by Shamim A. Sheikh “ous analytical models available in the literature far the conte fof concrete by resilinear tee ore studied. ‘These moses plied 10 the specimens tested by the author as well 25 by other Investigators to predict the results. Leadinst on these specimens in ‘lugs axis! and combined aval and bending, monotonie 2s wll 33 mined using the anshtitsl models. Experimental resuls are red with results predicted by various models I & concluded {natin addition to the commonly acknowtedsed variables sich at ‘he amount of Iatera! reinforcement and steel sttength, two other sariables play important role in determining. the behavior of confined eonsrete. These.varsbles are the distribution of the. fon “Meat steel around the core perimeter and the resulting tle con Liye] on, and the spacing of tes, Better diatibution of steel sad let spacing of ties along the volumn fonsitudinal exis (for the same amount of relavorcement) recut in higher concrete surength and shri. matical results fom the model accounting for these vate lables show the best agreement with experimental results, Circular spirals confine concrete much more effec: tively than rectilinear ties, and the mechanism of con finement for circular spirals is"better understood than for ties. But their’ relative ease in detailing makes the use of ties more attractive than spi When reinforced concrete sections are subjected to large deformations typical of seismic motions, their ability to carry load depends primarily on the behavior of confined concrete within the core. Numerous studies'"" have been reported on the behavior of con- rete confined by rectilinear ties. Several analytical models with various degrees of sophistication have been proposed, Some models predict only the ascend- ing part of concrete’s stress-strain curve, while others precit the curve up to a certain point on the descend- ing A few models predict only concrete strength ana sviresponding strain. Several variables have been considered in these ls. The amount of lateral reinforcement recei most attention. Some of the other variables which ap- 296 Pear in these models are strength of plain con: steel strength, distribution of longitudinal steel and the resulting tie configuration (steel configuration), te spacing, and section dimensions. In this paper the available analy sical models!" are applied to predict the resulis of tests reported by various investigators." The tests include speci mens subjected to axial as well as combined axial and bending loads. One'set of specimens was tested under constant “axial. loads sal of moments,” which case the.envelope moment-curvature curves Wee calculated and compared with the experimental results cis generally accepted thet the envelope curve under cyclic loading is almost identical to the curve obtained from the monotic lo: SUMMARY OF MODELS Almost all the analytical models for confineméat aft — based’on experimental results, Most experimental d2tt Were obtained from small-scale tests on simple tie com figurations. A summary of previous representative is given in Table 1, In most of the tests the ratio of the area of the core bounded by the center line of the perimeter tie to the gross area of the specimen We small compared with the values commonly used i practice, ‘The models proposed by the following researche"* are used in this paper: Chan,’ Kent and Park,t Roy aed Sozen," Sargin,” Sheikh and Uzumeri,!" Soliman 6” Yu,” and Vailenas, Bertero, and Popov.” Variable considered in these models are summarized in Table Direct application of these-models to all tests consié- ered in this study was not possible. Therefore, @ fe assumptions were made to adopt models to tests WHET ever required. Brief details of these models are 8i** below. The assumptions made for the application © ication poli Pout net ps abate | | weé the te sd by speci Vand ander wed | sults, ander ained tare cata oof ain ets and and bles sid few me ae ets ree asst Sr eae maple eet ite enn “sot model are also explaned thas been allempted iar ial ey ene fan niece ee iis Tea tee gee pee ete er carn astae es ace ye otf permete te shelkh and Uzumeri (1980) In this model the increase in strength of confined concrete Was calculated on the basis of effectively con- u fined concrete area, which is less than the core concrete area enclosed by the center line of the perimeter tie. ‘The theoretical-cum-empirical relation used the test re- sults from 24 to 12 in, (305 mm) square, 6 ft 5 in. (1960 mm) high reinforced concrete columns tested. under monotonic axis! compression. Minimizing the cumu- lative error for all 24 columns was the only criterion used in determining the empirical constants. No effort Was made to minimize the difference between exper imental