ein pap Load. \cing Method for Design and Analysis of Prestressed Concrete Structures” 8, HORST, BERGER: JOHN FGROTCHIE, CS. HANDRASERHAR, C ‘VEERATAH, and KS. KAJACOPALAN: NORMAN B. GREEN: T KATOW. ROBERT L: KOONS and CERALD |. SCHLEGEL: FRITZ LEONHARDT: COIN, ROZVANY: KEITH. THORNTON: D. VANDEPITTE: Y. C. ‘YANG: and AUTHOR By HORST BERGER! ‘The idea of balancing a portion of the superimposed lod by the action of draped prestessing tendons is indeed an ingenious one which has found many interesting and useful applications Uroughout the last decade, ‘The author presents the load-balancing concept as a general design theory for prestresed concrete structures, demonstrating Its exireme Simplicty on several numerisl examples The main advantage is not only the simpliity of analysis of complex structures, but the fact that the loed-balancing method leads to a meaningful tendon layout and helps the visualization of prestressed action. ‘A numberof comments seem to be necessary. ‘The load-balancing method takes care of a portion of the load only For the rest of the load other design methods have to be applied. The load-balancing method, therefore, isnot a theory ranking with elastic ‘ress and ultimate strength theories, which give complete answers for Toad cates, It is however, a most helpful design method, comparable ‘with the concept of dividing sn asymmetrical load into symmetrical tnd ansiymmetrical load eases. ‘The author does not provide definite. criteria for the load-balancing ‘method. It is left tothe designer's judgment to decide how much of the Toad should be balanced, and what to-do with the rest of the load. He mentions that elastic stress and ultimate strenglh requirements need to be checked, And. in is numerical eramples he uses live loads which ‘conveniently produce stresses within the allowable limits of present ode requirements, If such requirements are tobe respected, and if no ‘ational criteria are given to Felate the amount of lod to be balanced fo these requirements, we are left with a teal and error method. How144 JOURNAL OF THE AMERICAN CONCRETE INSTITUTE Oxember 196 ever, such rational criteria ean be developed. For example, for contnd ous beams with uniform loads and a given allowable tension in the pie compressed tension zone, the loa to be balanced is “146 we Se o where With the help of Eq, (1) and similar ones for other eases the load balancing method leads directly tothe complete solution ofthe problems The loed-balancing method shares the disadvantage ofall specialized rethods of not being generally applicable. The lou-belancing, method does not apply when stright eables are used, when eables end ecoene trleally at free ends, when cables deviate from parebolie or trapesoial shapes, when spans’or leads vary considerably, ete. The designer hat to Keep this in mind and should not force his design into the straight jacket of balanced design, even when ite application is not meaningfl However, in almost all case the load-balancing method provides « mas Powerful tol for preliminary designs 1f applied properly, with knowledge of is advantages and limitations the load-balancing method is a most useful concept, ‘The author ig to be commended for his presentation By JOHN F. BROTCHIE! Professor Lin is to be congratulated for presenting an important new ‘concept in structural design. He has shown thatthe additional varlables Present in prestressed structures, namely, cable curvature and cable fension, need not be a compliating factor in desig, and can in fact lead Zero deflection might be considered to be produced by balencing the external loads a5 described, or by balancing the external moments pro Alced by these loads, that is, by dreping the cables go thatthe moment produced about the center of gravity ofthe section by the cable tension Jn just equal and opposite at every point to the moment produced Be Yond. Thus there is zero moment inthe conrete and therefore zero de 1 beam these coeepis are Identical, but in » threedimensonal stricture they are not In a pate strucire lad balancing allows an infinite number of solutions, but a unique solution may be blaine with moment balancing In shell srture ot all of the load need be alanced by the eae, since part may be balanced by the direct fore nthe concrete so that moment balancing may be more appliable here ‘Moment balancing ai allows solutions in which the cable is curved only in pla, for example in an axisymmetric! circular pate the fs nee be placed ony in concent crcl although this sof theoretical ‘The concept of moment balancing may be simpler in some case in that boundary condton suchas seo moment ata see eg ae rectly tise Profesor Lin's ertteron of sero deflection as sitios) significance in that it maybe ued als atone several eter forthe rect design i plate and hel structures This concep ie desribedelewhere?* ad tad ben teal bo Got foe expectant Mt Fst Teg Pig. Ar ‘The structure was constructed and loaded in Mare, 183 nd {fiction throughout bas remained st zero since1846 JOURNAL OF THE AMERICAN CONCRETE INSTITUTE Oman 1963 REFERENCE ASCE S16, Dee 192 RB ap se By C5. CHANDRASEKHAR, C. VEERAIA', and K. 5. RAJACOPALAN" Profesor Lin's paper is unique in is simplicity of exposition and the writers praise him highly. A beginner would welcome ths presentation mort, because of its lucidity and its capeity to male him understand the Behavior of a prestressed eonerete structure more easly. ‘The method expounded by the author, in fact, is oly an extension af the so-called method af “imaginary loads” or the fuleular curve method. ‘The limitations of the method have been unreservedly stated by the author himeelt, However, the writers have the following comments to after LOAD TO BE BALANCED ‘As explained by the author, the amount of toad to be balanced is crux of the problem, which requires of the deslger, proper judgmen tnd imagination. The factor ks of live load has to be predetermined before selecting cable profile and/or the prestressing force. ‘The load to be balanced can perhaps be estimated by consideration ‘governing permissible eresses or perisble deflections Jn an atiompt to arrive ata deflate criterion for the load to be bale anced the writers conclude th following When no tensile stresae are allowed at the bottom fiber of the ste tion under design lve load and since the stress are vniform uncer balanced load, the unbalanced load must be such thatthe streses ‘duced at the top and bottom fibers (for symmetrical sections) are nol more than th J", where "fs the permissible stress in concrete. ene the unbalanced load and consequently the oad to be balanced could be obtained. It is interesting to note that since the stress distribution al balanced load is uniform and of magnitude ft, and since the sres se tribution due to initia prestress could vary frvrn zero at the top tbe to f.t at the botiom fiber, the stress distribution due to the balan load should obviously be the same as the stress distribution due to 1 unbalanced load, Hence the foregoing arguments lead to the eoncus that only half the total dead and design lve load should be balanced From stiffness