VOL. 114 NO. 3. AUG. 1988 ISSN 0733-9445, CODEN: JSENDH Journal af Structural Engineering AMERICAN SOCIETY OF CIVIL ENGINEERS STRUCTURAL DIVISION |/PuoReTICAL STRESS-STRAIN Mops FOR CONFINED CONCRETE By J. B. Mandes,' M. J. N. Priestley,? and R. Park,’ Fellow, ASCE “Rasiancr: A stress-strain model is developed for concrete subjected to ‘uniaxial compressive loading and confined by transverse roinforcement ‘The concrete section may contain any general (ype of confining steel either spiril or citcular hoops: or rectangular hoops with or without Supplementary cross ties. These cross ties can have either equal or tunequal coafining stresses alovg each of the transverse axes. “\ single ‘equation is used for the stress-strain equation. The model allows for toc loading and inclides the effect of strain rate. The influence of ‘Various types of confinement is taken into account by defiiing. a1 effective iteral confining siress, which is dependent on the configura tion ofthe transverse and longitudinal reinforemen'. An energy balance approach is used to predict the longitudinal compressive strain in the Conerete corresponding to first fracture ofthe tran-verse reinforcement ‘by equating the strain energy capacity of the transverse reinforcement to ‘the strain energy stored in the conctvte asa resul: of the confinement, IntRoDucTION In the seismic design of reinforced concrete columns of building and bridge. substructures, the poteatial plastic hinge regions need (0 be carefully detailed for ductility in oder to ensure thatthe shaking from large carthquakes will not cause collapse, Adequate ductility of members of reinforced concrete frames is also necessary to ensire that moment Tecistribution can occur. The most important design consideration for ductility in plstic hinge regions of reinforced concrete columns is the provision of sufficient transverse reinforcement in the form of spirals or circular hoops or of rectangular arrangements of sleel, in order to confine the compressed concrete, to prevent buckling ofthe longitudinal bars, and to prevent shear failure, Anchorage feilure ofall reinforcement must also be prevented. ‘Tests have shown that the confinement of concrete by suitable arvange- ‘ments of transverse reinforcement results in a significant increase in both the strength aad the ductility of compressed concrete. In particular, the strength enhancement from confinement and the slope of the descending branch ofthe concrete stress-strain curve have a considerable influence on the flexural strength and ductility of reinforced eoncrete columns. ‘Theoretical moment-curvature analysis for reinforced concrete columns, indicating the available flexural strength and ductility, can be “aiViitng Asst. Prof. of Civ. Engrs, State Univ. of New York at Butalo, Bufo, NY 14260, “Prof. of Struct. Enyrg., Univ. of California, San Diego, CA 92037 ‘Prof, and Head of Civ, Enarg., Univ. of Canterbury, Christchurch, New Zeal ole. Discussion onen until January 1, 1989. Separate dscussioas should be submitted forthe individual papers inthis sympesium. To extend tho closing date fone month, a written request ust be led vith the ASCH: Manger of Joureals. The ‘manuscrint for this paper was submited for review and possible publication on Desembee 30,1986 Ths paper is part ofthe Journal of Seuenwal Engnceing, Vol 114, No. 8, August, 1988, ©ASCE, ISSN O735:94450008- 1604510 + $15 per page. Paper No. 22686 1904 conducted providing the stress-strain relation for the con” and steel are known. The moments and curvatures associated with increasing few deformations of the column may be computed for various column axial Joads by incrementing the curvature and satisfying the requirements of strain compatibility and equilibrium of forees. Tho cover concrete will be ‘unconfined and will eventually become ineifective after the vompicesive strength is attained, but the core concrete will continue to carry stress at high strains. The compressive stress distributions for the core and eover concrete will be as given by the confined and unconfined concrete stress-strain relations. Good confinement ofthe core concrete is essentia| if the column is to have a reasonable plastic rotational capacity to maintain flexural strength as high curvatures. In general, the higher the axial compressive load on the column, the grester the amount of confining reinforcement necessary to achieve ductile performance. This is because & high axial load means a large neutral axis depth, vhich in turn means that the flexural eapacity of the column is more