and analytical values for individual columns, A stress-strain relationship for confined concrete, as, shown in Fig. 1(a), was proposed. The first part of the curve, up to a strain of ¢,, is a second degree parabola. Between ¢, and ¢,, the curve has a horizontal straight line portion. Beyond c, the descending part of the curve is @ straight line which is suggested 10 continue decreasing to 30 percent of the maximum stress, after which a horizontal line represents the concrete behav- Table 1 — Summary of available tests : - Devas of te speamens Loe Site of section, Loins Reteence|__Researener | Number | “in sieel | ge pecene 9 [king 0986) wee | as corner bars | 0267.60, 10 | Ring 0946) 16 | oxi ‘comer bars | 030.58 5s | chan 0955) 9 | 6x6 | 06.092 | 4 comer bare | 0.84 4.13 7 | 6x36 0.92-0.96 | 4 comer bars | 0.85. 4.35 2 | Gin as oo7 | a bare | ee. 3.28 2 | Brester and j 2 | oexs os] sean | oat 0.62 iver igen | | axe os} stan ui | Pater 36s) 2 fone | | oan | a 1 gx y 12 bare foe | wx bars 12 | Roy and Sozen | sss 4 corner bars 368 . 1 | Benero an 2 | aa one Feippa cise «| os | 3x3 4comer bars | 0-26 72 | ssa none 6 | bexan 4 comer bart +6 | Husson 1566) 2 | axe lossoe | “tes | 0-032 2 | 6x6 | osn065| soa 0 069 17 |soiminandys | 3 | sxe |osao) ates | 9-040 (38 1 | es | nasise | ¢ comer bars | 0.6 - 3.43 1 | 6x3 ot | & comer bare 16 | ors 0.93 | 4 corner bars as 1¢ | shahang Ranesn |r| 2x 083 | none o- Lor «sro 18 | Somes 11970 a2 | ee jossos2 | none | 0.67: 9.08 8 | seein ast a1 | ses Jossoss| sone | 0.55. 5.32 4 | Bordewe and 1s | sxs [orzo | none 0. 3.70 Pisdort cst) 4 | since oo | one | 18s: 6 3 | Bonar a995) 4 | oses” |'ossos0| none ° so | sxs | oss.oss | «come bers | 0 6.58 7 | ase et at nh | aor 16 | 0468.02 | 4 comer bare | 0.96. 2.86 asm 6 | sxe 0.70 | 4comner Sars | Gor 1.72 19) | Valens, 3 | 10x10 o7 | “aban 7 - Bertero and 3 | sxe 096 | tars 140 Popor (1972) 3 | 1x10 078 | none 3 | sxe 095 | fore 1s | sheikh ana 9 | en 078] fare | 0.60. 2.32 nuns (1980) 6 fer 078 | izbars | 1.60. 240 s | ae a7 | sean | 0.76. 230 21 | seou ett t | uteita [ores | foam | 134 293 38) | nasin |orsas | roan | 140. 2097 | | & * Table 2 = List of variables for different models Model Sheikh and Yolumettic ratio of fateral ste! to cone Unumesi® steel around the core perimeter and the resting chant Roy and Sozen'* (0 te spacing | Soliman and Yu" -| Sargia”™ Volurieire ratio of lateral steel to concrete core, ratio of width of conere Variables distribution of longitudinal ie contiguration, tie spacing, characteristics of lateral steel, ane strength of plain concrete. Volumetric rato of lateral see to constete coe, Volumetric ratio of tetera steal to concrete sore and ratio of setion dimension a of te stel bar, tie spacing, and section geomety sore toe spacing, stel strength, and strength of plain sonszev. snd Park! | Volu ratio of lateral steal 10 concrete core, ratio of width of concrete 0re (0 te spacing and strength of plain concrete, Vatlenss, Berero, and Popov jor. Four equations were given to define the curve completely. The four values are K, ratio of the strength of confined concrete to the strength of un- confined concrete; ¢4 and ,, the minimum and max- imum strains corresponding to che maximum siress in concrete; and fq, the strain corresponding to 85 per- cent of the maximum stress on the descending part of the curve. The following variables were considered in the de- velopment of this model: volumetric ratio.of lateral re- inforcement to vonerete core; distribution of longitu- inal steel around the core perimeter and-the. resulting tie configuration; tie spacing: characteristics of lateral steel; and strength of plain concrete. The amount of longitudinal steet was recognized as having no signifi- cant effect on the behavior of confined concrete. Chan (1955) oo, Chan's! equatisns to predict the stfength of confined conereie and the corresponding strain were based on tests in which toad was applied with a small eccentricity on 6 X 6 X 11% in, (152 152 x 292 mm) and 6 x 3% % 52 in, (152 x 92 x 1321 mm) specimens. Two equa- tions, one for K//K, equal to K,, and the other for ¢, the ultimate strain in the column when concrete catties the maximum load, were suggested as functions of the Volumetric ratio of tie steel to concrete core. No other variables were believed (0 affect the strength of con- fined concrete and the corresponding strain value. The suggested trilinear curve for unconfined and confined concrete is shown in Fig. 1(b). OAB approx- imates the curve for unconfined concrete. In plain con- crete, slope of BC Is‘negative. In confined concrete, with suitable lateral binding, A, can be positive, with e, attaining values much greater than for unconfined con- crete, Roy and