design & Construction, V. 8, No. 8, Aug. 1986, pp. 49-58, 17. Rogowsky, David M.; MacGregor, James G.; and Ong, See Y., “Tests of Reinforced Concrete Deep Beams,” ACI JouRKAL, Pro: ceedings V. 83, No. 4, July-Aug. 1986, pp. 614-623, ACI Structural Journal | March-April 1988