Tie Requirements for Prestressed Concrete Columns Journal/1987/July-Au… · ity. Ties and square spirals do provide some confinement for concrete at corner and central core locations;

^)a_research_fellowshi

Tie Requirements forPrestressed Concrete

ColumnsGrant T. Halvorsen

Associate ProfessorDepartment of Civil Engineering

West Virginia UniversityMorgantown, West Virginia

Craig A. CarinciStaff EngineerDewberry and DavisFairfax, Virginia(Formerly PCI Research Fellow atWest Virginia University)

P revious research on the behavior ofprestressed concrete columns has

been directed toward obtaining rationalmethods for predicting the ultimatecolumn strength based on slenderness,eccentricity, level of prestress, andstrength of steel and concrete. A numberof further concerns remain. The precastprestressed concrete industry has indi-cated a need for better assessment of theeffectiveness of lateral reinforcement.'Research in the PCI Specially FundedResearch and Development (PCI-SFRAD) program, as well as this study,will serve to answer some of these con-

without reinforcement, design en-gineers are often reluctant to omit tiesfrom prestressed columns. One author ofa current textbook on prestressed con-crete states that "lateral reinforcementshould always be provided just as forreinforced concrete columns." 2 Pres-ently, there is not a rational basis forevaluating the contributions of lateralreinforcement,* and the current provi-sion of tie reinforcement is likely to beuneconomical.

For prestressed concrete, ACI 318-833

cerns. *R. G. Anderson, Concrete Technology As-Although wall panels perform well sociates, Private Communication, 1979.

46

states in Section 18.11.2.2, "Except forwalls, members with average prestressequal to or greater than 225 psi (1.55MPa) shall have all prestressing tendonsenclosed by spirals or lateral ties." Inthis Code Section, "lateral ties shall beat least #3 (10 mm) in size or weldedwire fabric of equivalent area, andspaced vertically not to exceed 48 tie baror wire diameters or least dimension ofthe compression member." The tiespacing limit of 16 longitudinal bar di-ameters, given in Section 7.10.5.2 ofACI 318-83 is omitted for prestressedcolumns.

However, there is a clause in ACI318-83 Section 7.10.3 (Lateral rein-forcement for compression members)which states that "lateral reinforcementrequirements of Sections 7.10, 10.14 and18.11 may be waived where tests andstructural analysis show adequatestrength and feasibility of construction."In the 1976 report of the PCI ColumnCommittee, 4 some guidance is providedfor design engineers who wish to satisfythe escape clause of ACI 318-83, Section7.10.3. Chapter 12 of that report con-cerns the design of columns. Sections12.1.4 and 12.1.5 of the report permit thedesign of columns with or without lat-eral reinforcement. Lateral reinforce-ment is required only where nonpre-stressing reinforcement exceeds 0.01times the gross column area, or wheresuch reinforcement is used to carrycompression loading. The provisionsalso state "when lateral reinforcement isnot provided, computed design capacityshould be multiplied by 0.85."

In order to justify the omission of tiesthe designer must understand theircontribution to the behavior of thecolumn. Lateral reinforcement in theform of ties and spirals may perform fourprimary functions in conventionallyreinforced concrete columns:

— Confine concrete— Increase transverse shear strength— Prevent premature buckling of lon-

gitudinal bars

opsis

This study was initiated to evaluatethe effectiveness of lateral ties in pre-stressed concrete columns. Thirty-six8 x 8 in. (200 x 200 mm) columns werefabricated and tested with a range ofparameters for longitudinal and lateralreinforcement, column length, and ec-centricity of load.

For these columns, lateral tiesprovided in accordance with theminimum requirements of ACI 318-83made no difference in failure load, orthe overall violence of the failure.Some details of member failure areinfluenced by the presence ofminimum ties.

The results of this test series alsoconfirm the applicability of modifiedmoment magnifier methods in pre-dicting column load capacity whenload capacity is at least 20 percent ofthe axial capacity with no eccentricity,Po.

— Hold bars apart during castingThe use of square hoops or ties is not

the most effective method for confiningconcrete and increasing member ductil-ity. Ties and square spirals do providesome confinement for concrete at cornerand central core locations; however,members containing such reinforcementare generally believed to fail imm-ediately after the peak load has beenreached. Circular spirals can produce amuch more ductile and toughermember, since a "continuous confiningpressure" is applied to the core of thecolumn by the spiral's tension force.

Lateral ties in prestressed concretecolumns may increase the horizontalshear strength but may or may not effec-tively confine concrete. The ability oflateral reinforcement to effectively con-fine concrete may depend on columnslenderness and eccentricity of the load

PCI JOURNAUJuly-August 1987 47

as well as the lateral reinforcement sizeand spacing.

The clear advantage of prestressing aconcrete column is where the failureload is governed more by bending thanby axial force. This will be true if thecolumn is slender or is subject to largebending moments caused by eccentric-ity of load or transverse loads.

Most prestressed concrete columnsare precast and many serve as membersin precast structures where beams maybe framed into columns by means ofbrackets or corbels on the outside of thecolumn. Such details can produce ec-centric load on the column, possiblyproducing large column moments. Sinceprecast structures are often only severalstories tall, high dead load axial forcesfrom stories above may not be encoun-tered. A prestressed column is in a stateof significant axial compression due toforces induced by the pretensionedstrands. Since pretensioned strands tendto stiffen the member against bending,either from lateral or eccentric loads,these members are ideally suited forprecast building systems.

As bending increases and the lon-gitudinal steel on the tension face of thecolumn reaches higher stress levels, theneutral axis shifts toward the compres-sive face of the column. Depending onthe length, and ultimately the deflec-tion, the position of the neutral axis maybe located so that outward transversestresses are developed outside of the"effectively confined core" of thecolumn. Consequently, lateral rein-forcement provided for confinementmay not be necessary or effective in themidregion of the column. However, lat-eral ties may be necessary in columnend regions, especially in shortercolumns where transverse stresses maybe large.,'

It has been believed, generally, thatsince prestressing strands exist in a stateof high axial tension, there is no possi-bility of premature strand buckling asthe column approaches failure. In addi-

tion, as strands are tensioned they arepositioned by a bulkhead in their exactlocation in the column section. Thesereasons, combined with the ability oflong prestressed bearing walls "to per-form well without ties," as well aseconomic pressures, have motivated theindustry to evaluate the effectiveness ofties in prestressed concrete columns.

Placing lateral ties is a labor intensiveand time consuming activity which de-creases productivity. Columns could becast faster and more economically if tierequirements were more liberal. On theother hand, if lateral reinforcement isfound to be beneficial, fabricating tech-niques might be adjusted, focusing onthe square spiral or welded wire fabric.This latter possibility is perhaps attrac-tive to producers and fabric manufactur-ers, since fabric bending jigs are fre-quently present in the precasting plant.

In any event, a more refined assess-ment of tie behavior is crucial to im-proved designs that are more competi-tive with alternate materials. This ex-perimental study will attempt to answersome questions about the effects of lat-eral ties, as used in prestressed concretecolumns, and to direct designers andproducers to a safe and economical so-lution.

The immediate objective of thisresearch is to observe and evaluate theeffect of lateral tie reinforcement on thestrength and behavior of prestressedconcrete columns. Parameters studiedinclude:

— Slenderness— Eccentricity of load— Level of prestress— Number of longitudinal strands— Tie spacingDuring the experimental stage of the

investigation, loads, column deflections,and member behavior at ultimate loadwere recorded for 36 columns. Analysisof the column strength data also providean important opportunity for evaluatingproposed moment magnifier methodsfor predicting load capacity. Conclu-

48

lions, discussion of design implications,and recommendations for further studyconclude this report.

LITERATURE REVIEWStudies of prestressed concrete

columns have been conducted fromboth analytical and experimental view-points. Analytical research efforts havebeen directed at predicting the ultimatestrength of the members. Most methodsaccount for the effects of slenderness,eccentricity of load, amount and level ofprestress and concrete strength.Methods of achieving a realistic predic-tive equation have been based on theusual approaches from the theory ofelasticity and strength of materials.

Several authors have approximatedthe column deflected shape with a sineor cosine wave, then used this deflec-tion shape to determine column curva-tures, or plot load-deflection relation-ships. Investigators have frequently de-veloped computer programs to predictload-moment curves based on the inter-action diagram for members of anylength.