dependent om the contsibution of the conerete compressive stress distribution. Clearly it is important to have accurate information concerning the complete stress-strain curve of confined conereie in order to conduct reliable moment-curvature analysis to assess the ductility available from columns with various arrangements of transverse reinforcement. Jn this paper, a unified stress-strain model for confined concrete is developed for memters with either circular or rectangular sections, under static or dynamic loading, either monotonically or cyelially applied. ‘The concrete section may coniain any gencral type of confinement with cither spirals or eirevlar hoops, or rectanguler hoops with or without supplsme tary cross ties, with either equal or unequal confining siresses along each ‘of he transverse axes. The model includes the effects of cyclic loading and Strain rate. Full details of the proposed model is discussed elsewhere (Mander et al. 1984) Jn a companion paper by Mander ct al. (1988), the theoretical model Presented herein is compared with the results of an experimental program ‘of some 40 concentric axial compression tests. This program consisted of nearly fullsize circular, square, and rectangular reinforced concrete Columns teste at either slow or fast (dynamic) rates of strain, with oF without cyelic loading. Past INVESTIGATIONS OF THE BEHAVIOR AND MODELING oF CONFINED Concreve. Barly investigators showed that the strength and the corresponding longitudinal strain at the strength of concrete confined by an active hhystrostatic fluid pressure can De represented by the following simple relationships: Fee fio hifi eee oO Si) ain #) =) Where f, and 2.. = the miximum conerete stress and the corresponding | Strain, fespectively, under the ‘atera! fd. presture fj: fer and. ta 808lune _1ed conerote strength and corresponding strain, respectively; and. ( k; and ky = coefficients that are functions of the concrete mix and the lateral pressure. Richart et al. (1928) found the average values of the coeificients for the tests they conducted to be A, = 4.1 and k; = 5k, Also, Balmer (1949) found from his tests that ky varied between 4.5 and 7.0 with an average value of 5.6, the higher values occurring at the lower lateral pressures. Richart et al, (1929) also found that the strength of concrete with active confinement from lateral (fluid) pressure was spproximately the same as for concrete with passive confinement pressure from closely spaced circular steel spirals cousing an equivalent lateral pressure. Different investigators, such as Mander et al. (1984), Scott et al. (1982), Sheikh and Uzumeri (/980), and Vellenas et al. (1977), have carried out ‘numerous tests on nearly full-size specimens end have demonstrated that confinement is improved if (1) The transverse reinforcement is placed at relatively close spacing; (2) additional supplementary overlapping hoops or ‘cross ties with several legs crossing the section are included; G) the Tongitudinsl bars are well distributed around the perimeter: (4) the volume of transverse reinforcement (0 the volume of the concrete core or the yield strength of the transverse reinforcement, is increased; and (5) spirals or circular heops are used instead of rectangular hoops and supplementary ‘eross ties, Clearly it is important to be able (o quantify these effects of ‘confinement on the stress-strain behavior of concrete. The complex endochronic mathematical model developed by Bazant and ‘Bhat (1976, 197) appears to be the only constitutive model that describes the stress strain response under monotonic, ¢ © loadings ‘of confined or unconfined conerete with any state of multiaxial stress. However, endochronic constitutive models were developed using data ‘based primarily on biaxial and triaxial tests with active confinement provided by mechanical means. Therefore, at this state of development, no rational allowance can be made for the passive confinement from the many different configurations of transverse reinforcement that are possible using various hoop shapes and spacings. Early research on confined reinforced concrete behavior was generally carried out on small-scale concentrically loaded specimens at quasi-static rates of strain, The stress-strain model of Kent and Park (1971) for concrete confined by rectangular transverse reinforcement was based on the lest results of Roy and Sozen (1964) and others available at that time. This early model neglected the increase in concrete strength but took into account the increase in ductility due to