You are on page 1of 15
Ae Title no. 79-2 al TECHNICAL PAPER Stress-Strain Behavior of Concrete Confined by Overlapping Hoops at Low and High Strain Rates 6 Kk a Wa by B. D. Scott, R. Park, and M. J. N. Priestley An experimental investigation into the behavior of short reinforced ‘concrete columns is described. Twenty-five concrete units, each 450 ‘mm (17.7 in.) square by 1200 mm (47.2 in.) high, containing either 8 or 12 longitudinal steel bars and different arrangements of square ‘or octagonal stee! hoops, were subjected to concentric or eccentric loads to failure at different strain rates. Results presented include fan assessment of the effect of eccentricity of load, strain rate, amount and distribution of longitudinal steel, and amount and dis tribution of transverse steel. A stress-strain curve for concrete con fined by hoop reinforcement and loaded at a high strain rate (com arable with seismic loading) is proposed and compared with an existing curve based on previous tests conducted at low strain rates. The available ultimate compressive strain for concrete confined by hoop reinforcement is also discussed. Keywords: colams support): compressive strength: confined concrete de- formation; centric loads fire; oops: loads (orcs); reinforced concrete seinforcing steels; strans; stresses; stressstrain relationships; es Flexural strength calculations for reinforced concrete members are generally conducted assuming an extreme fiber concrete compressive strain of 0.003 and a con- crete compressive stress block based on the concrete stress-strain curve up to that strain.' However, under seismic loading high ductilities are often demanded of members, requiring an ultimate concrete compressive strain of much greater than 0.003. This is particularly the case for columns with high axial load levels, where the available section ductility is very much dependent on the shape of the stress-strain relation of the concrete at high compressive strains. It is well known that the ability of concrete to carry significant stress at high strains can be improved by providing arrangements of transverse and longitudinal reinforcement which effec- tively confine the core concrete,” increasing both the compressive strength and ultimate strain. At stresses approaching the uniaxial compressive strength f, the transverse strains become very high because of the pro- gressive internal cracking, and the concrete bears out against the reinforcement.* The confinement of the concrete is provided by arching between adjacent transverse bars and also to some extent by arching be- tween adjacent vertical bars. If the bars are close enough, the confinement of the concrete can approach ACI JOURNAL / January-February 1982 that provided by uniform lateral pressure. The concrete cover (outside the reinforcement) will crack longitudi- nally and separate from the concrete core at compres- sive strains in the order of 0,004 or higher and hence cannot be relied on to carry stress at high strains. Early research on confined concrete, leading to em- pirical stress-strain relations, was generally on small scale concentrically loaded specimens, often without concrete cover. The loading was generally applied con- centrically and at low strain rates. Behavior under these conditions has been used to predict behavior of full-scale members under seismic conditions which are characterized by high strain rates, repeated load ap- plications, and eccentricity of loading. Recent tests conducted by Kaar, Fiorato, Carpenter, and Corley (1978); Vellanas, Bertero, and Popov (1977);* and Sheikh and Uzumeri (1979 have involved more real- istically sized specimens of confined concrete, but the tests have not been conducted under high strain rates representative of seismic conditions. ‘At present, the recommendations made by the seis- ‘mic design codes of various countries show vast dif- ferences in the quantity of transverse reinforcement required for ductility in the potential plastic hinge re- gions of columns and piers. Park and Priestley’ com- pared the different requirements of various U.S., Jap- nese, and New Zealand codes in this respect and it is evident that the required quantity of transverse rein- forcement is still a matter of some controversy. Recent tests conducted at the University of Canterbury*” on nearly full size reinforced concrete columns, containing either spiral or rectangular hoop reinforcement and subjected to slow cyclic loading, have shown that the “Park, Re; Presley, M. J. Nii and Gil, W. D., “Ductlity of Square Confined Reinforced Concrete Colurons." accepied by the Structural Dive sion, American Society of Civil Engine’, ia pes. Received May 4, 1961, and reviewed under Insite publication polices ‘Copyright © 1982; Americas Concrete Intute. All eh reserved, including ‘het making of copes unless permision is obtained from the copyriekt pro Dritors, Pertinent cscusion will be published in the Noveniber December {982 ACI Jocasat if received by Aug, 1, 1962 002-8061/42/010015-18 $250 13 B. D. Scott completed his BE and ME degrees in civil engineering at the University of Canterbury, New Zealand. This paper describes the experimental ‘esearch work conducted for his ME in 1980-81. Hei currently in the United Kingdom gaining farther experience in structural design. 