Sozen (1964) Tests" on $x$ % 25 in, (127 x 127 x 635 mm) prisms led Roy and Sozen to conclude that the confinement provided by rectilinear ties does not enhance the strength of the confined concrete. They proposed a aon Volumettic rato of lateral stel to concrete sore, rio ofa steel to the area of cross section, sizes of We bar and longitudinal bar, ratio | of core dimension to ie spacing, steel strenath of longitudinal nd sirength of plain son stress-strain relationship of concrete, shown in Fi H(@), in which the coordinates of the peak point are fy 002, where f, is the strengih of concrete in a plain specimen. This means K,= 1.0. An equation for the strain value corresponding to°50 percent of the max’ ‘mum stress was suggested to define the descending pai of the bilinear curve. The only variables considered 10 affect the ductility of the confined concrete are the vol umetric ratio of tie steel to concrete core and the ratio. Of the shorter side dimension of the compressed con- cretevsection to the tie spacing, Soliman and Yu (1967) A stress-strain. relation for confined concrete was Proposed” in which four points were defined (Fiz. 1(@)]. Four equations were empirically developed usin. the data obtained from eccentric compression tests 09 3% 6 in, (76 * 152 mm), 4 * 6 in, (102 » 152 mm) and 5x6 in. (127 152) specimens. ‘The initial part of the proposed relation consisis of a parabolic curve with the peak value of_f,, ci: The second part of the relation {@ horizontal straight line up to a strain value of f« The final part is a straight line with a aegative slope representing the descending branch. This was defined by a strain value e, corresponding to 80 percent of thé maximum stress, All test specimens used simple tie arrangements with only one tie at one level, The variables considered were area of tie steel bar, tie spacing, and section geomettY: For the purpose of applying the proposed equations 1° the tests with complex tie configurations, the laters) steel is assumed to be distributed around the core Pe rimeter. This assumption is necessary 0 calculate * reasonable value of area of tie bar to be used in the proposed equations. Sargin (1971) sere proposed to predict the sith ng <6 ie and one to predict tiie" Corresponding. sical eu The empiric equations are based on analysis of results ©! tests conducted on $ x 5 x 25 in. (127 x 127 x 635m) nag.2 (pacimens under concentric and eccentric compression loads. As observed previously by Hognestad et al., sargin also found no significant difference in strength between concentrically and eccentrically loaded speci- mens. The following variables are recognized in Sar- gin’s equations: volumetric ratio of lateral reinforce ment 10 concrete core, ratio of tie spacing to the width fof concrete core, yield strength of steel, and concrete cylinder strength. The strain value £,, at peak stress is also assumed to depend on the sirain gradient at the section in addition to the above variables, A general ‘equation was proposed to give a continuous stress- sirain curve of confined concrete as shown in Fig. 1(), Kent and Park (1973) On the basis of existing experimental evidence, Kent and Pgrk proposed the stress-strain curve shown in Fig. 1(2) for concrete confined by rectilinear ties, The suggested curve combined many features of previously proposed curves, The ascending part of the proposed curve is unaf= fecied by confinement. Pezk stress and strain values are given as £/-2nd 002. The falling branch of the curve is suggested 10 be a straight line whose slope is i function of concrete cylinder strength, ratio of width of confined concrete to spacing of ties, and ratio of volume of tle steel to volume of concrete core: The descending part of the curve extends to.0.2 4, beyond whieh @ horizontal line represents the concrete beh: jor, The data reported by Rey and Sozen," Bertero and Felipa,* and Soliman and Yu" were used in developing the proposed curve This model suggests thai confinement due to recti- linear ties does not enhance concrete strength, which means K,= 1,0. The effect of confinement én ductility is recognized. Vallenas, Bertero, Popov (1877) The stress-strain curve proposed by Vallenas et al." IFig. 1(g)) is similar in form to Kent and Park's model,* shown in Fig. 1(f). The main difference be- tween the two is the inclusion of concrete strength en- hancement due to confinement in the model proposed by Vallenas et al. The ascending part of the curve is represented by a second degree parabola. The descend- ing branch of the curve consists of a straight line with 8 given slope extending to 30 percent of the maximum stress, beyond which the curve continues in the form of a horizontal straight line, The following variables are