rectangular confining steel. More recently, Scott ct al. (1982) and Park et al. (1982) have tesied near full-size specimens based on real building columas and modified the Kent and Park (4971) stress-strain equations to take into account the enhancement of both, the concrete strength and ductility due to confinement and the effect of strain rate. Monotonic stress-strain equations for concrete confined by fectangular-shaped transverse reinforcement include those proposed by Vellenas et al. (1977) anst Sheikh and Uvumeri (1980). Siress-strain equations for concrete confined by spiral reinforcement have been pro- posed by Park and Leslie (1977), Desayi et al. (1978), Ahmad and Shah (1982, 1985), Dilger et al. (1984), and others. ‘The flexural strength and ductility of confined reinforced concrete 1806 sections computed using thse stress-strain eauations shoy Particular, the equations are groused into those applicable © grouped into those applicable. rectangatay shaped confining steel and those applicable co excular shaped confog steel tis evident tata united approach applicable toll conigrations ef circular and rectangolar-sbaped transverse reinforcement, ait incien the effects of cyclic loading and strain rate, is required. oe ferences. In Unirieo Stess-Srrain Apenoaci Fon ConriNeo Cor Monotonic Loapina at Siow SrRAIN RATES gol! ‘The Bai Equation fo Monotonic C onl Compreson Long Mander etal (1980) have proposed suniedstesestrain approach for conied concrete appleabie tooth cia and ecangue shaped ranseicrinoreement. The steostan tod sls in ae ae ioe an. eaeiid ‘suggested by Popovics (1973). For a slow siti) stun le and monotone oar, thefongiaal com sive conerete stress f, is given by : oe @ “) where ¢, = longitudinal compressive concrete strain, 6 Compressive Stress, f- Compressive Stroin, &¢ FIG, 1. Streae-Strain Model Proposec: for Monotonic Loadin a, Sage Stn roposes for Monotonic Loading of Confined and 1807ay’ gested by Richart et al. (1928), where f/,, and e,. = the vnconfined/ o...cte strengit and corresponding strain, respectively (generally «.,, 0.002 can be assumed), and E Pp 6 where Fe =S,00\/Fig MPA. ccc : oD is the tangent modulus of elasticity of the concrete (1 MPa = 145 psi), and Fug oe 6 To define the stress-strain behavior of the cover concrete (outside the confined core concrete) the part of the felling branch in the region where e, > 2ceg is assumed to be @ straight line which reaches zero stress ai the spalling strain, ¢,. Effective Lateral Confining Pressure and the Confinement Effectiveness Coefficient ‘An approach similar to the one used by Sheikh and Uzumeri (1980) is adopted to determine the effective Iuteral confining pressure on the concrete section. The maximum transverse pressure from the confining, steel can only be exerted effectively on that part of the concrete core where the confining stress has fully developed due to arching action. Figs. ? and 3 show the arching action that is «ssumed to occur between the levels of transverse cireular and rectangular hoop reinforcement, Midway between the levels of the transverse reinforcement, the area of inaflectively confined concrete will be Jargest and the area of effectively confined concrete core A. will be smallest. When using the stress-strain relation, Eq, 3, for computing the strength and ductility of columns itis assumed for convenience that the urea of the Confined concrete is the area of the concrete within the center lines of the Perimeter spiral or hoop, A... In order to allow for the fact that A, < 4,., itis considered that the effective jateral confining pressure is Si fike 0 where f; = lateral pressure from the transverse reinforcement, assumed to be uniformly distributed over the surface of the concrete core; A, kage eee oe sae yorinc (OO) = confinement effectiveness coefficient; A, = area of effectively confined concrete core; Acc Ad ped) « an Pc ™ ratio of area of longitudinal reinforcement to area of core of section; and Az'= area of core of section enclosed by the center lines of the Perimeter spiral or hoop. 1808 Effectively confined core Cover concrete (spalis off) Ineffectively confined core — Ta~ SECTION A-A FIG. 2 Etlectively Confined Core for Circular Hoop Reinforcement Confinement Effectiveness for Sections Confined by Spirals or Gireular Hoops If in Fig, 2 the arching action is assumed to occur in the form of a second-degree parabola with an initial tangent slope of 45*, the area of an cffectively confined concrete core at midway belween the levels of {transverse reinforcement is, tHe) nial