1, Park, FACI, is professor and head of the civil engincering departaent at the University of Canterbury, New Zealand. His research interests include the inelastic behave of reinforced and prestressed concrewe beams, columns, ‘ames, and slabs, and earthquake engineering. He i 8 member of joint ACK [ASCE Commities 382, Joins and Connections in Monoithie Concrete Siu ‘res; 428, Inelastic Behavior of Reinforced Concrete Structures; 44, Reit- forced Concrete Columns; and of the ACI International Activities Commitee 1M. J. N. Prestey isa ceader in evil enpinering atthe University of Canter bury, New Zealand, and was formerly In charge of the Structral Research Laboratory of the New Zealand Minsiy of Works and Development. Hit research interests include earthquake resistance of Bridges, masonry structures and building frames, and thermal effects in concrete bridges and water retain. ing seactures draft New Zealand concrete design code requirements! for special transverse reinforcement for seismic design result in available displacement ductility factors of at least eight in columns. This paper presents the results of an investigation in which a range of nearly full-size specimens were tested under conditions that simulated seismic conditions, Twenty-five concrete units containing either 8 or 12 longitudinal steel bars and different arrangements of square or octagonal steel hoops were subjected to either concentric or eccentric loading to failure at dif- ferent strain rates. Full details of the tests may be seen reported elsewhere.” DETAILS OF TEST UNITS The test units had a section that was 450 mm (17.7 in.) square and 1200 mm (47.2 in.) high. Twenty-five units were tested. Two distributions of longitudinal reinforcement, each representative of current practice, were used. The arrangement of longitudinal bars in the section is of interest because it has been shown that the presence of, well-tied intermediate column bars between the corner bars significantly improves the confinement of the con- crete."* The draft New Zealand concrete design code! requires that in the potential plastic hinge regions of columns in seismic design, the center-to-center spacing of longitudinal bars across the section shall not exceed one-third of the section dimension in that direction or 200 mm (7.9 in.), whichever is larger. The two arrange- ments of longitudinal reinforcement used in the tests are shown in Fig. 1 and 2, The arrangements consisted of either 8 24 mm (0.94 in.) diameter bars giving @ 0.0186 from Grade 380 steel (specified f, = 55 ksi), or 12 20 mm (0.79 in.) diameter bars giving o, = 0.0179 from Grade 275 or 380 steel (specified f, = 40 or 55 ksi). ‘The arrangements of transverse hoop reinforcement used are also shown in Fig. 1 and 2. These arrange- ments are typical for 8-bar and 12-bar columns. The quantities of transverse reinforcement were varied by adjusting the spacing of the hoop sets and are typical of those required by the draft New Zealand concrete 14 Fig. 1 — Typical details of test units 12-bar column Fig. 2 — Typical reinforcement cages design code! for a range of axial load levels. The quan- tities also varied between about 60 and 140 percent of those required by the UBC" provisions. All transverse reinforcement was from plain round bars of Grade 275 steel (specified f, = 40 ksi). The hoop bars were an- chored normally by a 135 deg bend around a longi- tudinal bar plus an extension beyond the bend of at ACI JOURNAL / January-February 1982 Fig. 3 — Typical molds with reinforcement cages least eight hoop bar diameters embedded in the con- crete core. The spacing of transverse hoops was re duced by one-half for the 200 mm (7.9 in.) at each end of the test units to provide extra confinement and in- sure that failure occurred in the 800 mm (31.5 in.) long central region. Table 1 gives the center-to-center spac- ing of the hoop sets and the volume ratios of the trans- verse reinforcement for the test units Note that there is a difference between the detailing practice in the United States and New Zealand. U.S. practice" permits column transverse reinforcement to engage only alternate bars if they are within 152 mm. (6 in.) of a tied bar. New Zealand practice’ requires the center-to-center spacing across the cross section be- tween tied bars to not exceed the larger of one-third of the cross section