included in the model: volumetric ratio of lateral Heel to conerete core, ratio of area of longitudinal steel 'o the area of cross section, sizes of tie bar and lon- situdinal bar, ratio of tie spacing to core dimension, crete strength, and strength of tie steel. A major Zitference between this model and the one proposed by Sheikh and Uzum that in the Vallenas model strength enhancement of confined concrete is consid fred proportional to the volumetric ratio of longitu- inal steel to concrete, whereas in the Sheikh and ACI JOURNAL / Julv-Auqust 4ero (2) Sveth one Uromail) ose (0) Sarg Contines concrete R000 sou “soe “a0 “ (8) Kent one Pork!) (9) Voltenas et oi !8) Fig. 1 — Some proposed stress-strain curves for con- ‘rete confined by rectilinear ties : Uzumeri model strength enhancement depends upon Gistribution. of the longitudinal steel APPLICATION OF MODELS The seven confinement models discussed above were Used to predict the resulls of tests reported by various i iwe Cee Fig. 3 — Comparison of experimental K, values with the K, values calculated from various models for 24 col- researchers. These tests included specimens under mon- ‘conic axial compression and eccentric compression, ‘One.set of. specimens was tested under axial load. and reversed-cyclic bending, in which case-it was attempted to predict the moment-curvatuce envelope curves from the analytical models. Monotonic Compression The results of tests reported by Roy and Sozen," liman and Yu,” Vallenas et al..” and Scott et al.” were predicted using the Sheikh and Uzumeri model. All seven models were applied to predict the results of tests reported by Sheikh and Uzumeri* in which a large va- riety of test variables were studied. ‘These variables could be isolated by dividing the 24 specimens into subsets where only one factor varyed between different specimens. Various arrangements of longitudinal and lateral reinforcements, tie spacing, amount of longi- tudinal and lateral steels, and characteristics of steel were studied for their effects on the behavior of con- fined concrete. The amount of longitudinal steel was found to have no significant effect on properties of confined concrete. The four steel configurations tested are shown in Fig. 2. The large size of these columns, appropriate ratios of the area of core to gross area of the specimens, and complex steel arrangements resulted in large increases in concrete. strength (up 10-70 per- ceni) and ductility due to confinement, In most other tests reported in the literature, small increase in con- crete strength was observed due to rectilinear confine- ment. It has been demonstrated" that in specimens with only four corner bars, the effectively confined concrete area at the critical section between the ties would be very small compared with the core area bounded by the center line of the tie. This will result in poor confinement of the core concrete, Small spec- 300 Fig. 4 — Comparison berween experimencal K, values for various tests and K, values calculated from the model proposed by Sheith and Uzumeri imen sizes, simple steel arrangements, low volumetcie ratios of lateral steel to concrete core, and low ratios of core area io gross area of columns would result in only small increases in confined concrete strength and ductility, Checking the models against these test results will not distinguish between the accuracies of various models. A comprehensive study of the models is, thete- fore, based on test resulis ceporied by Sheikh and Uzumeri. The significant values related to concrete behavior were determined for all 24 columns, using the seven models. The experimental and analytical strength val ues, represented by K, are compared in Fig. 3 for six models. The analytical K, values calculated using the model proposed -by Sheikh -erid Uzumeri"® are com pared with the experimental K, values in Fig. 4. The strength of unconfined conerete in the columns was observed to be equal to 0.85 f/. The analytical curves from soime of the models are compared with the expetitnental curves in Fig. 5 for nine representative columns, which include all four steel configurations. Size of che tie bar, spacing of tits volumetric ratio of tie steel t0 concrete core ay ratio of longitudinal steel area to gross cross-section area @ and arrangement of longitudinal and lateral steel a? also given in Fig. 5 along with the curves for each cok umn, The models proposed by Kent and Park,! ROY and Sozen,” Sheikh and Uzumeri,"* Soliman and Yur and Vallenas et al." were used for this compariso™. Comparisons between experimental and analytical curves are presented for only nine columns. A vet! similar end is observed in the other 15 columas, # can be seen in Fig. 6 where areas under experimental curves