oo esc where s’ = clear vertical spacing between spiral or tioop bars; and d, diameter of spiral between bar centers. Also the ares of concrete core is Aa =F ~ pod «) Therefore, from Eq. 10), the confinement effectiveness coefficient is for cirular hoes 1809Effectively | confined core —. SECTION Y-¥ FIG. 3. Effectively Contined Core for Rectangular Hocp Reinforcement Paar ie a4), os) ‘The lateral confining pressure may be found by considering the half body sonfined by a spiral or circular hoop. If the uniform hoop tension Aleveloped by the transverse siecl at yield exerts a uniform lateral stress on whe concrete core, then equilibrium of forces requires that hn Ass = fisd, wees beeereee on 16) where fn = yield strength of the transverse reinforcement; A,» ~ area of transverse reinforcement bar; f; = ltera) confining pressure On concrete and s = center to center spécing or pitch of spiral or circular hoop, Now if p, = ratio of the volume of transverse confining steel to the volume of eonfined concrete core, *hen «n 1810 Substituting Eq. 17 into Eq. 16 and rearrauging gives fi 1 2 Pha = 8) Therefore from Eq. 9, the effective la:eral confining strss on the concrete 1 lass Rll ie r easy where k, is given by Eqs. 14 or 15. Confine:nent bffectiveness for Rectangular Concrete Sections ‘Confined by Rectangular Hoops with or without Cross Ties In Fig. 3, the arching action is again assumed to act in the form second-degree, parabolas with an initial tangent slope of 45° Arching, ‘oveurs vertically between layers of transverse hoop bats and horizontally between longitudinal bars. The effectively confined area of concrete at hoop level is found by subtracting the area of the parabolas containing the ineffectively confined concrete. For ene purabola, the ineffectual area ix (0) 716, where wis the ith clear distance between adjacent longitudina} bars (see Fig. 3). Thus the total plan area of ineffectually confined core Concrete at the level of the hoops when there are 1 longitudinal bars is - 20) Incorporating the influence of the ineffective areas in the elevation (Eig. 3), the area of effectively confined concrete core at midway between the levels Of transverse hoop reinforcement is to(om $2) -a)led) where b, and d. = core dimensions to centerlines of perimeter hoop in x and y directions, respectively, where 6, = d_.. Also, the area of concrete ore enclosed by the perimeter hoops ia given by &q. 1. Hence from Eq, 10 the confinement effectiveness coefficient is for rectangular hoops It is possible for rectangular reinforced concrete members to have different quantities of transverse confining steel in the x and y directions. ‘These may be expressed as + @D + @2) Pe pee eer ere F 3) 1811Ay { where Ay and Ay, ~ the total area of transverse bars running in the x and » directions, respectively (see Fig. 3). The lateral confining stress on the concrete (total transverse bar force Aivided by vertical area of confined conerets) is given in the s direction as, Aw Sin= FF Son = cB - @5) and in the y direction as Ay Typ = OE hacer corsc crane Bs 26) From Eq. 9 the effective lateral confining stresses in the x and y directions are Sic ken and. fh, = keoshn ‘where kis given in Eq, 22. Compressive Strength of Confined Conerete, f. To determine the confined concrete compressive strength f'-, & consti tutive model involving a specified ultimate strength surface for multiaxial compressive stresses is used in this model. The ‘‘ive-parameter”” multi- axial failure surface describec by William and Warnke (1975) is adopted, since it provides excellent agreement with triaxial test data, The calculated ultimate strength surface based on the triaxial tests of Schickert and ‘Winkler (1977) is adopted here. Details of the calculations have been given by Elwi and Murray (1979). The general solution of the multiaxia! failure criterion in ternis of the two lateral confining stresses is presented in Fig. 4. When the confined concrete core is placed in triaxisl compression with equal effective lateral confining stresses ffrom spirals or circular hoops, it can be shown that the confined compressive strength given is Ba, ft). @n + 28) 7 29) fan fil- 1.254 42.2544/14 ie ‘unconfined concrete compressive strength; and fis given by where f., E419. AAs a numerical example, consider a column with an unconfined strength Of f¢o = 30 MPa (4,350 psi) and confining stresses given by Eqs. 28 and 29 of fy = 2.7 MPa (390 psi and f;, = 5.1 MPa (740 psd. Then, by following the dotted in in Fg. 4, the conpressve strength atthe confined covers is found to be fi = 1.65 x 30= 49.5 MPa (170 ps, Monotonic Tensile Loading A linear stress-stiain relation is assumed in tension up to the tensile strength, provided the tensile strength has not been exceeded. ‘The longitudinal stress f. is given ky 1812 38 Confined Strength Rl en/t, Ss 10. 