dimension in that direction or 200 mm (7.9 in.). In the test specimens all longitudinal bars were tied (see Fig. 1). According to New Zealand prac- tice all bars needed to be tied. U.S. practice would re- quire all the bars of the 8-bar column to be tied, but would allow one of the intermediate bars in each face of the 12-bar column not to be tied ‘The test units were cast in the vertical position in batches of 10. The construction sequence consisted of assembling the reinforcing cages, fixing the strain gages, and placing the cages in the painted and oiled plywood molds. The 16 mm (0.63 in.) diameter hori- zontal bars at the third points of the test units (see Fig. 1) were passed through the sides of the mold and bolted to position the cages in the mold accurately. The concrete was placed in three lifts and well vibrated. Fig. 3 illustrates some of the molds. Table I also lists the measured yield strengths of the steel found from tension tests, and Fig. 4 shows the stress-strain curves for the bars. The concrete used was normal weight, had a maxi- mum aggregate size of 20 mm (0.79 in.), used ordinary portland cement, and had a slump of 75 mm (3 in.). The cylinders were cured at 20 C (68 F) in 100 percent relative humidity, while the test units were covered with hessian and polythene with the top surface kept moist. After 7 days of curing, the cylinders and units were stripped and stood in the laboratory until testing. ‘The average compressive strengths given by the 200 ACI JOURNAL / January-February 1982 oe oa aes ae aos 0 Fig. 4 — Stress-strain curves for steel reinforcement mm (7.9 in.) high x 100 mm (3.9 in.) diameter cylinders at age 6 weeks, which was close to the time of testing the units, are shown in Table 1. TESTING PROCEDURE Testing machine and strain rates The tests were conducted in a DARTEC electrohy- draulic universal testing machine which has a vertical load capacity of up to 10 MN (2248 kips) and a max- imum available clearance between the top and bottom steel platens (bearing plates) of 4 m (13.1 ft). The DARTEC machine can be either load or displacement controlled. The machine is capable of recording the load carried during the whole strain range, including the descending portion of the load-strain curve after maximum load. The applied load was measured to a precision of 1 KN (225 Ib) and checked by strains mea- sured on the columns of the testing machine. The tests were conducted at a controlled rate of lon- gitudinal compressive strain of either 0.0000033/sec, 0.00167/sec, or 0.0167/sec, normally up to a maximum strain of about 0.04. The high strain rate is represen- tative of that expected during seismic loading. The low strain rates were for comparison. Instrumentation Longitudinal concrete strains in the test units were measured using linear potentiometers over the central 400 mm (15.7 in.) gage length on each vertical face of the units. The potentiometers were attached (screwed) to the 16 mm (0.63 in.) diameter horizontal bars which were at the one-third points of the units (see Fig. 1). The average of these strains was also checked against the overall longitudinal strain observed from the stroke displacement of the loading jack of the DARTEC ma- chine, and good agreement was found when allowance was made for the flexibility of the machine. Electrical resistance strain gages were attached to the underside of the transverse hoop bars at three different levels within the central 400 mm (15.7 in.) of the test units. These strain gages measured the stresses induced in the hoop reinforcement by the confining pressure. For the tests conducted at the slow strain rate, the 15 Z z ase m /8 3 yoeds 150) 50 ‘sunse1 380) 16 strain gage positions were as shown in Fig. 5. For the tests conducted at the fast strain rate, only the gages numbered 2 and 10 were present at one one-third point, 4 and 11 at midheight, and 7 and 12 at the other one-third point. The strain gages used had a gage Iength of S mm (0.20 in.). For tests conducted at the low strain rate (0.0000033/sec), the loads and strains were recorded manually. For tests conducted at the higher strain rates (0.00167 /see or 0.0167/sec), all measurements were re- corded electronically. Test unit preparation The concentrically loaded test units had thin layers of plaster placed between their ends and the steel plates at the ends of the testing machine. The spherical seat- ing normally located at the top platen was removed since it was considered that a more uniform compres- sion strain would be maintained at high strains if ro- tation of the platen could not occur. For the concentrically loaded test units, a steel roller bearing was inserted at each end of the unit between the end plates and the platens of the testing machine. The roller bearings were set at a predetermined eccen- tricity