are compared with areas under analytical curves. For ail the columns, these areas are calcula up to the strain values at which the experimental curve terminate. In the case of a few columns, tests were te inated somewhat prematurely, for fear of damaging the equipment. For other columns, tests were termi nated when either the load suddenly dropped 10 2 !o¥ fraction of the maximum, resulting in a complete &© struction of the specimen, or no instrument was lft ® ACI JOURNAL / July-August 19%WERAGE COLUMN STRAW Race coun sraAN or oe o aa: oF svenace couWMN StAaM seman cou STRAIN Fig. § — Comparison of experimental and analytical stress-strain curves 1) tutvtAuoust 1982 301i i 1 | Fig, 5 (cont.) — Comparison of experimental and an- lytical stress-strain curves 2 eo feemie 4 Me {ogee EXPERIMENTAL AREA ANALYTICAL AREA Fig. 6 — Comparison between areas under the exper- imental stress-strain curves for 24 columns and the cor- responding analytical areas calculated using various models a . position to give further readings. Therefore, the areas under the curves plotted in Fig. 6 do riot necessarily indicate the relative toughness of the columns. Comparing experimental and analytical results shown in Fig. 3, 4, 5, and 6 with the nature of the equations, the following points can be made about var- us models. Chan's equation’ overestimates the strength increase of confined concrete in several cases, particularly for the specimens of Configuration A and those with large tie spacings. As no consideration is given to te spacing and steel configuration, the equation underestimates conerete sirength for columns with tightly Knit cages. Similar comments can be made about the equation pre- dicting the ultimate strain and hence ductility of con- fined concrete. The models proposed by Roy and Sozen" and Kent and Park! do not recognize any strength increase in 302 E REFER 10 Mecicico ken é Park ie wid THERE |S_ A STREUTH oS u confined concrete. Therefore, a complete lack of agceemant is found in Fig. 3 between experimental and ‘analytical. K, values. “A: strain’ of .002 at maximum sess in concrete, as suggested by the two models, underestimates the corresponding experimental strain values in all the columns. The slope of the descending part of the curve is reasonably estimated by Kent and Park's model in most columns, but the descending parts of Roy and Sozen's curves show a slower rate of drop in siress than shown by experimental curves in many cases, particularly in columns with small tie spacings. The energy absorbed by the columns up to ceriain strain values, a5 represented by the area under the curves, is underestimated by both models, as shown in Fig. 6, mainly because of lower predicted concrete ength, The increased concrete strength due to confinement is underestimated for all 24 columns by the model pro- posed by Soliman and Yu." The difference between the experimental and analytical K, values increases with the increase in the aumber of laterally supported longitu- This is because no consideration is given to steel configuration in the model. Except for columns with C configurations and high volumetric ratio of lat- eral sizelt0 concrete core, the model predicts the max imum strain corresponding co the maximum stress rea: sonably well. The descending parts of the curves are too steep: compared with the experimental curves for almost all the columas. This results in the lowest pre- dicted values of energy absorbed by the columns, 2s shown in Fig. 6. Close agreement is found between K, values given bY ‘Sargin’s three proposed equations for most of the cal- umns. Compared to the. experimental values, these equations underestimate the strength of confined com crete for most of the columns. The difference bewwéen experimental and calculated values.is higher for the columns of Configurations B, C, and D than for Type A columns. The effect of tie spacing, it seems, is not appropriately accounted for in Sargin’s equations. Ia- crease in concrete strength is considered co be directlY proportional to the stress in tie steel. The experimental data does not support this assumption, The equation proposed by Vallenas et al." underes- timates the strength of confined concrete in all col tumns. The difference between experimental and ant Iytical sirength factors X, increases with increase in the number of laterally supported longitudinal bars. EX perimental data does not support the assumption that Strength of confined conerete is dependent upon 107" situdinal steel content. This is clearly demonstrated from comparison of experimental and analytical curves in Fig. S(a) and (b). The model predicts the results (OT Column 445-9 (@ = .033) quite well, but the prediction does not remain as accurate for