15 20 Biaziar Largest Confining Stress Ratio 5 g 0 ar Smallest Confining Stress Ratio, f/f, 0.2 03 FIG. 4, Confined Strength Determination om Lateral Confining Stresses for Rectangular Sections So= Ete when f. ( ~ fg) ctacks open and the tensile sirength of concrets for sll subsequeat Joadings is assumed to be zero. Reloading Branches Fig. 7 shows the stress-strain curves including unlosding and reloading branches. The coordinates of the point of reloading (e,f,.) may be fron either the uniosding curve, o: from the cracked state in which ty. = (y= e))and f,, = 0, as shown in Fig, 7. A linear stress-strain relation i assusied between e,. and ey t0 a revised stress magnitude to account for cycle degradation, The new stress point (neq) is assumed to be given by the equation Son = OD fag + 0.08 ry ooo ve ees ; -. 40) ‘The same experimental data used to calibrate Ea, 46 was used for Eqs "A parsbolic transition curve is used between the linear relation Sof + Elce~ en) ae . 4) e16 l€c FIG. 7, Strese-Strain Curves for Reloading Brench where a. bite (48) And the monotonic stress-strain curve (Fa. 3) return coordinate (ey 5 fy) ‘The common return strain (¢,,) is assumed to be given by the following equation (49) where Es given by Ea, 48 The parabolic transition curve is then described by fat Byes $A? (50) where 7 Ere). sea : eee Z (51) cr) = MFrow ~ Soo) = Elna = rd Eye and fre = the common return point tangent modulus and the stress determined from the return strain, ¢,.. usiny, the monotonic siress-strain relation (Eq. 3), respectively. Errect oF RATE oF Stra on STRESS-STRAIN RELATION Concrete exhibits a significant increas: in both the strength and stifiness when loaded at an increased strain rate. Experimental data on the 181718) 16| he 12 19 le 10® 70° 104 10% 10° 10 1 10 Average Rate of Strain per second I&c! £1G. 8. Dynamic Magnifiostion Factors D, to Allow for Strain Rate Effects on Sirongts properties of concrete subjected to high strain rates has been reported by Watstein (1953), Bresler and Bertero (1975), Scott etal. (1982), Ahmad and Shah (1985), Dilger et al. (1984), and others. ‘The stress-strain relations given by Eqs. 3-52 have been written for slow (quasi-static) strain rates. However these equations will also apply 10 ‘concrete loaded at high strain rates providing that the control parumeters Feo, Eo. and E,o of the unconfined concrete are modified s0 as to apply 10 the relevant strain rate e, Relationships for the strain-rate dependence of these parameters, estab- lished by Mander et al. (1984) from the experimental results, are ss follows, Dynamic Strength (Fedlin = Difco ee ors) where f2, = the quasi-static compressive strength of concrete and (54) = rate of strain in s~! ; and /?, is in MPa (1 MPa = 145 psi). The dynamic magnification factor D, was found by regression analysis of the experimental results of Watsicin (1953) on plain concrete specimens of different strengths. Fig. 8 shows a plot of Eq. 54 compared with those ‘experimental results for two concrete strengths. Good agreement was 1818 © fe17.2MPa © fe 2h6.8MPo — Theory tr oF 10° 10% 10° 10? 107 1; 10 Average Rate of Strain per second \&c| FIG. 9. Dynamic Magnification Factors D, to Allow for Strsin Rate Effects on Stifiness provided by the limited data on large concrete specimens tested by the writers, Dynamic Stiffness Ela DER ee eee : 5) where E, = the quasi-static modulus of elasticity; and (56) where e, = rate of strain in s~' and f!, = the quasi-static compressive strength of concrete in MPa (1 MPa = 143 psi). The dynamic magnification factor Dp was found by regression analysis of the experimental results of ‘Watstein (1953). Fig. 9 shows a plot of Eq. 56 compared with those results for two concrete strengths. Dynamic Strain at Peak Stress (cddgn= Dates « wn Whore ey = quasi-static strain at peak stress; and z oe (58) The results of experiments by vatious investigators appear to show no consensts onthe value ofthe strain at peak steos for ign rates of strain . 