Researchers have addressed the ef-fectiveness of lateral ties in test speci-mens, though these tests are extremelylimited. The following discussion willbriefly review the literature pertinent toprestressed columns tested without ties,approximate method for evaluatingslenderness in prestressed columns, andsome basic questions of effectiveness oftie reinforcement. Ref. 6 provides amore detailed literature review.

Tests of Prestressed ConcreteColumns

Jernigan 7 tested prestressed and con-ventionally reinforced concrete columnsto: (1) compare ultimate strengths of thetwo column types for similar size andproperties; (2) determine the pointwhere additional prestressing resulted

in an ultimate strength reduction; (3)establish the effects of lateral ties; and(4) establish a relationship betweenaxial load capacity and slenderness.

Forty-seven columns were tested.Both 8 x 8 in. (200 x 200 mm) and 6 x 6in. (150 x 150 mm) cross sections wereexamined with four and eight strandsteel patterns. Forty-one of the 47 totalcolumns were prestressed, of which 26contained ties throughout the columnlength. The columns contained either5/16, 3/8 or 'hs in. (7.9, 9.5 or 11.1 mm)diameter seven-wire strands with an ul-timate tensile strength of 278 ksi (1920MPa) or Grade 60 (420 MPa) #5 (16 mm)bars. Number 3 (10 mm) bars were usedas tie reinforcement in the prestressedcolumns and 1/4 in. (6.4 mm) diameterwire was used with conventionally re-inforced columns.

Results showed that prestressed spec-imens without ties developed a lon-gitudinal crack at midheight. Pre-stressed and conventionally reinforcedmembers without ties are reported tohave failed "violently without warn-ing." At a slenderness ratio of Lid = 30,prestressed concrete columns withoutlateral ties were found to deflect twicethat of tied columns at ultimate load.The author recommended the use of lat-eral ties based on reduced column de-flections. The ultimate load capacity ofprestressed concrete columns with steelpercentages between 0.68 and 2.00 wasfound to be predicted closely by theEuler equation.

Extensive research on prestressedcolumns was conducted at NorthCarolina State University in the late1960's. s ' & 10 Zia and Andrews studied theultimate strength of prestressed con-crete columns, considering columnslenderness and load eccentricity.Eleven 5% in. (145 mm) square columnswere tested with four 1/4 in. (6.4 mm)strands used as longitudinal reinforce-ment and six #2 (6 mm) ties with a 2 in.(50 mm) spacing in the end regions.Slenderness, Lid, was 10, 20 or 30, and

PCI JOURNAL/July-August 1987 49

relative eccentricity, eld, was 0, 0.25 or0.5. The authors noted that eccentricallyloaded columns with a slenderness of 20or 30 failed through instability followedby material failure, whereas concen-trically loaded columns failed by a sud-den violent material failure with "se-vere longitudinal splitting."

Zia, Andrew and Chawla 1 ° extendedthe previous work and analyzed the ef-fect of additional parameters on the ul-timate strength of prestressed concretecolumns: type of cross section (hollow orsolid), type of lateral reinforcement(hoop and spiral), degree of prestress,and eccentricity.

Twenty-four 5 5/s in. (145 mm) squarecolumns were prestressed with four 1/4

in. (6.4 mm) strands. The slendernessremained constant, Lid = 20, while ec-

centricity, eld, ranged from 0.00 to 0.50.Lateral ties (hoops) consisted of six #2(6 mm) bars spaced at 2 in. (50 mm) ateach end and intermediate ties spaced at4 in. (100 mm). Spirals were 5 gauge(5.25 mm) steel with a 2½ in. (65 mm)pitch supplemented by four #2 (6 mm)bars spaced at 2 in. (50 mm) at each end.

For the concentrically loadedcolumns it was reported that "ductilityof the columns was greatly improvedand longitudinal splitting preventedwith the addition of lateral reinforce-ment." It was also noted that "nearly thesame ultimate strength and longitudinalstrains were developed whether thecolumns were reinforced with ties orspirals." The mode of failure for all con-centrically loaded columns was "con-crete crushing."

The eccentrically loaded columnscontinued to have a ductile mode offailure. Strains measured in the spiralreinforcement for eccentrically loadedcolumns were negligible at ultimateload. This was due to the neutral axisshifting toward the compressive facelying inside the tension strand line. Asthe load approached ultimate, com-pressive strains were developed on thecompressive side of the column, result-

ing in a small strain being observed inthe spiral steel. Experimental resultsagreed closely with predictions madewith the equations of Zia and More-adith. 8 Theoretical calculations werefound to be more conservative for con-centrically loaded columns.

Smaller scale prestressed concretecolumns have also been studied. Aroni"tested thirty-six 2.02 x 3 in. (51 x 75 mm)columns prestressed with four 0.198 in.(5.0 mm) diameter wires. Test parame-ters included slenderness, eccentricityof load and the effect of prestress. Ex-perimental results were used to confirmthe analysis of 300 columns which con-sidered additional material and memberparameters.

Since eccentricity of load increasedfrom near to 0.75 eld, the largest nom-inal prestress for which load capacitywas improved increased from 0.1f to0.3f,'. The effect of prestress level wasgreater as column slenderness in-creased.

Design Requirements andApproximate Evaluation ofSlenderness Effects

The PCI Column Committee° pre-sents design recommendations andcommentary for prestressed concretecolumns and bearing walls. Includedamong the design requirements is a pro-cedure for the approximate evaluation ofslenderness effects based on work byChaudwani 12 and Nathan. 12,13, 14

Chapter 12, Design of Columns, pro-vides a reduction factor of 0.85 to beapplied to the ultimate capacity if thedesigner chooses to omit ties. The com-mentary states that this 15 percent re-duction factor is based on the mode offailure of the members, not differencesin the ultimate capacity. The commen-tary cites the test by Lin and Itaya ls andthe work on load capacity of prestressedbearing walls by Baha 16 in support ofthis provision. The reduction factor wasintroduced to permit the treatment of

50

columns and loadbearing walls with thesame set of design provisions.

The approximate method to accountfor slenderness is based on the momentmagnifier method of ACI 318-71, exceptthat the coefficient C. is modifiedslightly, and the EI term revised to ac-count for the fact that the ACI momentmagnifier approach derives EI for rela-tively high values of axial load, as com-pared to the uniaxial capacity of the sec-tion. The effect of sustained loads istreated the same as in ACI 318-71 (sincethe data of this paper concerns onlyshort-term tests, coefficients which ac-count for sustained loads will not be car-ried through the discussion which fol-lows).

Rather than a uniform 2.5 reductionfrom the gross section stiffness to effec-tive stiffness, a sliding reduction factor,A, was proposed, with:

A=17.5-37.5(P/Po)>2.5

when there is no compression flange.For prestressed columns or walls with acompression flange, a slightly differentexpression for A was proposed.

Nathan" later indicated that due todifferences in the moment-curvaturecharacteristics of prestressed and rein-forced concrete columns, both the ACI318-77 and British CF 110 Codes appearunconservative for design of prestressedconcrete columns. Prestressed concretecolumns typically contain a lower steelpercentage and are more slender thanconventionally reinforced columns.These characteristics lead to a higherbalanced load on the interaction dia-gram and an increased likelihood of in-stability failure.

The ACI simplified approach to longcolumns (ACI 318-83, Section 10.11)uses a moment magnification factor in-versely proportional to the columnrigidity or curvature constant, EI. Na-than 13.14.17 recognized that "the mo-ment-curvature relationship becomesnonlinear when following the load pathat the column midheight, whereas the

ACI procedure predicts a linear re-lationship."