considerations also, the load to be balanced could obtained. Since the deflection under the balanced load is zero, the nitude of the unbslanced load can be obtained given the permisal eflectons, Total dead and design lve loads minus the above unbalanced vad fs obviously the load to be balanced. ‘Combining the above two considerations of strength and stiffness @ ‘most optimim depth to span ratio can be derived. It has been found by the weiters that for simply supported beams of symmetrical cross rion this ratio ls about 1 in 25 for # permissible deflection of 1/400 span assuming fis 2000 pa, and the modulus of elasticity of eonerete ‘cls 200 kp pee 5g in. "The foregoing arguments and conclusions hold good at long as the tive loads are fairly uniformly distabuted. Most of the structures, how- ver, could be generally classified under bulldings and bridges. In build- ings the live Ionde may be assumed to be faily uniformly disteibuted Gnd henee this load-balancing method of design and analysis would prove to be more advantageous However, in the case of bridge structures, the live load generally consists of concentrated loads varying in postion and magnitude and venice each section of the bridge allains a maximum moment for a dif ferent position of ive Iosd. So tis the maximum moment curve which ‘sore significant for design and in the ease of continuous structures Its the range of moment variation which governs the design. Thus for ‘hese eases it may not be posible to balance any portion of the live load tall The writers fee, therefore, that fo al such cases itis not advis: ‘ble to balance any portion of the live load. This is also jusifiable ‘ince the prestressed conerete structures “arey thir dead weight gratis” ind the dead load controls only the cable location and not the stresses Tn such circumstances, it is doubtful whether the losd-balancing method of design would be more economical, particularly, when the ‘maximum live load moment is grester than the maximum dead load moment (see arguments rating that the balanced load should be equsl ‘othe unbalanced load for symmetrical sections), In Fig. 6 on p. 724 of the author's paper, the cable profile does not conform with the span and loading shown, Unless L is greater than Ly ‘here would be & pont of inflection in the anchor span which requires ‘he eps to eross the eae. “The philosophy of the author's method ie similar to that of partial brestresting, Perhaps his method could be named as partially deflected Drestressed conrete structures’ analysis, The writers feel that the Luthor's method is a versatile tool in the analysis of prestressed conerete Staticaly indeterminate structures. The way in which the involved math- matics ofa fully concordant cable are done away with, is what makes ‘he author's approach most welcome. "The writers thank the author fr his valuabe contbution.1866 JOURNAL OF THE AMERICAN CONCRETE INSTITUTE By NORMAN B. GREEN* Professor Lin isto be commended for an excellent presentation the load-balancing method for the design of presressed concrete s tures. ‘The author has a talent for a clear and concise approach problems of structural analysis. ‘This method has been extensively used for the design oft prestresed flat plate floor and roof slabs and may in fact be consi ‘standard procedure for such designs” For school buildings of ‘buildings and apartment houses, the working prestress is usually Justed to balance the dead load, plus ¢ small fraction of the live hat may be considered as permanent, This fraction is designated, ‘Ky in the authors poper snd it will be about 1/10 forthe types of Pency mentioned above. Consder the common case ofa slab of unl thickness and a series of equal or nearly equal slab spans. Then a balance canbe attained fo the interior spans. If the prestress cables ‘continuous across the slab and maximums cable sags are used in spans for minimum prestress,it is necessary to provide cantilever of proper length around the perimeter of the floor or root, In to balance the loads inthe end spans. The cantilever slabs may be ei ‘ated and the end spans balanced, if these spans each have extra stress eables that are anchored atthe firt inferior support As an ll native design without cantlavers, the same cable sag may be used both the interior spans and the end spans, but the interior spans ‘then have less than the maximum sug.'This will require about a 0 third increase inthe prestress, orhich will not usually be economies the above described loading conditions therefore and ater all pre losses, the slab will remain practically fat, whichis very desirable the installation of floor eoverings, furnishings, equipment and partit For these types of oceupancy the live load required by the code, ‘it is reduced in accordance with the loaded ate, is much less than dead load. The concrete tensile stresses under the dead load, plus Ii oad, plus prestress ean then be limited to aero fa proper slab thick Ss used. I s the writer’ belief that this limitation fs advisable, a these slabs have a tendency to erack during the euring period and bel prestress is applied 1n the examples given inthe paper, the prestress cables have what writer has termed the “design profil." Professor Lin ealls attention the fact that, “Since the actual eable placement may not possess Sharp bent shown over the suppers, the effect of such deviation mt bbe investigated accordingly.” The writer has investigated the effect ‘his deviation and has published diagramn from which an approxima correction ean be made ‘A post-tensioned cable will actually have @ draped profile with In- ‘cetion points, and the error involved by use of the design profile will ‘depend on the position of these inflection points with relation to the Slab supports, the closer they are to the supporis the smaller will be ine error It is the writer's opinion that itis better to detail a pre- Stressed member with eable inflection points 20 elote to the supports that the resulting error is negligible, rather than to base the stress ‘aleulations on the actual draped profil. Apparently the oaly Litations ‘this regard are the minimum permissible radius to which the eable an be bent and the maximum permissible bearing pressure on the ‘onerete tender the humped partion of the enble. Even with these limltax tons, the Inflection point can probably be moved in to abput the 1/20, point of the span from the suppor, nese the span it very short, and {his will reduce the error to an acceptable value. In any case, the re= ‘lred postions of the eable inflection points should be clearly shown Gn the design drawings and nat lft to chance ‘Although the load-balancing method of design for flat plate slabs sives very good results and fas been widely uted, itis not as accurate fan analyse for this purpose as itis when applied to one-dimensional prestresting as for 4 beam. The usual procedre in the design of the plate is to employ the design cable profile and edjust the sag end pre- tress forces of the cables ia a column bay, so that the total upward force exerted by the sagging portions of these cables exerts an upward force on the concrete that is