Column 2A5-14, i* which the longitudinal steel content (@) is reduced (© 017. An examination of the experimental and the lytical curves for Columns 2A5-14 and 2A6:15 shows that the effect of tie spacing is more pronounced tha recognized by this model. This observation was 35° ACI JOURNAL J July-August 1987‘ os Cong a Sermeter ney ‘est OM/Anaitieal BM (é, +0005) ao PMA) Fig, § — Comparison beween experimental moment values and moment values calculated from various models for four specimens under different axial loads e o “a 1900 -spean urea u dicted and experimental moment capacities, how: is quite small. Studies of the effects of strain gradient on concrete behavior have shown varied results." Sturman et al.” found in their tests that the peak of the flexural stress-strain curve was located at 2 strain about 50 percent higher and at a stress about 20 per- cent larger than the peak of the curve for concentric | compression. Ford et al." and Kent and Park,' how- ever, indicated that the presence of a transverse strain gradient is not a significant parameter. It is evident that more experimental data is required to investigate the effects of strain gradient on the behavior of con- fined concrete. This should include specimens with var~ ious arrangements of longitudinal and lateral steel un- der various levels of axial load. In the meantime, che monotonic stress-strain relationship far concrete con fined by rectilinear ties, proposed by Sheikh and Uzumeri,” can be used to conservatively determine the ‘envelope moment-curvature curve for members under cyclic bending with reasonable accuracy. a corurite Hivetomn) eel Fig. 9 — Prediction of envelope moment-curvature relationships for specimens under eyclic bending 304 ‘AGI JOURNAL / July-August 198° |/istntinaatnnintn innit in oom 32 wns 8 O03 STRAW ei = | & os! 5 os 2 fom teor | . 7 poet nese PE AN ors cee Boz wee 7 : . o ear 608 008 Sra Fig. 7 — Experimental and analytical core concrete stress-strain curves for spec mens tested by Scott et al." made in several other pairs of columns. Experimental gvidence shows that increase in concrete strength is not, Girectly proportional t0 the stress in lateral steel, as given by the equation, The slope of the descending part of the curve is the same as given by Kent and Park," but due to higher calculated strain values correspond. ing to the maximum stress, the strain-characteristics of most columns are better predicted by this model. With the exception of the model proposed by Sheikh and Uzumeri," this model predicied the resulis of the tests, better than other models studied here (Fig. 5 and 6). ‘The analytical behaviors of confined concrete as cal- culated using ‘the model proposed by: Sheikh. and neril® show’ a close agreement with the experimen- tal results for all 24 columns. In addition to the 24 tests, the proposed model was used to predict confined concrete strength for tests reported by various other researchers." Experimental and analytical X, values for these tests are also compared in Fig. 4. It is ap- parent that most of the previous tests were done in low ranges of K, values. Simple tie arrangements with only four corner bars, low ratios of core area to gross area ~ of the specimens, and small-scale specimens result in small increase in concrete strength. Therefore, the total ceepacity of the specimen, after the cover had spalled off, would not exceed the unconfined specimen's ca- pacity. This seems to be the main reason for the dis- agreement among researchers about the increase in strength of concrete confined by rectilinear ties. The proposed model was also applied to predict complete behavior of two specimens recently reported by Scott et al." The 450 mm square columns were tested under concentric loading at a strain rate of 3.3 x 10 per sec. Fig. 7 shows the comparison of the experimental and predicted curves. Considering the scaiter in the exper~ imental results, the performance of the model is quite ~satisfaciory. Axial. load and eyelic flexure The four models, Kent and Park," Roy and Sozen” Sheikh and Uzumeri,” and Vallenas et al." were used to predict the resulls of tests conducted a1 the Univer- sity of Canterbury, New Zealand.