Since the benefit of prestressing con-crete columns lies in the long columnrange, many failures are governed bymember instability. In order to give thecolumn load-moment curve the appear-ance of material failure, the rigidity termmust be adjusted. Nathan retained therigidity term as EI/A; however, the fac-tor, A, is redefined to be a function ofboth loading and slenderness:

A=^B

where71 = 100 for (PIP.) < 0.02;

= 3/(PIP0 ) – 0.021(P/P0 )for 0.02 <P/P0 < 0.42; and

71 = 7 for PIP, > 0.420 = 17/(Llr) for sections with

no compression flange, and0 = 31/(Llr) – 0.09 for sections

with a compression flangeLimitations are also recommended for

these equations; the initial columnnominal prestress must be below 600 psi(4 MPa), the steel should be stressed toat least 50 percent of its ultimate tensilestrength, and slenderness, L/r, shouldbe less than 125. The reason for thesestipulations is that high levels of pre-stress induce large initial compressivestresses in the concrete. These largestresses tend to reduce the balancedload by initiating concrete crushing be-fore the steel can reach its theoreticalyield point. The balanced load is alsoreduced if excess steel is used at a lowstress level, where the steel will neverreach a stress level close to yield beforethe concrete begins to crush. Nathan's1983 paper" serves as a basis forrevisions to the PCI Column CommitteeReport, which is presently in the finalstages.

Ties in Concrete ColumnsThis section highlights two investiga-

tions specifically pertinent to the role oftie reinforcement in concrete columns.


Additional useful background for thisdiscussion is found in Refs. 18, 19 and20. A more detailed review is providedin Ref. 6 and a significant body of liter-ature relative to confinement as per-taining to seismic resistant design is notconsidered at all.

Bresler and Gilbert 21 reported that noprior analytical work had considered theeffects of lateral ties on the behavior ofreinforced concrete columns and thatexisting tie requirements were simplytaken from "other codes." ACI 318-56required "lateral ties at least '/4 in. (6.4mm) in diameter, not spaced over 16 bardiameters, 48 tie diameters, or the leastdimension of the column." The tie re-quirements of ACI 318-83 are almost thesame as the 1956 provisions, except thatthe 1956 requirement for every bar to beenclosed by a 90 degree bend of a tiewas relaxed to every other bar in ACI318-63, a code change supported by thework of these authors.

The authors recognized that the clas-sic 1907 tests conducted on plain con-crete by A. N. Talbot show that mem-bers which exhibit a "shear cone" haveroughly a 20 percent higher peak loadthan members whose failure zone ex-hibits longitudinal cracks. It was as-sumed that ties spaced at less than twicethe dimension of the core preventedthese longitudinal cracks from occur-ring. This spacing was determined to beroughly the least column dimension,which was then incorporated into ACI318-63.

A theory was developed to determinethe required tie spacing that would pro-duce longitudinal bar buckling just asthe longitudinal steel began to yield. Asecond theory was developed to deter-mine the required size of a tie to preventbuckling of the longitudinal bars due toinadequate tie stiffness. To evaluatethese theories, four 8 x 8 x 60 in. (200 x200 x 1525 mm) columns were tested.Longitudinal steel was provided by #5(16 mm) deformed bars with yield andultimate tensile strengths of 44.5 and

77.5 ksi (307 and 534 MPa), respec-tively. Ties were #2 (6 mm) deformedbars or '/i6 in. (1.6 mm) diameter steelwire. Tie spacing at midcolumn was 8in. (200 mm), with all longitudinal barsenclosed by a 90 degree corner of a tie.

Based on the test results, the authorsacknowledged that the "maximumstrength of the concrete core wasachieved when the tie spacings did notexceed the least column dimension,"although they suggested that reducedtie spacings be required for memberscontaining high strength longitudinalsteel. They also reported that tiessmaller than the minimum required sizeperformed satisfactorily.

Sheikh and Uzumeri 22.23 studied theability of laterial ties to provide effectiveconfinement for concrete. Increases incolumn strength and ductility were re-lated to the area of concrete that was"effectively confined." Twenty-four 12x 12 x 77 in. (300 x 300 x 1955 mm) rein-forced columns were tested under an in-creasing monotonic concentric load withstrains measured at designated load in-crements. Test parameters includeddistribution of longitudinal steel, tiespacing characteristics of lateral steel,and amount of longitudinal steel.

Results of these experiments showthat increases in the number of longi-tudinal bars tied effectively by lateralreinforcement increased the confinedarea. Strength increases of up to 70 per-cent resulted, due to "well distributedlongitudinal steel," although theamount of longitudinal steel, as a per-centage of core volume, did not affectthe behavior of the confined core.

Decreased tie spacings also increasedthe area of the effectively confined core.When the tie spacings were reduced, in-creased strength and ductility was ob-served. This result was more pro-nounced for members with smaller butmore closely spaced longitudinal barsthan larger, more widely spaced bars. Itwas also noted that an increase in vol-umetric ratio of lateral reinforcement re-

52

8 IN. 8 IN.

zOD

-5/8 IN.(TYP.)

1/4 IN.PLAINROUND

TIE

}EIGHT

—STRANDSONLY

1/2" 0 STRANDS

1/4" 0 STRANDS

Fig. 1. Cross section details of columns tested.

suited in a less than proportional in-crease in strength.

EXPERIMENTAL MATERIALSAND METHODS

For this investigation, 36 columnswere cast and tested in the Major UnitsLaboratory at West Virginia University.The column cross section was 8 x 8 in.(200 x 200 mm), with lengths of 14, 18and 20 ft (4.3, 5.5 and 6.1 m), for result-ing slenderness ratios, Lid, of 21, 27 and30 (L/r for 70, 90 and 100), respectively.

The column identification system liststhe significant column properties in se-quence, such as H1.1-18-END, the"first" 18 ft (5.5 m) column of Batch 1,with 1/2 in. (12.7 mm) diameter strandsand ties only in the end region. In order,the elements of the column identifica-tion scheme are:

Strand diameter:His s in. (12.7 mm)Q is l in. (6.4 mm)

Batch number: 1-9Column number: 1-2Length: 14, 18, or 20 ft

(4.3, 5.5 or 6.1 m)Tie configuration: 4, 8, 16 in.

(100, 200, 400 mm)END region only or NONEEight ties in each end region spaced

at 4 in. (100 mm) were incorporated intoeach column except for two columnswith no ties at all.

Four corner strands were used in eachmember; columns with '/4 in. (6.4 mm)strands were cast in Batches 7 and 8with four additional strands, one at themiddle of each side of the section. Forclarity in the text of this paper, the "(8)"designation will follow any reference toa column in either batch. Fig. 1 showscross section details. Since the samebulkheads were used for all strand ten-sioning operations, the nominal distanceto the strand centroid was the same for'/4 and ½ in. (6.4 and 12.7 mm) diameterstrands. Prestressing steel was 1/4 in. di-ameter Grade 250 (6.4 mm, 1725 MPa)or stress-relieved 1/2 in. Grade 270 (12.7mm, 1860 MPa) seven-wire strand.Stress-strain properties of the Grade 270(1860 MPa) strand were not available tothe investigators. The actual load-straindiagram for the Grade 250 (1725 MPa)strand, and bilinear approximations ofthe stress-strain curves used later inanalysis are given in Ref. 6.


v0J

300

200

Deflection, in.

Fig. 2. Load-deflection curves for concentrically loaded columns.

Table 1. Cylinder compressivestrength of nine batches of concrete.

Batch f/, Psi

1 86602 75503 85304 57905 61306 65907 54408 58609 6100

Note: 1 psi = 6.895 kPa.

Ties were fabricated from 1/a in. (6.4mm) diameter 1020 carbon hot rolledplain round steel. The yield strength ofthe material was 43 ksi (296 MPa), withan ultimate tensile strength of 65 ksi(448 MPa). Ultimate elongation in a 2 in.(50 mm) gauge length was 36 percent.

Concrete was supplied by a localready mix plant. Mix proportions were385 lb per cu yd (228 kg/m 3) water, 752

lb (446 kg/m 3) Type IA portland cement,1663 lb (987 kg/m 3) Ohio River sand, and1188 lb (705 kg/m 3) 3/s in. (446 kg/m 3) (9.5mm) crushed limestone; a high rangewater reducer (Master Builders 400N)was used at the rate of 15 fl oz/cwt ofcement (5.1 1 /m3 of concrete). Average28-day cylinder strengths ranged from5820 to 8910 psi (40 to 60 MPa) asreported in Table 1.

The range of compressive strengthencountered here is wider than may beexpected. Factors implicated in thisscatter are: use of a batch size muchsmaller than delivery truck capacity;problems with homogeneity with dis-tribution of the high range water re-ducer, and intermittently frozen aggre-gate stockpiles.

The prestressing bed consisted ofsteel cross beams post-tensioned to anassembly of reinforced concrete thrustblocks. This configuration permittedcasting four columns in plywood formsfrom each batch, two 14 ft (4.3 m)columns and two longer columns.