exactly equal to the dead or permanent Toad on the bay. The eables in the column strips are assigned more prestress than those in the middle stip. I is then assumed that under ‘dead lond the center of pressure is atthe center of the slab and there fare no bending moments” However, each cable also exerts a downward force of the sume magnitude ar the upward fore, where it humps up ‘long the column center lies and since there are no rigid supports on these ines, bending moments are produced in the slab in the general area of the ealumn strips, These downward forees are indicated in a [general way in Fig. 17 but the author does not discus thelr effect on an fetual slab design Such a discussion would be very interesting Te appears fo the writer that the good results secured by this method of design pplied to plates, is based on the fact that about 84 percent more prestress {s usually employed in the eslumn strips than in the middle strips, Obvi- ‘ously this prestress distribution is not consistent with the basic assump- tion, thatthe prestreas balances a uniformly distributed dead loa, since this would require « uniform prestress distribution across the sab. Tt ‘Seems apparent that the load-balancing method of design es applied to ‘at plates involves important empirical factors and it doesnot give exact stresses for one load condition, in accordance with the usual assumption.1880 JOURNAL OF THE AMERICAN CONCRETE INSTTUTE _Oecente REFERENCES 9, Rise, Edward K, and Kulka, Fells, “Design of Presrssed Lift Saba efit Contra." ACI Toten, Proceedings V8 No Feb, 198, bp OO 10. Green, Norman B, “Pico im Design and! Constucion. of Ltt ‘uilnga” ACT Jovtwa, Proceedings V6 No. Ape 12, pp 857-908 By T. KATOW* Tt would be interesting to attempt an application of the two-dimensi load-balancing method to the design of prestressed continuous labs. This might be done similar to the two examples by the au In Fig. B, wniformly distributed eables which are laid out in directions produce downward line loads along the column lines. downward line londs can be neutralized by the upward pressure large cables along the column lines. The net effect of all the cables therefore, a uniform pressure over the entie flor ate. This proc 4s general and can be readily applied whatever the span verlations slab edge conditions. As in the two-way slab, an infinity of eable layout fare possible for any given set of conditions, Since it isnot placed a5 closely as possible to the ealumn line. Practical considers Dreclude the concentration of cables to any great extent, making esigns theoretically quite unsatisfactory. However, ifthe column ca fare confined to within the “column strip,” this is at least as accuta nd considerably more logical than the presently used. beam th ‘method. It should be noted that the question of the distribution cables betoveen the “column” and “middle” strips (6535 percent, 75 Percent, ete.) does not arise here in thi load-balancing method, {tis actually not necessary to get a close approximation tothe ideals cable layout. Certainly, deflection control does not require any gral accuracy. However, approximating the idealistic layout as clovely ag ‘Possible makes the unbalanced portion ofthe load as uniform as possible ‘and thus, possibly justifying the use of the well-known approximations for moments ina uniformly loaded elastic fet sla, ‘Axial compressions in the slab could be approximated from some ‘manner of two-dimensional anchorage zone stress theory. Here agala, almost any intelligent guess would be at least as accurate asthe presente ly used beam theory method in whieh the usual assumption is that the ‘ial compression remains within the “column” or “middle” sicip in Which the cable anchorage eeurs ie inicirnoe koala ame eer ere pena eee Sle tha a fail goo tempt balancing the ded ond of the ab eee Sector eee Se cea cee See ee ee Sees eles ee eee Peg eye ee oes 2s ee a ae bslancing the given load by the net tanevesepreaures exerted by sae eee ee | Leowemw" canses Fig, BUniforly distribated cables132 JOURNALOF THE AMERICAN CONCRETE INSTITUTE. cab {for positive sign with respect to the coordinate system shown, Si ‘the Vertical forees we obtain wi) = eles wear nF the cas ofa straight conerete member, the corresponding equa we a7 where u/ Is the cable eurve for this case lace 384 (or— vd sttisty the same differential equation and boundary conditions, ‘must be equal. . © pawn a. (4) oF () must be aplnd in acordance with sg. The re cane contr that thie the aubor al of prover This type of proof may net be very convincing to thee unanl with fret equations and one might Ick In te fl Wray. Th p curve isthe slgebrle sum of cures yan yo The ye ton ofthe curve balance the anes peature eee by he cued toneet member andthe portion ote carve bln the given Tei supra fo note that the linear tasfrmaion fle pls tothe conerete member ot wel atthe cable Thi flows from Moin In Fig tt the atone ah the let oer In ering transverse press, the ane except that oe ie enon and the other is in. estes: pees “ Handling this problen by eonalering the eccentric af he exile component of the rete force i very cumbermme, The mead LOAD-ALANCING METHOD 183 ‘this discussion is much more logical, simple and revealing in physical action 11, Neer, Andrew R, and Jon, Cal B. “Semlgrapicl Analysis of Long attired Concrete Wasted Shall” ACI Sousa, Proceedings V. 82, Xo. 3, May 188, pp. 650-67 By ROBERT L. KOONS,” and GERALD J. SCHLEGEL’ Professor Lin isto be complimented on his concept of load balancing ravity loads by prestresing concrete. The writers prefer to consider the continuous structunes dicusted in Profesor Lin's paper as sean Chored able structures, Considering the structure as eable-supported provides a method of visualizing the reaction of the exble on the con= rete due fo the prestressing or tensioning of the stool. The uplift forces treated then es be tented lads acting opposite to the gravity Toads. ‘These loads can be converted to moments and shears and equated to the moments and shears caused by the gravity loads. "The paper is s0 broad in scope, itis dificult to cover many of the details of design, Somne of the details Uhat need tobe discussed are 1, The effet of reversing the tendon over the supports. 2. Allowable tensile streser in the concrete for continuous girders and the use of mld steel to accommodate these stresses, "The ultimate strength of a continuous girder. ‘The continuous beam illustrated in Example 8, Fig. 12, will be de- signed taking Into sccount the reversal of the tendon. As Professor Lin points out, the most economical design uses a maximum drape ‘The beam with the drape used by Professor Lin and reversed over the center support at O40 of the span is drawn in Fig. D. It can be seen thatthe area over the support has a uniform downward pressure while the concave upward section hs a uniform upward pressure. Two charts fre included to compute the fixed-end moments due to these combined Toads, Fig. Eis a chart for asymmetrical tendon; that is, one which has ‘the same total drape in each half span. Fig Fis chart for an unsym= Iettcal tendon, or a tendon which has different drapes in each half span, These charts (Fig. E and F) were developed by the weters!