* Four specimens 550 mm (21.6 in.) square and 3300 mm (10 ft 10 in.) long. ‘ACI JOURNAL / July-August 1982 were tested under combined axial and bending loads. Each specimen was subjecied to a cyclic lateral load sequence, while the axial load remained constant throughout the test, Ratios P,/A,f¢ for four specimens were 0.21, 0.26, 0.42, and 0.60, where P, is the axial load and A, is the gross area of cross section of col- umn. Since confinement comes into effect only at large strains, moment capacities of the sections were caleu- [ated mostly at relatively high curvature values. Ex treme fiber compressive strains of .004 of larger were used'for the concrete core. Iterative procedure was used to find the depth of the neutral axis required for the internal forces to balance the external applied load. Cover concrete was assumed to be effective up 10 2 strain of ¢ (the strain corresponding to $0 percent of the maximum stress in cylinder’on. the descending part of the curve) and to be lost at higher strains. The stress-strain curve proposed by Kent and Park! for un- confined concrete was used along with the other models to calculate cover concreie's contribution. The moments were calculated about the plastic centroid of the original section. jg. 8 shows @ comparison of experimental and an- lytical moment values when the extreme fiber com- pressive strain for the concrete core was .005. The pre dictions from the model by Sheikh and Uzumeri" re- main within 4 percent for all four specimens. At lower levels of axial load, all four models give fairly eccurate results. At higher levels of axial load, only the model by Sheikh and Uzumeri remains accurate. Fig. 9 (a) to (d) shows the experimental moment-cur- vature behaviors of the four specimens® under cyclic loading, The envelope moment-curvature curves as de- termined from four analytical models are also shown for each specimen. The section capacity calculated ac- cording to the ACI Code’ is also given. In all four specimens, the model by Sheikh and Uzumeri gives re- sults consistently betier than other available methods. However, in all cases predictions are conservative. 1 ‘appears that the strain gradient in the column section results in additional confinement of concrete which seems to be responsible for slightly increased moment capacity of the section, The difference between the pre~ 303SUMMARY AND. CONCLUSIONS: al models for concrete confinement by rec 7 scat teinforcement proposed by various fsearchers re studied. Seven models were applied £0 specimens by the author and by other investigators. The imens were tested under axial loads only, a5 well 3 under the combined effects of axial and flexural 1] jgads, Analytical evaluation of the envelope curves for dist of specimens tested under fixed axial loads and gseli bending was also inchided in this study. The fol fowing conclusions can be drawn from this work. ‘The model préposed by Sheikh and Uzomer™ pre diets test results better than the other models studied, | eth for axial load only and for combined axial and fesural loads, This is te only model that considers the {fsribusion of longitedinl ste! and the resling tie fonfigurations asa variable affecting the mechanism of SGnfinement. Experimental results confirm the bene- fal effets of better distribution of longitudinal and Intra steel, Volumetric ratio of lateral steel to con- rete core is considered in all seven models as a sig tificant variable affecting concrete Behavior. Other fariables affecting concrete behavior aie tie spacing Jecion dimensions, evlinder strength, and steel fvengih. From the limited work reported here, t ean tise be concluded that the envelope moment-curvature fone for reinforced concrete: section under eyelie bending cen be determined with reasonable accuracy ty using Sheik snd: Uzumerts etresrstrain relation 3 for confined concrete | ACKNOWLEDGMENT ‘The author woulé like 19 thank Dr. Thomas T. C. Hsu, Chai nan, Depertment of Civil Engineering, the Unlveisty of Houston. for his vluable-suggestions. Some ofthe data reduction for this pa per wat done by Mary Robbins, a senior year student 4 NOTATION A, = stots ates of erous tection of column fm airngihvat plain concree as measuted from a standard cinder (6 * 12 ind test | te strengin of coatined concrete foe fy > sttength of concrete in plain specimen F) "= ratio of confined concrete sirengih in a column to frength of plain conerie ina sandats Ke = ratlo of plain coneretesirength In a column to plain conerete strength ina standard specimen 1 ky = = ratio of confined ebserete stengih 10 plain concrete i] sicength in the specimen of similar size and shape = | fell i = slit load on specimen a = concrete strain values” (Fig. 18) sverage longitudinal srain corresponding to the max as = miimom and maxionum average longitudinal stain corresponding to the maximum sires in conerete (Fig. Jey) ACI JOURNAL / July-August 1982 ut fox a= average longitudinal srain-corresnonding 10 88 percent ff the maximum stress in concrete on the unloseins part of the curve (Fi. 13) & ‘slime strain in the column when concrete earies the maximum loss e * ratio of the area of longitudinal steel to the eros-see- sional ated of the colomn fs = ratla of the volume of lateral steel 10 the volume of REFERENCES: 1, Bertero, Vielmo V., and Felippa. C.. Discussion of “Ductilty of Concrete” by H. E. H. Roy and Mele A. Sozen, Flexurel Me chanics of Reinforced Concrete, SP-12, Americen Concrete Inat= tote/American Society of Civil Engineers, Detroit, 1965, pp. 227. 