54

Y

O0J

150

100

50

Deflection, in.

Fig. 3. Load-deflection curves for eccentrically loaded columns.

Strands were tensioned individuallyusing a center pull jack. Initial tensionlevels were 6300 and 17,280 lb (28 and76.9 kN) for 1/a in. and 1/2 in. (6.4 and 12.7mm) strands, respectively. Loads weremeasured with 10 ton (89 kN) load cellsaffixed to individual strands. Initialstrand stresses were 176 and 113 ksi(1213 and 779 MPa) for the 1/4 and 1/a in.(6.4 and 12.7 mm) diameter strands.With an estimated loss of prestress of 20percent, nominal column prestress isapproximately 865 psi (60 MPa) for col-umns in Sets H1, H2, H3 and H4; 315psi (2.2 MPa) for Sets Q5, Q6, and Q9;and 630 psi (4.3 MPa) in Sets Q7 andQ8.

Due to height limitations in the MajorUnits Laboratory at West Virginia Uni-versity, the column tests were con-ducted in a horizontal position. Thisrequired design and fabrication of a loadframe to be used in conjunction with thestructural floor. Horizontal alignmentsimplified member placement into thetest frame, improved control of end ec-

centricity and improved the safety ofworking conditions.

The eccentricity of applied load wasthe same at each end of the column.Before applying any load to the column,a stringline was stretched along themember and the sweep (member mis-alignment or initial crookedness per-pendicular to its longitudinal axis priorto any application of load) measured.

Load was applied in increments of 25kips (111 kN) until the midcolumn(since columns were tested horizontally,"midcolumn" is used in these descrip-tions, rather than the more familiar"midheight") deflection became notice-ably nonlinear, i.e., began to "creep,"then further readings were taken at 10kip (44 kN) intervals. Midcolumn lat-eral deflection was recorded at 5 kip (22kN) intervals throughout the ascendingbranch of the load-deflection curve.When deflections began to increasewith essentially no increase in load, de-flections and corresponding load cellreadings on the strain indicator were


recorded until failure of the column.Deflection measurements were takenfrom a stringline and plastic scale in 0.1in. (2.5 mm) increments after the dialgauge capacity was exceeded.

TEST RESULTSIn the final report of this PCI Fellow-

ship study, 6 a narrative description isprovided for the mode of failure, anddistinctive aspects of the behavior ofeach of the 36 columns. For purposes ofbrevity in this paper, discussion will belimited to typical load-deflectionbehavior and failure modes, and predic-tion of load capacity through applicationof approximate (moment magnifier)methods. Table 2 reports the initialcrookedness and eccentricity of load foreach column, along with load magnitudecorresponding to the noticeable onset ofnonlinearity (creep load) and maximumand failure loads and deflections. Inmost cases, the loading system per-mitted observation of a portion of thedescending branch of the load-deflec-tion curve.

The load-deflection curves reflect as-pects of column length and load eccen-tricity in a manner that is generallystraightforward and essentially intui-tive. Graphs of load-deflection curvesfor columns with concentric and eccen-tric loads are shown in Figs. 2 and 3.The deflection indicated in each case isthe lateral deflection of the columns,measured in the plane of bending half-way between the ends of the column.The reader should note that differencesin prestress level and relative eccen-tricity, in addition to column slender-ness, cause the 18 and 20 ft (5.5 and 6.1m) columns to behave more similarlythan might be expected.

If ties make a definite contribution tomember ductility and toughness, thiscontribution should be evident in theload-deflection curves. If lateral ties caninfluence load-deflection behavior, it

seems that this effect should be notice-able for relatively short, concentricallyloaded columns. Fig. 4 shows load-de-flection curves for 14 ft (4.3 m) concen-trically loaded columns with four 1/2 in.(12.7 mm) strands. Here the columnswithout lateral ties do have a lowercapacity, but this is probably due to in-creased initial crookedness for thesecolumns. For the shortest columnsloaded eccentrically, there is not muchdifference in the load-deflection curveswith or without ties, as shown in Fig. 5.A similar trend is noted for 18 ft (5.5 m)columns, as shown in Fig. 6. Here thestrength of Batch H3 is significant.

Overall, the authors believe that thesecurves indicate that ties have no in-fluence on the shape of the curve in theregion of the peak load, and minimaleffect otherwise. For Column H4.1-18-4(see Fig. 6), toughness was phenomenal.However, near the peak load the load-deflection curve was essentially thesame as for columns with ties spaced at 8in. (200 mm), and columns with end ties,only. Because of the many factors thatmake comparison of load-deflectioncurves more complicated than antici-pated, it seems that the best means ofevaluating behavior of a particularcolumn is to develop a theoretical load-deflection curve, and to compare ob-served and theoretical curves.Aburazza24 has developed one such pro-cedure; however, the authors have yet toimplement that method and makedetailed comparisons.

The influence of ties may also be dis-covered through examination of failuremodes. Here, there were some dif-ferences, although the significance ofthis difference is open to discussion.The 14 and 18 ft (4.3 and 5.5 m) columnsfailed by crushing and diagonal crackingat the midcolumn region. This is true forboth columns with concentric and ec-centric loads. Generally, there was not anoticeable difference in the violence offailure for columns without ties in themidcolumn region, and columns with

56

Table 2. Column loads and deflections.

Column

Eccentricity, in. Load, kips Deflection, in.

MaxInitial* Loath Creep Maximum Failure Load Failure

H1.1-14-END 0.25 1.75 105 124 108 1.91 2.7H1.2-14-END 0.19 0 265 280 259 1.29 1.85H2.1-14-END 0.19 0 250 296 290 1.05 1.4H2.2-14-END 0.23 1.75 115 126 107 1.53 3.07H3.1-14-8 0.13 1.75 125 142 136 1.5 2.13H3.2-14-8 0.15 0 285 327 327 1.2 1.25H4.1-14-4 0.13 0 220 250 248 0.95 1.3H4.2-14-8 0.13 1.75 110 118 107 1.4 2.4

Q5.1-14-8 0.19 1.75 80 93 59 1.5 3.2Q5.2-14-END 0.19 0 210 244 244 1.3 1.3Q6.1-14-8 0.13 0 175 240 - 0.91 -Q6.2-14-END 0.19 1.5 100 115 92 1.1 2.1Q7.1-14-8(8) 0.06 1.5 95 120 102 1.2 2.4Q7.2-14-END(8) 0.00 1.5 115 120 99 1.3 2.5Q8.1-14-8(8) 0.06 0 225 276 270 0.85 1.1Q8.2-14-END(8) 0.13 0 180 231 228 1.2 1.4Q9.1-14-NONE 0.10 0 210 270 - 1.0 -Q9.2-14-NONE 0.10 0 195 242 238 1.3 1.5

H1.1-18-END 0.44 0 - 200 164 - -H1.2-18-END 0.44 1.75 - 88 - - 5.5H2.1-18-END 0.40 0 - 185 150 1.50 2.77H2.2-18-END 0.38 1.75 - 88 67 1.66 4.77H3.1-18-8 0.38 0 - 245 245 2.0 2.0H3.2-18-8 0.31 1.75 - 108 81 2.5 4.47H4.1-18-4 0.38 1.75 - 84 32 1.7 7.0H4.2-18-8 0.25 1.75 - 88 66 1.9 4.2

Q5.1-20-8 0.44 0 135 140 140 1.4 1.4Q5.2-20-END 0.44 0 100 148 147 1.5 1.7Q6.1-20-8 0.41 1.0 80 85 46 1.0 3.0Q6.2-20-END 0.42 1.0 85 90 42 1.34 4.7Q7.1-20-8(8) 0.24 0 100 153 147 1.6 2.0Q7.2-20-END(8) 0.20 0 105 156 142 1.6 2.3Q8.1-20-8(8) 0.39 1.0 90 92 53 1.5 5.0Q8.2-20-END(8) 0.32 1.0 93 100 51 2.0 5.6Q9.1-20-16 0.50 0 135 156 156 2.0 2.0Q9.2-20-16 0.37 1.0 85 86 36 1.5 5.5

*Column sweep, or initial crookedness.tEccentricity of load, applied equally at both ends of column.Note: 1 in. = 25.4 mm; 1 kip = 4.448 kN.

ties spaced at 8 in. (200 mm), the ACICode maximum for this column con-figuration.