® 2°, of 040 was chosen to eliminate the problems of friction during the stressing operation, Onetenth is considered to be conservative. A dius of 280 im over the support would be quite adequate for 8 two- ‘pan beam stressed from both ends— : caer io eer 6s Fig, D_Contnout beam shoving cable rovers ‘he fxed-end moments for the frst span are iustrated below. The s span and system of rapes Is considered tobe unsymmetrical 60 aaa aa ‘The formulas for caleulting the moments on the chart are Symmetrical case b= 080 Pa) 5s ‘neymmeteicl case sete 8 — sah gy (RY aK 4 aK 80 = BE weak 3K) Mem yg Pgs 08129 — arg gy CE 2) a = BE cK aK From Fig. with Ky ~010 the fxed-end moments ase SE ae a0 ‘om 0008 Go 01s 0m om moe K Sort maon L=Point oF nEvEREAL oF conve oF Ten0on Saneh Ge acer x a4 Mote = ovens j= ‘The ixed-end moments ar balanced as follows P AS (eal ee KE Tiger por “ane ie iume | tas" mn © ‘vise | FEM MetMe = CPa ‘The M/P= 16.704P in.=< or “c" line, the effective postion of the tendon over the support: The “e line ofthe tendon is sketched in Fig. G ‘The "cline can also be called the true moment digram of the tendon. Fig, Char fr caleating ised and moment for asymmetrical tendon1056 JOURNAL OF THE AMERICAN CONCRETE INSTITUTE Omenbe LOAD-ALANCING METHOD 97 ‘The moment produced by 1.09 kips per ft atthe % point is 1805 f = 229 kips (see Fig. G), The stress at the points is obviously P/A since ‘or 2165 in-kips, Balancing the load could be accomplished by bal the moments have been balanced. Then the stress is 229/960 — 635 psi ‘the moments produced by the load provided a tendon can be placed ‘compression over the entire section. At the center support, the external provide equal and opposite moments at all points. However, this ‘moment from 1.08 kips per ft is $220 ftps or 3860 in.-kips. Here the Impossible as demonstrated in Fig. G. The load can be balanced ‘restress moment i 229 (16.70) = 3620 in-kips The resultant stress from proximately by producing an equivalent moment with the pres ‘the tnbalanced moment is = 22 ps, tension on top and compression on force, at points of maximum moment. The force in this case is 2168 the bottom, Agsin there is an axial force of 229 kips. The combined ef fect ofall the stresses at this point is 618 pei top and 687 pal bottom all f°. compression. The results at the two points investigated indicate the Stresses to be nearly uniform across the entire section. This will not be ‘necessarily true at all point in the span since the tendon doesnot follow S the gravity load-moment curve precisely. However, the beam will be ‘nearly level for the loading condition considered, ‘The additional load of 057 kips per ft causes addtional streses of + 068 plat 3 L. 0 20) —_—2t ey a a z A | = 7 25 Fig, @—Connae bam ted fr ayy a2 as sig ge tendan 4 ane 3 gers taicaer 24500 amr x = i Stet pend $6005" E aC P, «Poe Mya P9) +6, Pag a Fig. FChart fr calculating sedand moment fr an ansyrmatrical tendon Fig, H—Stren volume cleltion8S8 JOURNAL OF THE AMERICAN CONCRETE INSTIUTE yaceres Secor ADA O05FE dag + (58000, Solve fra chi 16 00 ne TA SS ee TB 00 eateed) Fig. |Diagram fr fining ultimate moment of lad ‘The final stresses in the member under total load are: sm — 4ats.2 45 = { + 8s comprion op ins } Epa tetond comet 48h | ap en) 1p soe stems Wh dng nad on the member vain win tosis aed omg empties indent Sanaa ae ne center support has a conaldersly larger tenale Suess The new AGE piven ceases ote IEG) gems lore ei Sew 6 ret conditions, The writers feel it is desirable to ac ital when {Else stove 20 pl when ante cond ott a an ‘beams Its tlt however thatthe tele stems In flat sabe sould be ite to at whe tense acne el ssresng steel should be added The ultimate strength check might Tera tate grea meta otters {IM se lb red compen rhe eon seen a iors hare lean Smee i PB ‘loess of 300 parted tempus te eu a txt nd bn ann iis a inte rng heck mae by wg te | Toe ltnaiemcng mimes ton Rg tig omnes (4-4) +7, 1E the ultimate moment of the | siting moment as determined bythe meth the method in i sale Seme features of Fig. Tneels dacaton, enn Fit he 4, ofthe pests tee the A ofthe pret ele deemined by ding the“ balancing force by ie llowabe working rea, pe we of LOAD BALANCING METHOD "87 ‘Second, the f, ofthe mild steel can be chosen to equal that avaliable i {he designer's area, The waiters ws an f, of 162,000 psi for coated ten- ons since the shim-olf stress is generally 168,000 psi. They believe thet the wire will develop atleast this amount of stress near failure. Third, {he moment arm of ihe tendons isthe “e" line of the tendon, not the ‘hyaeal position of the tendon. The “line or effective position of the Bites force is independent of the value ofthe prestress force or the ‘eter loads For any aystem of drapes in a continuous beam, there i Ship one "cline Thee" line does et change its postion even though the eonerete may have cracked. "The ultimate moment of the load of the support is 1.8 (Bigo) = 10,810 mips ad of the A point is 18 (3360) ~ 6048 in-kips. The ultimate fealeing moment with the steel calculated using the criteria established hove was insufiient, Therefore, the mild steel was increased to 25 fq in at the support and 03 sq in. atthe % point. “The writers feel the methods discussed herein properly demonstrate 1a design approach taking into account the reversal of tendons. REFERENCE 12. Koons, Re, and Sehege, G. "A Practial Approach 10 the, Desi of coe tia in Pretsesed Coerei,” Jounal Presireared Concrete State, V8 No. Au 1063 po. 8-58. By FRITZ LEONHARDT* In his paper Profesor Lin desing with interesting aspects of pre~ stressed eonerete, expecally satcally indeterminate structures. Much ev poon written in the past 18 years about the analysis of such statically indeterminate structures, especially in Burope, end ome confusion ha ‘Poon created by introducing new terms lke “parasite moments” or “con- estency (ho Testraint forces due to prestressing) and similar concepts ‘The wrer designed in 1948 the first continuous ginder ridge of pre ‘Treated concrete ever three spans without considering concordancy and ‘Smputing the restraint reactions due to prestressing by the classic old Secthods Meanwhile many thousands of statically indeterminate strc Tents have bean designed and. successfully built In Europe with pre- ‘Stressed sonerte. One has found, thatthe old methods of statical analysis SE lly mulfcient and that-concordancy is no requirement of any practical importance ‘Designers alo found very ety, that in statically indeterminate struc: tures he tendons must be curved of bent and should have a curvature ‘Mar to the curvature of the moment line due to the loads, whieh pre~ ‘itucng must counteract In the ealeation of the bending moments"M40 JOURNAL OF THE AMERICAN CONCRETE METITUTE ene due to th pressing forces of our et res of ou first stately indeterminate Lous ides we already aed the transverse free tee sf rection ofthe pestresing tendon in tion fo the norma Acting atthe ends ofthe structure and coniered then oe a {ores lke ledings