of 49 mm (1.93 in.) for the 12-bar units and 33 mm (1.30 in.) for the 8-bar units. The eccentric loading produced a strain gradient which varied as the test pro- gressed, since the distribution of concrete compressive stress depended on the shape of the concrete stress- strain curve which changed as the test progressed. For the chosen end eccentricity (the end eccentricity was held constant during the test), the strain was 0 at one face of the concrete core when the compressive strain at the opposite face was about 0.01. Generally, rather than zero strain, there was initially a small compressive strain and then a small and eventually a high tensile strain at one face, while the compressive strain in- creased at the opposite face. TEST OBSERVATIONS The appearance of vertical cracks in the concrete cover was always the first sign of any distress in the test units. These cracks spread rapidly as crushing of the concrete cover caused the cover to become ineffec- tive. As expected, this was particularly evident for the units with closely spaced hoops, since closely spaced transverse steel causes a plane of weakness between the core and cover concrete. However, with the cover con- crete lost, the load still continued to increase as the core concrete became confined by arching between the hoops and longitudinal bars. Eventually this load de- creased. Buckling of the longitudinal bars occurred at higher strains; this was invariably associated with frac- lure of the hoops at or near the buckle. As the hoops snapped, the core concrete in the near vicinity was re- duced to fine rubble and flowed or was ejected from the core. After the tests, it was found that the horizontal bars which passed through the units at the one-third points ACI JOURNAL / January-February 1982 ’ 2 XY} 1? fx S : , EX 44 ; = Fig, 5 — Positions of electrical resistance strain gages for 8 and 12-bar units A ‘ 6 Bucking of longituaino! bar benaing support 24mm (0.94) 1a bor 18mm (0-63in)d0 support bor Fig. 6 — Bending of bar to which linear potentiometers were attached and supported the linear potentiometers had often de- formed near one end, indicating that initially plane sec- tions were not remaining plane for the duration of the test. This occurred despite the fact that the end sections of the test units remained plane. It was noticed that the horizontal bars bent only if they were near a buckle in the longitudinal bars, as illustrated in Fig. 6. It ap- peared that the very high localized strains associated with longitudinal bar buckling caused local distortion of the concrete section. However, because this phe- nomenon occurred toward the end of testing, errors in the longitudinal strains measured were not considered to be great. In the tests, a marked degradation in the strength of the conerete core was always initiated by fracture of an inner hoop. Generally, a considerable number of hoops fractured before testing was complete, and the concrete was still able to carry a significant load after the fracture of three or four hoops. Thus, it would be reasonably conservative to define the limit of useful concrete compressive strain as that strain at which fracture of a hoop first occurs. It should be noted that fracture of the outer (perimeter) hoops occurred later, if at all, than fracture of the inner hoops. This was due to the loss of bond at the outer hoops, caused by the loss of the concrete cover, allowing an averaging of the outer hoop strain across the width of the con- crete core. Fracture of an inner hoop, and the subse- quent local degradation of the strength of the concrete core, also caused a loss of anchorage at the embedded ends of the outer hoop bars, resulting in a tendency for the outer hoop bar to unwind rather than to frac ture. 7 te) ye 0.99 aeoonas (95.4: 1.48 ‘seconds (6) 09 1.88 aeconde (q) tim: 2.22 seconde (8) ena of Beat Fig. 7 — Unit 12 at various instants during the high strain rate test Fig. 7 shows a series of time-sequence photographs for Unit 12, which was loaded concentrically and at the high strain rate, The photographs were taken at high speed, and the time ¢ in the figure indicates the time in seconds after the start of loading. The hori- zontal metal bands seen in the figure near the center of the test unit were made loose fitting, with 25 mm (1 in.) clearance all around, to provide protection to the potentiometers; the bands provided no confine- ment. Fig. 7 shows that vertical cracking of the cover concrete had occurred by t= 0.39 sec, just before a peak load of 8.5 MN (1910 kips) was reached, and by the next photograph at ¢ = 0.79 sec the load had fallen to 6.4 MN (1440 kips). The next photograph at t= 1.13 sec was taken just before the first hoop fracture, and the photograph at t= 1.48 sec is at the third hoop fracture. By the last sequential photograph at ¢ = 2.22 sec, six hoops had fractured; two more would do so before the end of the test. The final photograph shows the condition of the concrete core at the end of testing. Fig. 8 shows selected features of the failures for var- ious test units. Fig. 8(a) shows a typical lightly con. fined unit (Unit 17) and Fig. 8(b) shows a typical heav- ily confined unit (Unit 15), both after high strain rate concentric loading. Failure occurred in the central re- gion of the units and the longitudinal bars buckled. It is also evident that the core concrete of the more lightly confined Unit 17 is significantly more damaged than that of the more heavily confined Unit 15. Fig. 8(¢) 18 shows a close-up of the fracture of three hoops from the lightly confined Unit 12 (see also Fig. 7) and also illustrates the loss of anchorage of the outer hoops. Fig. 8(d) shows the unreinforced Unit 11 after high strain rate concentric loading. The cone of failure at each end joined by a large vertical crack was typical of all unreinforced units tested, and the failure was always brittle with a sudden loss of load capacity. The eccentrically loaded Units 4 and 5 at failure can be seen in Fig. 8(e) and (f). Note the buckling of the longitudinal bars on the compression face and the large evenly spaced tension cracks on the opposite face. These tension cracks are slightly inclined to the hori- zontal due to the shear induced by the amount gradient resulting from high, but variable, P-A moments. Since P-A moments are maximum at mid-height and zero at top and bottom, a moment gradient with height, and thus a shear force distribution, is involved. It is sig- nificant that despite the squat nature of the test units, the lateral displacements finally induced P-A moments of the same order of magnitude as moments resulting from the end eccentricities. The extent of the final lat- eral displacement is clearly illustrated by Fig. 8(0). GENERAL TEST RESULTS ‘The main test results obtained for each unit were the relationships between total load versus longitudinal strain, the longitudinal concrete core stress versus lon- gitudinal strain, and the hoop reinforcement stress ver- ACI JOURNAL / January-February 1982 (a) tase 17 . 8 — Features of failure for various test units sus the longitudinal strain. These results are plotted for six typical test units in Fig. 9 to 14. The plotted results for all the units may be seen elsewhere." Some infor- mation on measured stresses and strains is given in Table 1. Concentrically loaded units Fig. 9 to 12 illustrate the results obtained from typ- ical concentrically loaded units with low and high strain rates. The load-longitudinal strain curves show the sequential fracture of the hoops as steps of de- creasing load. Superimposed on those figures are also the load-longitudinal strain curves for the longitudinal steel and the cover concrete. The load-strain curve for the longitudinal steel in compression was calculated using the stress-strain curves measured in tension at a Iow loading rate on bar samples. The load-strain curve for the concrete cover was calculated from the stress- strain curve measured up to a strain of 0.01 on the unreinforced Unit 1 which was tested at the low strain ACI JOURNAL | January-February 1982 rate Fig. 10 — Unit 6 — Concentric loading at the low strain rate rate. Some error is introduced in using the results of Unit 1 to represent the behavior of the cover concrete for the units with the high strain rate, but the error is small, For example, an error of 0.15 f’ in the stress carried by the concrete cover would cause a 2 percent difference to the stress carried by the concrete core. ‘The load-longitudinal strain curve for the core concrete was attained by subtracting the contributions of the longitudinal steel and the cover concrete from the total load-longitudinal strain curves. Dividing by the core area enabled the core concrete stress-longitudinal strain curve to be obtained; this curve has been plotted in Fig. 9 to 12 up to the stage of first hoop fracture as a fraction of cylinder compressive strength. The core area was measured to the outside of the peripheral hoop, in keeping with the definition in the ACI code.’ Examination of Fig. 9 to 12 indicates that substantial enhancement of peak concrete core compressive strength was obtained as a result of the confinement. For Units 2 and 6, tested at the low strain rate, the strength enhancement was about 20 percent, while for Units 3 and 15, tested at the high strain rate, the strength enhancement was about 70 percent, related to the unconfined cylinder compressive strength. The con- sistent strength enhancement found in these tests as a result of confinement contradicts earlier conclusions by 19 Fig. 11 — Unit 3 — Concentric loading at the strain rate rN Fig. 12 — Unit 15 — Concentric loading at the high strain rate 1 rere iD i I L. Fig. 13 — Unit 4 — Eccentric loading at the low strain rate 20 Fig. 14 — Unit $ — Eccentric loading at the high strain rate Roy and Sozen" that confinement did not increase compressive strength of rectangular sections. Fig. 9, 10, and 11 also show the measured hoop re- inforcement stress-longitudinal strain curves and indi- cate when the hoops have yielded relative to the lon- gitudinal strain. The hoop stress plotted for each unit is an average of all hoop stresses measured for the unit at that longitudinal strain. Eccentrically loaded units Fig. 13 and 14 illustrate results obtained from typical eccentrically loaded units with low and high strain rates. The load-strain curve in each figure plots the load carried versus the longitudinal strains at opposite extreme fibers at the outside of the core concrete and an average of these longitudinal strains. Note that ini- tially both extreme fibers were in compression. How- ever, as the test progressed the increase in curvature of the unit resulted in an increase in the eccentricity of applied load (P-A effect). This caused a change in the strain gradient resulting in tension at one extreme fiber, as may be also observed from the tension cracks in Fig. 8(€) and (f). Also plotted in Fig. 13 and 14 is a comparison of the measured and calculated load and moment at the midheight section of the units plotted against the av- erage longitudinal strain. The measured moment in- cludes the P-A moment which was found from the measured lateral deflection of the units as the test pro- gressed. The calculated values for load and moment, ACI JOURNAL / January-February 1982 = Unit 21 & = 0,0000033 /sec Unit 26 & -0.0167/se === Unit 27 0.001676 CONCRETE STRESS / 0 ‘a00s 00 STRAIN Fig. 15 — Stress-strain curves for unreinforced con- crete units loaded at different strain rates corresponding to a number of measured strain distri- butions, were found from a laminar analysis in which the section was divided into nine 50 mm (2 in.) deep strips. For a given measured strain gradient, the strain in each strip was taken as that at its midpoint. The stress in the concrete core at that strain in the strip was calculated from the core concrete stress-strain curve for the corresponding unit with the same longitudinal and transverse reinforcement which had been loaded con- centrically at the same strain rate. The stress in the cover concrete in the strip was calculated from the stress-strain curve of the corresponding unreinforced concentrically loaded unit. Load and moment at the midheight section were then calculated from the stresses found in each strip. It is evident from Fig. 13 and 14 that the measured loads and moments are much higher than those calculated from the concentric stress- strain curves for the concrete. Hence, the presence of the strain gradient has significantly improved the stress-strain curve for the concrete by reducing the slope of the falling branch of the curve. EFFECT OF RATE OF STRAIN The low rate of strain, 0.0000033/sec, was represen- tative of the strain rate used in many previous column research projects in which the columns were loaded to failure (typically over a period of 10 min to 2 hr). A. medium rate of strain, 0.00167/sec, was included to determine the sensitivity of the strength increase to changes in rate of strain, The high rate of strain, 0.0167/sec, could be regarded as being indicative of the strain rate expected during the response of reinforced concrete to earthquakes. This rate may in fact be ex- ceeded in some structures with a short fundamental period of vibration during earthquakes. However, the effect of increasing the strain rate by a factor of 1 to 4 times is likely to be small in comparison with the ACI JOURNAL / January-February 1982 Curve fer gn sian rte CORE CONCRETE S1RESS Fig. 16 — Stress-strain curves for concrete core of 8- bar units with similar transverse hoops loaded con- centrically at different strain rates aN Dad Pranend Curve hr hh sr ae eget 8) Fig. 17 — Stress-strain curves for concrete core of 12- bar units with similar transverse hoops loaded con- centrically at different strain rates effect of the variation from the low to high strain rates used in these tests, which involved a factor of 0.0167/ 0,0000033 = $000. Fig. 15, 16, and 17 illustrate the effect of strain rate on the stress-strain curves for unreinforced concrete (Units 21, 26, and 27) and for the core concrete of 8- bar units (Units 6, 7, 18) and 12-bar units (2, 3, and 13), respectively. All of these units were loaded con- centrically. The curves in Fig. 16 and 17 were found by subtracting the loads carried by the cover concrete and longitudinal reinforcement from the total load car- ried and dividing by the cote area, as described earlier. a

You might also like