Although the failure modes were gen-erally the same for these columns with

and without ties, some differences arefound in the details of the column fail-ure. For columns with ties, diagonalcracking progressed from the mid-column center. This formed the begin-


0 I e S

Deflection, in.

Fig. 4. Load-deflection curves for concentrically loaded 14 ft (4.3 m) columns, four 1/2 in.(12.7 mm) strands.

ND

0 I 2 3

Deflection, in.

Fig. 5. Load-deflection curves for eccentrically loaded 14 ft (4.3 m) columns, four 1/2 in.

(12.7 mm) strands.

0J

300

200

uf:

00J

150

100

50

58

ning of a longitudinal split which wasdirected by the ties out of the columncore and into the shell, forcing spallingof the cover. Even so, it is important torecognize that the columns with tiesspaced at 8 in. (200 mm) were effec-tively destroyed, and there was no re-serve capacity. The longitudinal splitpropagated a significant distancethrough the center of concentricallyloaded columns where no ties werepresent in the midcolumn region. Thissplitting action caused the strands tobow and buckle, forcing the concreteshell to spall away.

For the eccentrically loaded columns,the failure sequence was a three-stepprocess initiated by localized concretecrushing and more widespread spallingin the compression zone. The secondstage of failure was formation of diago-nal compression zone cracking throughthe column core. Finally, the concretecracked and fell away on the tension side.

The columns with a length of 20 ft (6.1m) (Lid = 30, L/r = 100) failed in a morecharacteristic stability mode. Generally,diagonal compression zone cracking wasnot observed; multiple flexure crackswere observed, although these wereusually dominated by a single crack atmidcolumn which passed straightthrough the section. These columnssimply snapped. A notable aspect of the20 ft (6.1 m) columns was the apparentincrease in core confinement apparentfor columns with eight strands. Photo-graphs taken during and after thecolumn test are useful in comparing thenature of column failure modes. Figs. 7through 10 compare sets of columns, andFigs. 11 through 24 will show additionaldetails.

Fig. 7 compares 14 ft (4.3 m) columnswith four '/2 in. (12.7 mm) strands withand without ties. One concentricallyloaded and one eccentrically loadedcolumn is shown in each set. The con-

D

0J

^oL

100

50

Deflection, in.

Fig. 6. Load-deflection curves for eccentrically loaded 18 ft (5.5 m) columns, four 1/2 in.(12.7 mm) strands.


Fig. 7. Columns [14 ft (4.3 m) long I with 1/2 in. (12.7 mm) strands.

Fig. 8. Columns [14 ft (4.3 m) long] loaded concentrically.

60

Fig. 9. Columns [20 ft (6.1 m) long] loaded concentrically.

Fig. 10. Columns [18 ft (5.5 m) long] loaded eccentrically.


Fig. 11. Delaminated tension face cover in column with eight 1/4 in. (6.4 mm) strands.

centrically loaded columns are in themiddle of the picture.

Fig. 8 shows concentrically loaded 14ft (4.3 m) columns with eight 1/4 in. (6.4mm) strands. In the column with endties only, shown at the right, concretedelaminated for some distance along thetension side strands. The column withties spaced at 8 in. (200 mm) developeda longitudinal split along the midsidestrand. For 20 ft (6.1 m) columns with anominally concentric load, effects oflength and initial crookedness create theappearance of a bending failure, withprimary spalling generally confined toone face, as shown in Fig. 9. Eccen-trically loaded 18 ft (5.5 m) columns areshown in Fig. 10. The column labeled"None" actually has ties in the END re-gion. The presence and spacing of tiesdoes change the appearance of the fail-ure zone without appreciable change inpeak load.

A close-up of Column Q8.1-14-8(8) isshown in Fig. 11. This is the columnshown to be left in Fig. 8. Note the ex-

tent of delamination along the tensionstrands where three strands on a face areprovided.

Fig. 12 shows concentrically loadedcolumn H4.1-14-4 and eccentricallyloaded column H4.1-18.4. The 18 ft (5.5m) column of this pair was perhaps thesingle most interesting test of this series.Although load capacity and the load-de-flection behavior near the peak were es-sentially unaffected by the close tiespacing, the column held together atvery large displacements. Fig. 13 showsthe midcolumn region of ColumnH4.1-18-4 just before the test washalted. At this point, the column hasdeflected nearly 7 in. (175 mm). Fig. 14shows the failure zone while some loadremains on the column. Note that at thespall, one 1/2 in. (12.7 mm) strand hasbuckled and opened to a width of about1'/2 in. (40 mm).

The strand buckling was noticed inthe failure of other columns as well.Figs. 15 and 16 show failures of eccen-trically loaded columns H2.2-18-END

62

Fig. 12. Columns [14 and 18 ft (4.3 and 5.5 m) long] with 4 in. (100 mm) tie spacing.

and H4.2-18-8.Severe longitudinal splitting was en-

countered in concentrically loadedcolumns with very wide tie spacing, orno ties at all in the column midlength, asshown in Figs. 17 and 18. Where no tieswhatsoever were provided, the splitextended through about one-half of thecolumn length, as shown in Fig. 19.

As slenderness and eccentricity in-crease, the presence of column ties ishardly noticed near the failure zone.Figs. 20 and 21 show eccentricallyloaded 20 ft (6.1 mm) long companionspecimens from the same batch Q7.1-20-END(8) and Q7.2-20-8(8). For thecolumn without ties in the failure re-gion, a single flexural crack seems topredominate, but the appearance isotherwise much the same. Similar be-havior is noted for 18 ft (5.5 m) columns.Figs. 22 and 23 show Columns H1.2-18-END and H4.2-18-8.

In some instances, ties were bentduring the column failure, and in thecase of one column with 1/4 in. (6.4 mm)

strands, a single broken tie was discov-ered, as shown in Fig. 24.

Although validation of analyticalmethods for predicting load carrying ca-pacity was not an original goal of thisstudy, it is a valuable benefit of the testseries. Table 3 compares the results forobserved maximum load with predic-tions made by the moment magnifierapproach and various methods for es-timating the column flexural rigidity, El.These are the two approaches given inACI 318 using Eqs. (10-10) and (10-11),the 1976 PCI Column Committeereport,' and the equation developed byNathan.'s

Although the work by Nathan 13 indi-cated that the usual ACI methods for es-.timating El are not applicable, theywere obtained for comparison. Assess-ment of Nathan's' s more recent ap-proach is of particular interest, since thismethod forms the basis for a pendingmajor revision of the PCI ColumnCommittee report. It must also be notedthat since the investigation is assessing


Fig. 13. Column H4.1-18-4 at large deflection.

Fig. 14. Column H4.1-18-4 failure zone.

64

Fig. 15. Column H2.2-18-END failure zone (eccentric loading).

1 sl1. ..

Fig. 16. Column H4.2-18-8 failure zone (eccentric loading).


Fig. 17. Column Q9.1-14-NONE failure zone (concentric loading).

Fig. 18. Column Q5.2-14-END longitudinal split (concentric loading).

66

Fig. 19. Column 09.2-14-NONE failure zone and longitudinal split (concentric loading).

Fig. 20. Column Q7.2 -14-END (8) failure zone (eccentric loading).


strength prediction, the capacity reduc-tion factor was taken as unity for thesecomparisons.

The estimation of load capacity wasaccomplished by developing a micro-computer program to compute the

load-moment interaction diagram for thecross section, and to then construct in-teraction diagrams which account forslenderness by the various momentmagnifiers. The basic section inter-action diagram was determined by ap-

Fig. 21. Column Q7.1-14-8 (8) failure zone (eccentric loading).

t

Fig. 22. Column H4.2-18-8 failure zone (eccentric loading).

68

plying principles of equilibrium, straincompatibility and knowledge of con-structive behavior for steel and con-crete. The ACI rectangular stress blockwas used to estimate the magnitude andlocation of the concrete compressionforce for cases where the compressionzone was wholly within the section.Bilinear stress-strain diagrams were as-sumed for the reinforcement. A straightline interpolation was assumed betweenthe capacity at zero moment (pure com-pression) and the point on the inter-action curve where the neutral axis co-incides with the lower extreme of thecross section.