We ised the influence ines for bending or shea free fnd those vals due fo presireing and ie thence of th rate Indteminacy was need witht Aitcuiea Th infuence nes are needed snow for mavng ie a ae avaiable ay deen handboks:Thaeto lin it seven the only simple solution to elelate prentened pistes Gly plates for skew bridges. It farther proved fo be meee a the tse of dig computers, when continuous beam of ames of sige Jar sructues have tbe cleated for presen rin now ug he ong mete, Bch ta snaveae component of the pressing force wih tc Dar of the wertng om tut the scare ed ea Sresey for a sort tine te this a practical approneh forthe deg Of pressed conereesirtures? Ar roe in shows i Pag Ye load balancing gives certain point ofthe oadidefletion linn he Point where there is no deflection after prestressing. i 1 the presiresing force is choen to fit ths pein, then other more Important requirements will in moe ass ot be filled, oF chee words thie load-balancing method normaly det aot let the fal Zolsion forthe neceary or ecnonical amount of prstresing fon, Profesor Lin himel ys on . 74, thatthe enginer must heey ind the sstacton of la srs limon eck snl en mate tength Fr lsance if enly «small part of ve ods tnened Ad the specified dsin liv load is much hier, ten the lod beleriog nethod give ato small prereming force for sisting the ened Safety agaist the total lad Hf, however, the ultimate loed conden is satnfid, then the pint of no deflection or the balsned loning hag ‘moved to a higher portion ofthe ive lad. Undoubedy, it Is more important to satiny the ultimate strength condition and tres limitations ike for insane no tenon Inte eae reson zone under pari dead lod or no high tension Inthe ton Zone under the fll eign fu, than to ave 8 belance between the ranaverse component ofthe prestening force and cia of the working load. se thy = ace The sim, which Proesor Lin obviously undesirable camber or deletion under nora in mind is to avoid conditions of a struc: W EMETESAMGNS argon Pete! couse Atma, eng ture, expecially under sustained loadings. This aim is reached also if there is small eamber or amall dfletion with low working streses, beesuse the deflections of prestressed concrete structures are very small 5 long as the bending stress are not high, like they are for instance in pretencioned girders which have been designed for a large but rare live load and for no tension under dead load plus live load ‘We have found, that this last condition, no tension under dead load plus lve loa, the so-called full prestressing, can be the cause of un- esirable deformations of prestresed structures If this condition is ful- filled fora structure which has to carry high le loads only oeasionlly, then the fension chord of the structure must be put under high com pression by a great presttesing foe and excessive camber, especialy fue fo ereep must be the consequence. In some cases, like beams for Tmovable crenes, this camber enused unserviceability. We therefore ound, that in such cases it isnot important to have no tension under the full design losd, and we allow tension because even eracks will not affect the serviceability, i they appear only under short-time extreme Tadings and are closed agrin by the eampression due tothe prestresing lunder the long-time dead load condition. In Germany, we therefore Introduced the so-called limited prestressing, a reduced degree of pre= firessing, for which a certain amount of tension is allowed under the fall design load, if this design losd does not occur too often. This Limited prestressing Is not allowed for ralroad bridges, for example, where ah excessive camber due to dead losd must be avoided by using ‘wide tension chords, a8 they are given by hollow box girders, by which fxtreme high compression stresses under sustained load conditions can be avoided. "Profesor Lin says, that “the engineer should exerise his judgment ‘when choosing the proper amount of loading to be balanced by pre- Gtrening” (p. 140). However, with this choice, usually other critical Conditions such as safety against ultimate load are not fulfilled and we trust se clearly, that this single point of no deflection ean not be made ‘criterion for design and itis not left to the engineer to choose this point, The alm of Profesor Lin can be better understood, ifthe degree St prestessing is ot fixed by the condition of no tension under dead {oad plus live loed. The degree of prestresing should be kept variable tnd dependent on the kind of usage of the structure, If fall ive load wwil oceur only rarely, then tension can be alowed and the prestressing Teel ean be supplemented by mild stel for getting erack control and the safety against ultimate load, ‘Further, the statement by Professor Lin, that if sustained loading Ss balanced, "the beam will femain level regardiss of the modulus of Clasteity ofthe flexural creep of concrote™ (p. 128 and 70) isnot cor- tect, beeause the prestessing foree ls not constant but decreases withshrinkage and ereep of the eonerete and therefore the transverse Ponent of the prestressing force is also variable, ‘The loss of Streming force due to these influences can be as high as 20 ithigh concrete stresses occur early, and if for instance, steel bare Wik medium strength are used. However, this influence will not be important. Generally, the engineer does not have the choice between the th fencepts, which Professor Lin describes at the beginning of his pa ‘The first two concepts must be used for any prestressed atructors check stresses and the safety and the third concept fs very useful If ‘is not confined to this one point of load balancing, but I iis ms {generally used asa method of analysis with the transverse componets of the prestressing force as it has been publied by the author In his book ‘Spannbeton fir die Praste and elsewhere long ago By ©. LN. ROZVANY! -Frofewor Lins new concept i one ofthe most important developments 4m prestressed concrete i ‘The writer has invertignted several sspects of the load-balancing technique since 1961. In this dlscusion, some of the proposed refinements are outlined. ‘TWO DIMENSIONAL LOAD BALANCING ‘The author states that n th case ofa two-way slab, “the most economs eal design fs to carry the load only inthe short direction (orto corty 850 in each direction in the ease of a square panel).” Professor Lin ‘Also mentions that the slab will not be level under uniform load if more ables are located along the middle stripe than along the edges. The writer has shown that, considering a uniformly loaded slab, varying tendon profiles and spacing ean result in: 1. Zero defection troshoat he sab 2 AA mare economical sltion than the one propre bythe author ‘The optimum design for minimum cable volume has also been dee termined LINEAR AND EXACT THEORIES Profesor Lin does not sate the general equation of load belancing explicitly. However, his examples seem to be based on the assumption, ‘that the resultant shear and bending moment are zero when 2 a eh LonpaaLancine Merton tenon force any i te ta which (2) he external nding ithe tn 7 ‘lau been the centroid of the meets mdm (0) ed th Serio