23a 2. Bresler, B., and Gilbert, P. H., "Tie Requirements for Reln- forced Concrewe Columns.” ACI Jounsan, Proceedings V. 58, No. SLND. 1961, pp. 555-565.° - 5. Bonni, N. G., "Rectangular Ties in Reinforced Concrete Col: Reinforced Consrece Columns, SP-S0, Ametiean Concrete Detroit. 1975. pp. 193-210, ie, Edwin G., tnd Hilsdorf, Hubert K., “Behavior of ally Reinforces Concrete Column," Proceedings, ASCE, ¥. 57, ST2, Feb. 1971, pp. $87-€02, 5. Chan, WWW. L.,"'The Ubimate Strength and Deformation of Plastic Hinges in Reinforced Conerste Frameworks,” Magazine of Concrete Research (Landon). V. 7, No.2!, Nov. 1985, ppi 121-132, 6 Hudson, Fred M., “Reinforced Concrete Columns: Effect of Lateral Tie Spacing.on Ultimate Suensth."” Smporium on Rein Jorsed Concrere Columns, SP-13, american Consreie Ini Wait, 1986, pp. 258-248 7. Kear. P. Hi: Fiorate, A. Ex Carpenter. 5.8.3 and Corley W.G., “Limiting Strain “of Concrete Confined by Rectangular Hoops." Research and Developmen! Bulletin No, RDOS3.01, Port land Cement Assosinion, Skokie, 1978, 12 pp. §. Kent, Ducley Charles, and Park, Rober with Confined Concrete.” Proceezinas, ASCE. 1971, pp. 1968-1990, 9, King. J. W. H.. The Btfect of Lateral Reinforcement in Rein- forced Concrete Columns," The Structural Engineer (Londen), V. 24, No.7, July 1986, pp, 355-386 10. King. 1. W. H., "Further Notes on Reinforced Concrete Col- lumins." The Structural Engineer (London). V. 24, No. Il, Now 1946, pp. 609-616 ML. Pfister, Jmes F., “influence of Tis on the Behavior of Rein- forced Conerete Columns," ACI JouRNat, Proceedings V. 61, No, 5, May 1964, pp. 421-536. 12, Roy, H. E. H., and Sozen, Mate A., “Duetliy of Concrete” Flexural Mechanics of Reinforced Conereie, SP-12, Arseican Con: ee Instuie/American Society of Civil Engineers, Deol, 1965, pp. 213224 13, Sargin, M., "SuessSirain Relationships for Concrete and the Analysis of Structural Cencrete Sesvions." Study No. 4, Solid Me- chanics Division, University of Waterloo, 1971, 167 pp 1H, Shah, S.. and Rengan, 8, V., "Ef feets of Reinforcement on Doailiy of Concrete,” Proceedings, ASCE, V. 96, STE, June 197, pp. 1167-1186 1S. Sheikh, 8. A, and Uzumeri, S. M., "Mechanism of Confine- ment in Tied Columns,” Proceedings, 7th World Conference on Earthquake Engineering (Istanbul, Sept. 1980), Kelaynak Printing Company. Ankara, 1980, pp. 71-78 16, Sheith, Shamim A., and Uzumeri, S. M, "Suength and Due- lly of Tied Concrete Columns," Proceedings, ASCE, V. 106, STS, May 1980, np. 1079-1102, 17. Soliman, M. T.-M, and Yu, C. W., "The Flexural Stes 208, ‘Flexutel Members v.97, STD, Julyi 1508, ‘ J 19. Vallenat, en a Strain Relationship of Concrete Confined by Rectangular Transverse + Reinforcement," Sfapazine af Concrete Rescazeh (London), ¥. 19, No. 61, Dee. 1967, 9p. 223238, Ia, Somes, Norman F., “Compression Tests on Haop- Reinforced Concrete," Proceedings. ASCE, V. 95, ST7, July 1970, pp. 1i95- Bertero, Vs Voi and Paper, £. P., “Concrete Confined by Rectangular Haops and Subjected to Avial Loads,” Rex ‘port No. UCB/EERC-17/13, Eacthquske Engineesing Retearch Cone (ee, University of Californin, Berkeley, Aug. 1977. 114 pp v_ 20. Gill, W. Di: Park, Ru and Prietiley, M. JN, “Dusty of Rectangular Reinforced Concrete Columns with Avial Loads,” Re- search Report No, 79-1, Department of Civil Engineering, University of Canterbury, Christchurch, Feb. 1979. 136 pp. 4 21, Seott, 8. Du: Park, Rui and Prlesiley, M.J/N., “StreseStain Behavior of Concrete Confined by Overlapping Hoops at Low and High Stain Rates," ACI Jouanat, Proceedings V.79, No. ty Jane 306 Feb. 1982, pp. 13.27 22. Sinha, B. P.: Gevsle, Kurt H.; and Tulin, Leonard 6., “SiressSurain Relations for Concrete Under Cyclic Loading,” acl Jounsat, Proceedings V, 61, No. 2, Feb. 1964, pp. 198-212 21. Hognestad, Elvind: Hanson, N. W.t and Metlenry, Dousls, “Concreis, Suess Distribution in Ulimate Strength Design,” ACK Joursat, Proceedings V. $2, No. 3, Dee. 1955, pp. 485480 28, ACI Comininee 318, “Building Code Requirements for Re. forced Conereie (ACI 318-77)." American Conerste Insitute, D teolt, 1979, 102 po. 25, Ford, J. $.; Chang,_D. Cx; and Breen, J. E., “Behavior af Concrete Colurnns Under Cantrolled Lateral Deformation,” ACI JouRsat, Proceedings V. 78, No. 1, Jan-Feb, 1981, pp. 3. 26, Surman, Gerald ML Shah, Surendra P.; and Winter. Georg “Effects of Flevral Strain Gradiente on Microcracking snd Stree ‘usin Behavior of Concrete," ACI JOURNAL, Proceedings V. 6, No. , Jely 1965, pp. 805-822 cnwat AAT IOAN 1 tates Maen