The comparison of predicted and ob-served load capacity is accomplishedmore easily by graphical means. Figs. 25to 28 represent interaction diagrams forcolumns in four of the nine batches ofconcrete in this investigation. Theseinteraction diagrams have been selectedfrom Ref. 7 to include the range of pa-rameters in this study: column length,lateral reinforcement and longitudinalreinforcement.

On each figure, three curves, four ra-

dial lines and four points are plotted.The curve farthest from the origin is theshort column interaction diagram for thecross section. The middle curve of eachset is an interaction diagram which ac-counts for length effects in the 14 ft (4.3m) columns, and the curve closest to theorigin accounts for length with either 18or 20 ft (5.5 or 6.1 m) columns, de-pending on the batch. Each point corre-sponds to the failure load of a particularcolumn, labeled in each case, and plot-ted on a radial line from the originwhich corresponds to the total eccen-tricity of load (the sum of the initialcrookedness and applied eccentricity).

These figures indicate that for the pa-rameters of this investigation, Nathan's16method for simplified calculation ofslenderness effects is an accurate meansof estimating load capacity of pre-stressed columns. It appears from thesetests that the simplified moment magni-fier method can be applied for levels ofinitial prestress substantially higherthan recommended previously.

Table 3 indicates that the methods ofACI 318 for estimating stiffness lead to

Fig. 23. Column H1.2-18-END failure zone (eccentric loading).


Fig. 24. Fractured tie in column with 1/4 in. (6.4 mm) strands.

quite inaccurate predictions of columnload capacity in many cases, and fairlyaccurate predictions in others. The rea-son for this is that the ACI 318 slendercolumn "curves" are nearly straightlines which run from the "pure mo-ment" point on the short column inter-action diagram to a point on the load axisat 20 to 40 percent of P o . The ACI 318prediction by use of ACI Code Eqs.(10-10) and (10-11) can be fairly accuratefor a limited range of behavior. ThePCI-76 prediction is calibrated for ex-tremely low axial loads and significantslenderness, and consequently performspoorly for the parameters studied here.Fig. 28 illustrates the predicted longcolumn behavior using various methodsfor an 18 ft (5.5 m) column in Batch H3.

SUMMARYIn this investigation, thirty-six pre-

stressed concrete columns were fabri-cated and tested at West Virginia Uni-versity. The investigation was initiated

to study the behavior of prestressedconcrete columns as influenced by thepresence of lateral reinforcement in theform of isolated ties. The results of theinvestigation also provide a means forevaluating existing theoretical modelsfor column load capacity. This discus-sion will summarize the investigation,present conclusions about the effective-ness of tie reinforcement, and recom-mend both a design interpretation of thestudy and aspects requiring furtherevaluation.

The thirty-six 8 x 8 in. (200 x 200 mm)columns tested in this study encom-passed a wide variety of parameters.The casting bed limited each batch to atotal of four columns — in each of ninebatches two 14 ft (4.3 m) columns werecast along with either two 18 or 20 ft (5.5or 6.1 m) columns. Consequently, six-teen of the columns were 14 ft (4.3 m)long; eight were 18 ft (5.5 m) long, andten were 20 ft (6.1 m) long.

Sixteen of the columns were rein-forced with four '/s in (12.7 mm) Grade270 (1860 MPa) seven-wire strands. In

70

Table 3. Test and predicted load capability.

Column f^, Test load,

Predicted load, kips

ACI ACIpsi kips 10-11 10-10 PCI-76 Nathan*

H1.1-14-END 8660 124 126 96 36 129H1.2-14-END 8660 280 231 150 231 319H2.1-14-END 7550 296 214 142 214 289H2.2-14-END 7550 126 116 90 34 126H3.1-14-8 8530 142 128 98 36 138H3.2-14-8 8530 327 233 151 233 324H4.1-14-4 5790 250 190 130 190 247H4.2-14-8 5790 118 101 80 32 118

Q5.1-14-8 6130 93 97 66 28 99Q5.2-14-END 6130 244 197 111 197 265Q6.1-14-8 6590 240 204 113 204 277Q6.2-14-END 6590 115 106 71 30 119Q7.1-14-8 5440 120 100 71 31 121Q7.2-14-END 5440 120 102 72 31 124Q8.1-14-8 5860 276 199 113 199 264Q8.2-14-END 5860 231 190 110 190 247Q9.1-14-NONE 6100 270 199 110 199 268Q9.2-14-NONE 6100 242 199 110 199 268

H1.1-18-END 8660 200 133 105 28 209H1.2-18-END 8660 88 90 75 24 82H2.1-18-END 7550 185 127 101 29 196H2.2-18-END 7550 88 84 72 21 87H3.1-18-8 8530 245 136 106 27 212H3.2-18-8 8530 108 91 76 24 89H4.1-18-4 5790 84 72 64 19 87H4.2-18-8 5790 88 74 66 19 92

Q5.1-20-8 6130 140 94 56 17 148Q5.2-20-8 6130 148 94 56 17 148Q6.2-20-8 6590 85 72 45 16 90Q6.2-20-END 6590 90 72 45 16 89Q7.1-20-8 5440 153 91 57 19 145Q7.2-20-END 5440 156 92 58 19 147Q8.1-20-8 5860 92 70 48 16 99Q8.2-20-END 5860 100 71 49 16 102Q9.1-20-16 6100 156 89 53 15 138Q9.2-20-16 6100 86 69 44 14 96

*As described in Ref. 13.Note: 1 psi = 6.895 kPa; I kip = 4.448 kN.

order to limit the level of nominal pre-stress in these columns, the tendonjacking force was significantly less thanpermitted by present design codes. Theremaining columns were reinforcedwith V4 in. (6.4 mm) Grade 250 (1725

MPa) seven-wire strand. These strandswere jacked to the permissible limit. Ofthese columns twelve had four strandsand eight contained eight, resulting in anominal prestress of 315 and 630 psi (2.2and 4.3 MPa), respectively.


500

00 500

Moment, in.—kFig. 25. Interaction diagrams for Batch 2.

500

Im

OOJ

00 500


M

OO

72

500

00 500


short column

500- –AC! eq 10-11i-^ A C I eq 10-10

PCI 76Nathan 83

0O

00 500

Moment, in.—kFig. 28. Comparison of predicted long column behavior.

W4

D0J


Lateral tie reinforcement was fabri-cated from 1/4 in. (6.4 mm) plain roundbar stock. Where provided, tie spacingswere 4, 8 or 16 in. (100, 200 or 400 mm).Eighteen of the columns had ties only inthe end region [eight ties spaced at 4 in.(200 mm), beginning 2 in. (100 mm)from the column end ], and two columnshad no ties at all. Fourteen of the eigh-teen columns with ties throughout themember used spacings of 8 in. (200mm) — the maximum permitted by theACI Building Code, ACI 318-83.

Approximately half of the columnswere loaded concentrically. However,all but one column had some initial ec-centricity at midcolumn. Initial midcol-umn eccentricity was significantlygreater for the longer columns. Columnfailure loads ranged from a low of 85kips (378 kN) in the case of an eccentri-cally loaded 20 ft (6.1 m) column to ahigh of 327 kips (1454 kN) with a 14 ft(4.3 m) column loaded concentrically.

Column load-deflection curvesshowed minimal dependence on thepresence of lateral reinforcement. Thatis, columns with and without ties be-haved approximately the same with re-gard to:

— Peak load— Peak load deflection— Ability to withstand deformations

on the descending branch of theload-deflection curve

The 14 and 18 ft (4.3 and 5.5 m) longcolumns failed by crushing at the mid-column region. This was true for bothconcentrically and eccentrically loadedcolumns, and independent of the pres-ence of ties. All of these columns werethus material failures. Generally, therewas not a noticeable difference in theviolence of failure for columns withoutties in the midcolumn region, and forcolumns with ties spaced at 8 in. (200mm), the ACI maximum spacing for thiscolumn configuration.

Although the failure modes were gen-erally the same for columns with andwithout ties, differences were noted in

the details of column failure. Forcolumns with ties, diagonal crackingprogressed from the center of the crosssection at midcolumn. This crackdeveloped into a longitudinal split, butthe ties apparently directed the splitoutside of the column core, and forcedspalling of the concrete outside the tie.Even so, it is important to note thatcolumns with ties spaced at 8 in. (200mm) were destroyed, and there was noreserve capacity. For some columnswithout ties at midcolumn, the longi-tudinal split propagated through thecore for a significant distance from mid-column. This splitting action appearedto cause the strands to bow and buckle,forcing the concrete shell to spall away.