he endoe ) G20) 1 wll be noted it: ee Tbe dr nt ge terns matheate ton PER ES Ss Seow ne Mest pacing engineers nthe would ko know 2, chnique for coretng the eur n case such as deper beams an Santi pane {olde pats peste In this discussion, the exact theory is compared with the simplified ‘method (B4, (6) ] which isrlerred to as ‘Iinear theory: EXACT EQUATIONS Bq. (6) would give the rigorous solution ifthe forces transmitted from 3 fendons (ote sneete beam were verti 7 ‘*Eilering ade suction ofan unbondd tendon (Fig. 0), the mag tude of this force: Bae OF Ee ob o elite aaah cl sf the friction is neglected. "The vertical eomponent win * 2 snd the horizontal component: war” YI eB Tar Fig. KEqulibrium of tendon element1864 JOURNAL OF THE AMERICAN CONCRETE INSTITUTE Section At ss Fig. LPratreied deep cantilever beam {Hf he Fomultant bending moment of the internal and external forces wd fasting oe iene ein ale eee ee ei Jn cates where the fron canon be nese the folowing equ (fase ne} Uf“tempee - [SECGREE) = sre cerante on J which F, and F, are the tendon force atthe end of the beam and at an arbitrary section respectively, i the coefficient of fiction betreos the tendon and the duct In the above equation, the wot oce ‘equation, the wobble effect Is ‘CANTILEVER BEAM WITH UNIFORMLY DISTRIBUTED LOAD Conser a dep catoer beam with «unary ribed lend w (Fig 1), af nee ‘The bending moment due to external loading: on According tothe linear theory (Ea. (8)], the “balanced load” tendon a vor a produces (1) zero shear and (2) zero bending moment throughout 1 , ng moment throughout the ERROR OF THE LINEAR THEORY ‘The exact balanced moment can be calculated from Bq, (10) and (18): 0g a) = f foretamre t— freee Alter integration, Bg, (1) reduces to ie ae Fig. M gives the balanced bending moments according to the linear ‘theory (Ba, (12)] and the exact method (Bq. (19)] if = 10. The error of the lines theory Is also shown for various depth/Tength (4/1) ratios. The shear force balanced by the same tendon (Eq. (13)] capcans wo Ate nogto, 2, bcos "Te a Then hy ges ned hfs Th magi gee eee ee ee Sites rng V pee cue in cs T T 7A 1 nt Me z 2 4 Of \+ x ; ne Fig. Mending moments balanced by parabalic tendon1866 JOURNAL OF THE AMERICAN CONCRETE INSTITUTE Osan TRY i Fs. foros balanced by par. Fig. O—Di Sr SachSie vaio Sp Orta otal Belaneed shart 8) balanced ton Fig. N shows the shear forces balanced by a parabolic tendon, LUnear and nonlinear equations are used, The later gives different vel for zero shear (A) and zero moment (B).. After differentiating Eq. (17) and (18), the loadings for balan shear and for balanced moment are: op oo respectively, Tho linear theory gives we, = w (see Fig. 0). EXACT THEORY, BALANCED SHEAR For zero shear, the tendon curve ha ost the following equation foreiore # == a After ditfereniation, Ba. (21) becomes ror = Hience, the “no shear” curve is: en co LoADALANCING METHOD 1867 Fig. PNeshear tendon profiles (A. ‘and boning momarts beled iby telinene nethesr profile (C) ‘The bending moment balanced by this tendon ean be obtained by eal- culating y’ and y" from Eq. (28) and substituting these in Eq (10): eo ‘For the given boundary conditions, ie uso he integration of Bq, (28) gives eo wewes fE * tothe near (Bq 1) Fig. @) ives the “noahee” tendon acorn tthe in and ext (Bq) thon andthe being manta blaneed by the later (Ba. (25)1868 JOURNAL OF THE AMERICAN CONCRETE INSTITUTE Decor LONE S Mees! it ‘The first and second diferential of Bg (25) and represent the shear force and the external loading, for which the shear tendon (Eq, (24)} gives zero moment (Fig. Q, B). EXACT THEORY, BALANCED MOMENT eet from Ba (11) that, onsiering uniformly Tedd Eh ever, the rent men ie 1 aie EC [fee eas [eres (ffrestome st - [arbre #) = 9 b 1 i ¥ 7 After integration for the % boundary conditions, Bq. (2) me] Fs-SSpeci pon of ol nbaning tendon cave (A and ies ee T=" = a 4 2. y(z) is only slightly greater than The wren sauna ht Ey woul sve angus statin at Setar conan yo ee = and ted several for ana vale of Sompitr etd Ce) To i upri,thene 2. Along ie curvet Ions ov eonting re _ After plotting the gradients wire SO ee ‘the slope is maximum. (Points of contraflexure, see Fig. S, Point P.) 4. Beyond ine 27 onto (2) Yeo main. (See Pe Punt) 4 2 a ie '5, Along curve 2*/2, the slope and curvature are zero and infinite Fa, RDite lng alned Yb reece. bite os pf (4) tnd (See Pig 8, Punt“ shear; (8) balanced memant “This method was waggeted by Canny “The exitence of such a curve war found by L. J. Gleeson and she writer lndependentiy;1970 JOURNAL OF THE AMERICAN CONCRETE INSTITUTE Dae LOAD-ALANCING METHOD (CONCLUSIONS LINEAR THEORY VERSUS EXACT THEORY Fig. M,N, and O indeate tht 1. The presently used linear theory gives quite accurate shallow beams (D/L < 1/20 and D/L <'1/40 for cantilevers and ‘supported beams respectively) 2. The error becomes more significant in the case of beams of epi 3, Por very deep beams the error ie extremely large. (Fig. Of D/L = 1/2, B = 1088 percent) However, the last cate is of no practical importance, because straight line theory does not apply to beams of these proportions therefore the lond-balancing method would not be justified in its form. It willbe noted thatthe exact “no-shese” tendon eurve (Fig, P(B ‘very ineficent for moment balancing (Fig. P(C), Q(B), and RB) For the ero bending eriterion, the exact theory gives an infinite ber of tendon curves which are quite differen in shape from the solution (Fig 8) Fig, T~Contnsus boa of variable ros selon ts loation to "The first problem isto determine a presress force and its Ica ust balance the dea load ofthe beam and portion ofthe ive lad Consider a continuous beam of varinbe eros section under the action ofits own weight ws, 8 superimposed load w, with a prestress force F “han spre ino so prs (ne Pig. U and), the ning cifect and the prestres effect. ‘Assume the prestressing foree F is located so that its effect (7) acting up equals the loading effect (37+ ws) aeting down at all points slong the beam. i fe eet ferro e, rea fava fran Fig UContnvout bea under sl aight and superimpoted led fae wrap oun 70 nevsyness allanee allnece Pallarcee Fig. Continuous beam under prestes: —— ACKNOWLEDGMENTS ‘he author wishes to thank HL Laslo and Te J. Gleeson, mathematics ‘Monash Univerty, Austral, for dei ceoperaion and vale, snd Miss J." Wittmer for pltting the computer agus REFERENCE 12, Rosary, G. IN. and Hampson, A.J. K, “Optimum Design of Pre Plate” ACT ovina, Proceeding VW, o. 8] AUE 1865p. 108-1068 By KEITH C. THORNTON* ‘The load-balancing concept as put forth by the author isa simplification of the design of prestressed indeterminate sutures renders the analysis of such structures to nothing more than the sates for the balanced portion of the load. The total stresses as the author points out can often be greatly spproxima ‘The waiter has used this concept now for several years and Uke to illustrate to what extent the concept can be applied in the Signof some Indeterminate structures. First, a general development of the concept will be presented. It be followed by several numerical examples. The