98 was derived assuming thatthe werk done on concrete to achiove strength is constant respective ofthe rate of strain. Generally good agreement i obtained with most observed result re19Ge fee) Low strain rate (quasi-static) Eo Strain ee FIG. 10, Effect of Strain Rate on Monotonic Stress-Sirain Relation for Concrete Fig. 10 shows the typical results obtained from the application of Eq 35-58 to predict the stress-stmin curve of concrete at high and lov: strain tes. Itwill be seen that an increase in the strain rate results in an increase in the strength f;. and the init! stiffness E,. and a decrease in the strain at peak stress c.... ‘There is also an increase in the steepness of the falling branch curve, such that the dynamic curve approaches the quasi-static curve at high strains. As an example of the inflience of Eqs. 53-8, consider concrete of luiconfined strength f;, = 30 MPa (4350 psi) tested at a strain rate of 1sélsce. Eqs, 53 and 34'predict a strength inerease of 27%, Eqs. $8 and So predict an initial stifness increase also of 27%, and Eqs. 57 ad 58 predict a reduction in strain at peak stress of 13%. Uctimare Concrete Compression StRAIN In orter to calculate the available ultimate rotation capacity at a plastic hinge in a reinforced concrete flexural member, itis necessary to be able to predict the ultimate concrete compressive strain e,,,. Early experimen. tal work on the deformability of compressed concrete in reinforced concrete members by a number of investigators resulted in the develop- ment of several empirical equations for e,,, A summary of some of those carly equations for s,, is given in Park and Paulay (1975). Recently Scott et al. (1982) have proposed that the ultimate concrete compressive strain be defined as the longitudinal strain at which the first hhoop fracture occurs, since that strain can be regarded as ths end of the useful region of the stress-strain curve for the confined concrete core Alter first hoop fracture there is a sudden drop in the compression load capacity of the core conerte duc to reduction in confinement, and there also a loss of buckling restraint for the compressed longitudinal bars, Subsequentiy, Mander et al. (1984) proposed a rational method for predicting the longitudinal concrete compressive strain at first. hoop fracture based on an energy balance approach. In this approach, the additional ductility available when concrete members are confined is considered to be due to the energy stored in the transverse reinforcement. 1820 a ss — Consider the stress-strain curves for unconfined and dyed concrete shown in Fig. 1. The area under each curve represents the total strain energy per unit volume required to “ail” the concrete, The increase in strain energy ai failure resulting from confinement (showa shaded in Fig. 1) can only be provided by the strain energy capacity of the confining reinforcement as it yields in tension. By equating the ultiraate strain energy capacity of the confining reinforcement per unit volume of concrete core (U,4) to the difference in area between the confined (U,.) and the unconfined (U.,) concrete stress-strain curves, plus additional energy required to maintain yield in the longitudinal steel in compression (U.,), the longitudinal concrete compressive strain corresponding to hoop frac. ture can be calculated. Thus Vay = Vee + Usp — Uso (59) ‘Substituting in Eg. 59 gives Pe [ties he [tae he ff fide dees [see (60) where p, = ratio of volume of transverse reinforcement to volume of Concrete core; A... = area of concrete core, f, and e, = siress and strain in transverse reinforcement; ey = fracture strain off transverse reinforce- ment; f_ and e, = longitudinal compressive stress and strain in concrete, oq = ultimate longitudinal concrete compressive strain: p,. = ratio of volume of longitudinal reinforcement t volume of concrete core, fay = stress in longitudinal reinforcement; and ¢,,~ spalling strain of unconfited. concrete. Ih the first term on the left-hand side of Eq. 60, the expression 1) is the total area under the stress-strain curve for the transverse reinforce- ‘ment up to the fracture strain ey. Results from tests caricd out by Mander et al, (1984) in New Zealand on yrade 275 (f, = 40 ksi) and srade 340 (f, = 55 ksi) reinforcement of various bar dismeters indicates that Uy i effectively independent of bar size or yield strength, and may be taken (within = 102%) as Uy = 110 Main? E @) For this steel ey ranged between 0.24 and 0.29. For the last (erm on the right-hand sie of Eq. 60, the ares under the stress-strain curve for unconfined conerete is required. Ic was found from analyses of measured data from a range of plain concrete specimens that the area under the stress-strain curve for unconfined concrete may be approximated as [ fede, = 0.01T\/Fo Mim? . 0 - 63) 1821