For eccentrically loaded columns,failure was a three-step process initiatedby spalling in the compression zone.The next stage of failure was develop-ment of diagonal compression zonecracking through the column core, fol-lowed finally by breakoff of concrete onthe tension side.

The 20 ft (6.1 m) columns appeared tofail in a stability mode. Generally, diag-onal compression zone cracking was notobserved; flexure cracks, usually domi-nated by a single crack at midcolumn,passed straight through the section.These columns simply snapped. A no-table aspect of the 20 ft (6.1 m) columntests was the significant increase in coreconfinement apparent for columns witheight strands.

Observed column maximum loadswere compared with theoretical predic-tions using the approximate evaluationof slenderness effects by the momentmagnification procedure ofACI 318. Forprestressed concrete columns, theflexural rigidity may differ significantlyfrom that predicted from the equationsgiven in Chapter 10 of ACI 318-83. Con-sequently, load capacity predictionsbased on the moment magnifier methodmay be inaccurate for columns with ap-preciable axial load if procedures basedon Eqs. (10-10) or (10-11) of ACI 318-83

74

are employed. For the columns tested inthis study, the procedures described byNathan's provided extremely accurateestimates of the column maximum load.Predictions based on other estimates ofthe column flexural stiffness were sub-stantially less accurate.

It should be emphasized that the pre-dicted column capacities cited in thisreport are based on the assumption ofpinned-end conditions, measured con-crete strength at the test age rather thanthe 28-day cylinder strength, and thattotal eccentricity (column crookednessplus load eccentricity) was used to com-pare theory with experiment. The use ofthe initial crookedness is particularlyimportant in the prediction of loadcapacity of concentrically loadedcolumns.

CONCLUSIONSConclusions to be drawn from this

study are directly applicable to the testparameters considered, namely squareprestressed concrete columns with aslenderness, Lid, in the range of 20 to30, lateral tie spacings in the range ofone-half to two times the column width,and concrete strength in the range of6000 to 9000 psi (40 to 60 MPa). Exten-sion of these conclusions to other con-ditions requires further evaluation andjudgment:

1. Ties do not lead to a significant in-crease in column capacity at peak loads.

2. Ties do not appreciably limit thecolumn deflections at peak load, and donot consistently influence the ability tosustain deformations on the descendingbranch of the load-deflection curve.

3. Ties do not influence the explosivenature of concentrically and eccentri-cally loaded columns which experiencematerial failures. This was observed incolumns up to 18 ft (5.5 m) long (Lid =27), and a total initial eccentricity ofabout 2.2 in. (55 mm) (eld = 0.275).

4. Ties do alter the details of the fail-ure region in the case of failure by mate-

rial crushing at midcolumn. Ties directcompression zone cracks outside thecore within one or two tie spacings of 8in. (200 mm), and limit the tendency forlongitudinal splitting. However, thefailure is just as complete with or with-out ties.

5. Columns [20 ft (6.1 m) ] loadedconcentrically or eccentrically (Lid = 30and a total eid up to 0.176) failed in astability mode, with tension side strandstypically snapping as large deformationsoccurred.

6. Within the range of parametersconsidered in this investigation, there isno logical justification for a strengthreduction if ties are omitted from pre-stressed concrete columns. While thereis some slight difference in the details ofan otherwise explosive failure, theauthors believe that the difference isnowhere near as significant as impliedby the 0.85 reduction factor suggested inthe 1976 recommendations of the PCIColumn Committee.4

7. Designers should be more con-cerned about the difference in capacityreduction factors between tied columnsand spiral columns, rather than intro-ducing a new reduction factor for omis-sion of ties. This whole discussion mustbe considered in light of the philosophi-cal considerations of the role of ductilityin structural safety requirements. It maybe simply inappropriate to link strengthrequirements and ductility. Perhaps,ductility should be addressed separatelyand explicitly, where the provision ofties might be based, in part, on a checkto see if the concrete neutral axis wouldlie outside the tie location at the strengthlimit state. If the tie does not cross theneutral axis, it cannot be effective.

8. The approximate method forevaluating slenderness effects of pre-stressed columns developed byNathan' s provides an accurate means ofaccounting for load capacity in rec-tangular prestressed columns. However,no tests in this study are available toverify the method where P/PO < 0.2.


DESIGN IMPLICATIONSIn cases where tied columns are ap-

propriate, columns without tie rein-forcement behave about the same ascolumns with isolated ties spaced in ac-cordance with the requirements of ACI318-83.

However, from the tests conducted inthis investigation, it is apparent that, ifhigh ductility or toughness demands areto be made on a prestressed concretecolumn, square tie or hoop reinforce-ment cannot provide adequate confine-ment of the concrete core if tie spacingsare on the order of the least columndimension. Judging from the behavior ofconventionally reinforced concretecolumns, a circular spiral would appearto be more suitable to applications witha high ductility demand. It seems un-likely that a square spiral could be aseffective, since the confining action of asquare spiral would be obtained throughthe legs of the spiral acting in flexure,rather than the hoop tension approachedin the case of circular spirals. Thesquare spiral may have advantages overisolated ties from a fabricator's stand-point. Placement and tying of tie rein-forcement is difficult and tedious, withboth factors affecting productivity. Asimilar advantage might be obtainedthrough use of welded wire fabric cages,where lateral reinforcement is to beprovided.

The failure modes observed forcolumns with slenderness in the rangeof Lid = 21 to 30 is such that there doesnot seem to be an effective contributionof isolated ties for a tie spacing on theorder of the least column dimension (8in. or 200 mm). For the one 18 ft (5.5 m)column with ties spaced at 4 in. (100mm) through the midcolumn, the abilityto sustain deformations on the des-cending branch of the load-deformationcurve was rather remarkable, althoughother aspects of that column's load-de-flection behavior were identical tocolumns with no ties in the midcolumn

region. This might suggest that someadvantage could be achieved if tiespacings were limited to something likethe least leg size of the tie or be relatedspecifically to the column core dimen-sion.

It is important to recall that Breslerand Gilbert 21 concluded that the "shearcone" was formed in members wherethe lateral reinforcement was spaced notfarther than the core dimension. Forprestressed columns, which often aresquare members, the minimum crosssection dimension may be appreciablylarger than the core. This is suggestedsince the columns studied in this inves-tigation had a relatively large cover onthe ties and prestressing strands, andsome commercially fabricated columnsappear to be similar.

Before urging wholesale adoption ofthis more restrictive tie spacing re-quirement, it must be recognized thatthe concentrically loaded columnH4.1-14-4 did not show an appreciableadvantage by virtue of a closer tie spac-ing.

This study has indicated that tie rein-forcement is generally not a satisfactorymeans to enhance the ductility of pre-stressed concrete columns. It should benoted that round spiral reinforcementmay also have liabilities. To achieve thebenefits of spiral confinement, a veryhigh spiral steel ratio is required to meetthe requirements of the "minimum spi-ral," described in Chapter 10 of ACI318. For example, in the case of a 16 in.square (400 mm) column with 8000 psi(55 MPa) concrete and the limiting spi-ral strength of 60 ksi (400 MPa), a 3 in.(20 mm) spiral with a 3 in. (75 mm) pitchwould be required.

The ACI Building Code, ACI 318,should recognize the use of equations,such as developed by Nathan,' 6 whichprovide a more accurate assessment ofthe flexural rigidity of prestressed con-crete columns, and which leads to abetter prediction of the load carryingcapacity.

76

FURTHER STUDYThis study has demonstrated that iso-

lated ties, placed at the maximum spac-ing permitted by ACI 318-83, have littlepractical effect on the behavior of pre-stressed concrete columns up to, andbeyond, peak load. Some differences arenoted in the exact character of the fail-ure, however, the essence is the same. Itappears that the use of isolated ties issimply an impractical way to effectivelyconfine concrete in these columns.

Square spirals or welded wire fabriccages may be considered as alternatemeans of providing lateral reinforce-ment. Speculation based on the experi-ence of this study would lead to the as-sessment that these schemes would bepractically and economically superior toisolated ties, but the effectiveness isopen to question, and substantiationthrough further testing and evaluation isrequired.