symbols used ia discussion are the same assed by the author. = Zp. Lm and Aci, Cniing sts gees, Mew Yok Fig. WConfinuoubaam under eambined loading and presto872 JOURNAL OF THE AMERICAN CONCRETE INSTITUTE LOADSALANEING METHOD v7 oe % Ss 4 Fig. Yeading diagram Note that ifthe e at the two supports were different (e, e) the omen loading diagram support reactions would need to be corrested by Combining the loading and the prestress effect, Rp (F, €) Rs (Wo, ws) at the supports and st»4 (,e) eaneels ws and oy the beam, leaving the beam unloaded along its spans (ste Fis, W), From the above the following observations ean be made for a ‘any cross section when the prestress ie designed to offs! the 1. The verti! component othe pretreaing fore takes the shear at Pt along the bem (20 at sy pl Rama + Bane 0 mets Eee (CS PLACEMENT BY 3 2 The pri af the pressing foe ete same shape asthe ‘As contro pan lotding moment dag 15 3, The reaction a each support equals the prersing force tines te BB son 5 «) = 1012 12— no 1467 — 0208 10 — 0088 x gular change over the suport (etl! component the fore) ‘The reaction ean, thertor be sited from one supper by rotation or liner trentormaton of the pestesing pros ‘Tue beam has 20 deletion or camber any pose 5 The sre at any pot i the bam te F/d 8, he beam is statcaly determinate snd the ctl prestzes prfle be pled ty taking 2M at eierent points ang te Beco ‘While the above concept seems simple, it application to special often demands ingenuity. The writer would like to usteete with Sat ao A Tao flanger Right 354 nck, ‘ctual eases: a continuous beam with variable eros section and & mamte Story rigid fame, co = BA — ropa 128i wy ‘CONTINUOUS BEAM WITH VARIABLE CROSS SECTION Left ‘The following example (see Fig. X end ¥), s taken from the Pay eta ase tak wena sctual design calculations for an interior éontinuous slab apes pve aaneAds Parking gnrsge permet) ‘Balance sab welght 45 10 pr sat eu = BBE. 306-1122 wp Required mint clarence of top in bottom 1% To control cracking, an average prestress of 335 pi is chosen the Sin. slab, Therefore, F per fof slab ia 389% § x. 12-— 2,000 I f= hips per te ‘The maximum eccentricity atthe supports is taken as ‘Asfar (516 in from lft supper!) 20 (306 4 ¢) = 10625 8 — 0026 x 30 — 0208 42-0008 12 00675 5 88°36 = 44 inkVe74 JOURNAL OF THE AMERICAN CONCRETE INSTITUTE LOADSALANCING METHOD ws ee ‘As a check, the gs ‘parabolic inthe uniform Sn, and the prestressng force s2g must equal the simple moment (Fa = wLi/8) ‘The sag « inthe 145-6 span 0m 4 20120, = Tin, (use Tn) Chee eight = 1728 59, (a 10% in ‘This isan approximate ur Was taken a the center ‘mal span and not the center 5 in. slab span. However, the flat profile in thi region, i is close enough. foam dein Fig. 2_Malistry frame pas! ‘Therefor, no prestresing is roqulred. [MULTISTORY FRAME BALANCED FOR LATERAL LOAD ‘The following calculations shown only describe the design of to balance lateral foree of 78 kipe applied at the top of a frame: Fig. ¥). Efforts to balance the vertical loads, such ae applied to top girder, are not described herein, Since the magnitude and location of prestres is designed to just. ance the 78 kips horizontal thrust, the structure is not under and is subject to only axial stress under the T6-kip load. Beams CD and EF Four 1% in. diameter 160 k post-tonsioning rods were used at 1125 kips per rod, P = 4 1425 = 570 kips AM capo PHM x on sn, By D. VANDEPITTE® Profesor Lin ie to be congratulated for his paper. He has made a ‘lear and gystematie presentation ofa very useful idea. ‘The same fundamental concept is present in embryonic form in @ paper published in 1852" ‘The writer uses the load-balancing method forthe design and analysis of prestressed concrete suspension bridges. Its use is almost obvious for ‘hat type of bridge In a further paper (1 +a) stands for the intial ‘alue ofthe upward acting force exerted by the cable on the stiffening (C65 PLACEMENT BY CColann deign (beth eslamn aa) ‘To control eracking under live load and keep eccentricities twenty-four % in. dlameter, Towire 270 k pretensioning strands ‘sed a 25 ips per strand, Then Pax 25 = 000 eps Section AC (es ei1976 JOURNAL OF THE AMERICAN CONCRETE INSTIUTE here ‘beam, and (14-1) w represents its final value, w betng the Hence the dead load is overbalanced, and the excess upward. ‘aw, which gradually decreases toward its ultimate value st, is to balance part of the live load. It is also the excess upward fore Droduces creep flexure of the stiffening beam. The lots of prt force resulting in the decrease of «toward « and due to creep and causes i studied in this paper. Reference 15 contains a specific applletion ofthe load-balancing od in one field. However, Professor Lin isthe first to sate the ln general terms and to show the wide range of problems tha 3 = | ot calle profile wih te Fig: AA-Simple parabolic cable Fig. BB—Actal cable profil ‘The cable at the support and midspan exert a uniform downward force w, and upward foree w,, equal to respectively handled efficiently by this method, a wu 1 den,“ oon tw hesnien ds No aaa “Anat dea Trova Public de Bue (Breas, Feb SE pe ‘he fxed-end moment a the apport de the ation of wy and w. ean 15, Vandpiti, D, ‘Prestressed Coneete Surpensicn Bridge ad the of Tine Biles ont Mapnitade of the Presming Free Tunlstions Monthly Reviews, Indian Roads Conte, Dee 8, pp. To, ByY.c.YANG* ‘The author’ load-balancing method is indeed an important con tion toward the design of prestressed continuous members. Ta the cation of this method variations and refinements are often needed writer wishes to bring out two items a follows be computed tobe rm = 22 a While i we use the author's assumption of one simple parabola, the fixed-end moment a the support will be ran = 27 ‘he tterence between the to methods CABLE CURVATURE ‘he author's paper gnerally assumes that the cable follows a parabolic curve Fig. AA. However, du tothe stifinens of» cable, Actual profile actually may fellow a reversed curve as shown in Pg Referring to Fig. AA and BB, le: iol dap na pia interior pan ae Ep Sa mew oc con PEGG! distance trom tho point of antafecure to tho cable Assuming that the curve at the support isa simple parabola downward and the curve at midspan is a simple parabola “upward, then Fn — FEM and the percentage deviation is (100 percent) (a) Fa co Prteuing es Fig. DD — Cable profile {h8eing manmar deviation Fig. EECable profile at rd spon cS cen Spa ts bet ger. na ee‘By the same reasoning it can be shown thatthe percentage devi at midspan is (200 pereen) (a). Evidently the deviation approsches 2ero as a approaches zero the maximum deviation is $0 percent when @—03, in which case cable will be as shown in Fig. DD. The veriter har found that the usual practice the value of @ ies between 0 and O15, hence error in flxed-end moment may be 10 10 15 percent if a simple pa ‘was assumed. However, due to the effect of linear transformation, ‘resulting effects onthe stresses are not quite 0 serious to be arrived at by balancing a predetermined portion of the load. The ‘thor would ike fo go one step further and suggest thatthe allowat ‘tresses should be varied depending on load-bslancing requirements, jit at they can be exceeded when Using ultimate strength method of design De. Brotchie's suguesion that we could often balance the external moments instead of the extemal loads is an important statement. The cuthor has used this approach and considers it especially convenient {or beams of variable section, which Is ala explained by Me. Katow for ‘urved members, Applying moment balancing to plates and slabs indo- Dendently forthe to directions will yleld a Unique solution, as pointed tout by De, Brotehle, but it should be noted that this may not be an ‘ptimim design, since other arrangements may be more economical. 1m some shell structures it is posible to balance the membrane siresses, instead of trying to balance the loads. But load balancing is the easiest ‘or the boginner, while eventually we may prefer to apply moment valancing and direct stress balancing for certain cases. The writer isto be congratiated for using this method as a direct design for his experi- rental plate. It is indeed the design feature, more than the analysis, hat males this method so sel ‘Mrs. Chandrasekhar etal have a good suggestion that the load to ve balanced could be estimated by stress or deflection considerations ‘Their value of If", however, is correct only when the ege of the setion s located at is middepth. Thelr attempt to obtain an optimum depth- ‘-span ratio is another good example of how the load-balancing method ‘cam be ulized for optimum desig. The waters are corzect that a point ff inflection should have been shown in Fig 6 on p. 24 DM. Green's diseussion deals essentially with prestressed lift slabs vehowe design criteria, using the beam method, were initiated by the suthor for epproximate design of regular spans" These erieria are ‘meant to be approximate; unfortunstely, some engineers who apply ‘hese erteria tend to regard them as perfect. Becaute of the nature of flat slabs on isolated columns, it is imposible to totally balance the ‘oad, since we eanaot bring all the maln eables directly over the columns In mart cases, we would end up with downward loads arcund the periphery of the columns and upward loads elsewhere. The degree af Sowaraey of the beam method will depend on many factors, and ean be ‘deenmined ifthe base principle of load balancing is applied, The author Sind is colleagues have uted the finite difference method? as a more ract method of two-dimensional Iosa balancing. Slabs designed by the beam method, particularly those with Tong cantilevers, cannot be ex pected to remain perfectly fat, as ean be easily proved by the basic END SPAN When applying the load balancing method to an end span, one realize that continuity terminates atthe end euppert, and its the practice to locate the end of the eable atthe ege ss shown in Fig. BE. It wo push the cable up by an amount of ad at the end, the rape is Increased. The gain in moment at the fist interior support. equal to Pha rsa A the same time thee is introduced a erancein moment at the equal to Fad. This moment when earried over tothe firs inevioe ‘ort, is approximately equal to Fd/2 ifthe carry-over factor 1s ‘Thus the net effect on the moment atthe frst interlor support De ero, By the same reasoning there may not be muich change in momnent the frst interior support if we locate the cable below the cge at the: support. However, shifting the eys at the end up and down will result ‘considerable change in streier in the end span and roy be cl but rather effectively. AUTHOR'S CLOSURE ‘The author is grateful to all the writers since thei discussions ha conteibuted much toward the concep expounded inthe peer. Me. Berger, while agrsing to the advantages of the method, ment that it is more limited when compared to the sare tessa ul strength methods which give complete stswers for all load eases author wishes to point out thatthe cle sree method. does apply much beyond eracking while the lina strength method Plies only atthe fatlre losd. Hence, none of the three methods Complete in tact, and each wil frequently need the supplementation the others. Se = Mr. Bergers equation relating the amount of balanced load to stress roqulrements is woeful enterion, which enables own sieRe eee sence sean at Biceiete seat ae ee ee eae pee See eee ais a are eee ae cee ree eases eis tee ave a ee ee pee oe ee Spee eee eee ee oe Sem ee a eee See ooo eee efficiently by means of it.” ms i zs a foe oy aa =e ere Deere Soe ee eed eee Sei a peer eee ee ee re Sena ee ee ac ne ee robs Gil Ti si ata Gre on the shape of the curve and the magnitude and significance of the ‘espective requirements In fact, Dr. Leonhardt inadvertently admits {his when he advocates the use of variable allowable stresses fo fit the ‘eqirements of deflection and eamber. ‘Dr. Leonhardt believes that “this load-balancing method normally doesnot lead othe final solution forthe necessary or economlesl amount ‘of prestressng.” This is ofcourse, a matter of personal eonviction and Skill Having started with the allowable stress and ultimate strength Inethods of desig, the author and his colleagues have now shifted over {o the load-balancing method forthe majority of statically indeterminate ‘designs, simply for the reaton that It does lead to final designs much faster and more directly. ‘The author wrote on p, 728: “Ifthe effective prestress balances the sustained loading, the beara will remain level regardless of the modulus tt elasticity or the flexural ereep of conerete”* Dr. Leonhardt quoted nly the second half ofthe above and eritilzed it as being incorret in neglecting the lose of prestre. The author used “flexural creep” de- berately to diferentiste i fom “axial creep” “The author concurs with Mr, Rozvany that a more rigorous mathe- matical solution is preferred when the depth-span ratio is unusually high. The author also wishes to eall the readers! attention to the in- teresting optimum design explained inthe paper by Rorvany and Hamp- ‘som in the August 1985 ACI JOURNAL ‘Messrs. Koons and Schlegel and Mr. Yang independently studied the ctfects of reversed curvature in a supposedly parsbolic tendon, The ‘curves and formulas developed for estimating these effets are very tuseful for beams, For similar curvature deviations in a flat slab, a finite difference method to estimate their effect might be efficiently spplied in a computer solution of the problem. ‘Mz. Thornton's multistory frame balanced for lateral loads is indeed ‘an interesting example. It indicates the transmission of a horizontal oad down through precast frame without producing bending in its ‘members The writer should be congratulated that he was able 10 make fiich an arrangement, not only a theoretically advantageous one, but flo practically feasible scheme, “The author wishes to mention In conclusion that while the concept of fond balancing has been established, its application is only beginning. Further refinements and extensions are being developed, expecially for SRatically indeterminate frames considering the effect of exeep and Shrinkage, for fat slabs of ll shapes, and thin shells of many types.