For cases where high ductilityor toughness demands exist, a round spi-ral is most probably the best solution,though spiral requirements may begreat. Since prestressed concretecolumns are manufactured under plantconditions with abutments holding thestrands in place while the concrete iscast, it seems that placing the primaryreinforcement in a circular patternshould not be a particular liability.

Further studies should be accom-plished to better establish the mecha-nisms of behavior for closely spacedties, welded wire fabric, and square orround spiral reinforcement in pre-stressed concrete columns. Thebehavior of round spirals may be some-what different than in conventionallyreinforced columns, since the slightbulging effect introduced into a pre-stressed column at transfer would tendto put a small pretension in the spiral aswell. However, this effect may be of lit-tle practical experience.

To date, tests have been made on lab-oratory scale columns. It appears thattests are needed on full scale members,or another type of specimen, so as toevaluate the effectiveness of lateral tiereinforcement when the tie crosses theneutral axis and as such may be moreeffective in confining the concretecompression zone. Many design engi-neers may conservatively insist on tiesin prestressed concrete columns whereseismic concerns are paramount. Theyshould recognize that ties in substantialexcess of the minimum requirementsmay be needed to provide acceptablemember behavior.

Tie reinforcement should also be pro-vided for prestressed columns with sig-nificant amounts of nonprestressed re-inforcement.

ACKNOWLEDGMENTThe investigation reported in this

summary paper was initiated under theauspices of the PCI Fellowship Pro-gram. Major donations of material andservices by Shockey Brothers, Inc.,Belot Concrete, Centurial ConcreteProducts, Florida Wire & Cable Co.,Bethlehem Steel, USX Corp., and Ener-pac, Inc., enabled this investigation tobe accomplished.

Additional support for the experi-mental studies was provided by the fol-lowing PCI Producer Members:

Bladholm BrothersBlakeslee Prestress Inc.Joseph P. Carrara & Sons, Inc.Southern Prestressed ConcreteForest City Dillon Precast

Systems, Inc.High Concrete Structures, Inc.Atlanta Structural Concrete Co.The authors would also like to recog-

nize Alex Aswad and A. E. N. Osbornefor their useful and thoughtful review ofthe PCI Fellowship Final Report.


REFERENCES1. Salmons, J. R., "Research Needs of the

Precast Prestressed Concrete Industry,"PCI JOURNAL, V. 26, No. 6, Novem-ber-December 1982, pp. 22-30.

2. Nilson, A. H., Design of PrestressedConcrete, John Wiley & Sons, Inc., NewYork, New York, 1978, 526 pp.

3. ACI Committee 318, "Building CodeRequirements for Reinforced Concrete,"American Concrete Institute, Detroit,Michigan.

ACI 318-56, 1956, 73 pp.ACI 318-63, 1963, 144 pp.ACI 318-71, 1971, 78 pp.ACI 318-77, 1977, 103 pp.ACI 318-83, 1983, 111 pp.

4. PCI Committee on Prestressed ConcreteColumns, "Recommended Practice forthe Design of Prestressed ConcreteColumns and Bearing Walls," PCIJOURNAL V. 21, No. 6, November-De-cember 1976, pp. 16-45.

5. Zia, P., and Andrew, J. R., "UltimateStrength of Rectangular PrestressedConcrete Under Concentric and Eccen-tric Loadings," North Carolina StateUniversity, Raleigh, September 1967.

6. Carinci, C. A., and Halvorsen, G. T., "TieRequirements for Prestressed ConcreteColumns," Final Report of PCI Fellow-ship Investigation, Department of CivilEngineering, West Virginia University,Morgantown, July 1986.

7. Jernigan, A. M., "Some Studies on theBehavior of Prestressed ConcreteColumns," MS Thesis, Department ofCivil Engineering, University of Florida,Gainesville, June 1956.

8. Zia, P., and Moreadith, F. L., "UltimateLoad Capacity of Prestressed ConcreteColumns," ACI Journal, Proceedings, V.63, No. 7, July 1966, pp. 767-787.

9. Zia, P., and Guillermo, E. C., "Com-bined Bending and Axial Load in Pre-stressed Concrete Columns," PCIJOURNAL, V. 12, No. 3, June 1967, pp.52-59.

10. Zia, P., Andrew, J. R., and Chawla, M. S.,"Prestressed Concrete Columns UnderConcentric and Eccentric Loadings,"North Carolina State University, Raleigh,July 1969.

11. Aroni, S., "The Strength of Slender Pre-stressed Concrete Columns," PCI

JOURNAL, V. 13, No. 2, April 1968, pp.19-33.

12. Chaudwani, R., and Nathan, N. D., "Pre-cast Prestressed Sections Under AxialLoad and Bending," PCI JOURNAL, V.16, No. 3, May-June 1971, pp. 10-19.

13. Nathan, N. D., "Slenderness of Pre-stressed Concrete Beam-Columns," PCIJOURNAL, V. 17, No. 6, November-De-cember 1972, pp. 45-57.

14. Nathan N. D., "Applicability of ACISlenderness Computations to Pre-stressed Concrete Sections," PCIJOURNAL, V. 20, No. 3, May-June 1975,pp. 68-85.

15. Lin, T. Y., and Itaya, R., "A PrestressedConcrete Column Under EccentricLoading," PCI JOURNAL, V. 2, No. 3,December 1957, pp. 5-17.

16. Baha, Z., "A Study of Ultimate Capacityof Prestressed Concrete Load-BearingWall Panels," PhD Dissertation, De-partment of Civil Engineering, NorthCarolina State University, Raleigh,1973.

17. Nathan, N. D., "Slenderness of Pre-stressed Concrete Columns," PCIJOURNAL, V. 28, No. 2, March-April1983, pp. 50-77.

18. Burdette, E., and Hilsdorf, H. K., "Be-havior of Laterally Reinforced ConcreteColumns," Journal of the Structural Di-vision, ASCE, V. 97, No. ST2, February1971, pp. 587-602.

19. Pfister, J. F., "Influence of Ties on theBehavior of Reinforced ConcreteColumns," ACI Journal, Proceedings, V.61, No. 5, May 1964, pp 521-538.

20. Ray, H. E. H., and Sozen, M. A., "Duc-tility of Concrete," Flexural Mechanicsof Reinforced Concrete," Proceedings ofthe ACI-ASCE International Sympo-sium, ACI Special Publication SP-12,American Concrete Institute, Detroit,Michigan, 1965, pp. 213-235.

21. Bresler, B., and Gilbert, P. H., "Tie Re-quirements for Reinforced ConcreteColumns," ACI Journal, Proceedings, V.58, No. 11, November 1961, pp. 555-570.

22. Sheikh, S. A., and Uzumeri, S. M.,"Mechanism of Confinement in TiedColumns," Preprint from the Proceedingsof Seventh World Conference on Earth-quake Engineering, Istanbul, Turkey,

78

1980. 1980, pp. 1079-1102.23. Sheikh, S. A., and Uzumeri, S. M., 24. Aburazza, Soliman, "Ultimate Capacity

"Strength and Ductility of Tied Con- of Prestressed Columns," MS Thesis,crete Columns," Journal of the Struc- University of Denver, Colorado, Augusttural Division, Proceedings, ASCE, May 1974.

APPENDIX - NOTATION

Cm = member curvature coefficientd = total depth of cross section in

direction of bendingE = concrete modulus of elasticityEl = column flexural rigidityf,' = concrete compressive strengthI = gross section moment of inertiaL = column lengthP = applied load

P,, = axial load capacity without bendingr = least radius of gyration for column

cross section= load ratio contribution to rigidity

modification0 = slenderness contribution to rigid-

ity modificationA = factor used to adjust column rigid-

ity from gross section properties

METRIC (SI)CONVERSION FACTORS

tin. = 25.4 mm1ft =0.305m1 kip = 4.448 kN1 ksi = 6.895 MPa1 in.-kip = 0.113 kN m

NOTE: Discussion of this report is invited. Please submityour comments to PCI Headquarters by April 1, 1988.

PCI JOURNAUJuIy-August 1987 79

Documents

Tie Requirements for Prestressed Concrete Columns Journal/1987/July-Au… · ity. Ties and square spirals do provide some confinement for concrete at corner and central core locations;