You are on page 1of 14
ASI STRUCTURAL JOURNA TECHNICAL PAPI Title no, 84-835 Effective Length and Rigidity of Columns by Bekele Mekonnen ‘Simple approximate equations are presented for determining effec: tive length and rigidity of columns for use n stability analysis and ‘rift calculations of frames, including effet of cladding. The range (of application of curent design aids and the proposed equations is determined. Outside that range, the Grinter substitute frame is rec ‘ommended and a simple, exact equation given for caleulating its de ‘Pections. Stailiy under the ation of eccentric lateral loads is treated sing the storystffness approach. Tables of effective length factor ‘or braced and sivay frames are sven concrete igi: stably sis; strut amass " Many design standards, including the ACI Building Code,' recommend comprehensive second-order analy- sis for consideration of slenderness effects in compres- sion members. However, in the interest of simplicity and speed in routine design, alternative approximate methods are also accepted by building codes, and nat- urally preferred by designers. The most important of these methods is probably the moment-magnifier method, which now has as wide acceptance in concrete design as in steel where it had its first application, iculation of the magnifier requires either the ef- fective length or the lateral stiffness of columns, which are difficult to determine for all but idealized isolated columns. For multistory columns, current practice is 10 use a conventional local model of one column and its restraining beams, on the basis of which design aids like the Jackson and Moreland alignment charts have been prepared In this paper, both this conventional model as well as the more powerful Grinter substitute beam-column frame model are reexamined and compared. It will be apparent that current design aids, like the above-men- tioned alignment charts, should have limits set on the range of their applicability. Outside that range, the Grinter substitute frame may be used with considerable improvement in accuracy. Finally, the concept of story rigidity is applied to calculation of the moment magni- fier, in the case of eccentric lateral loads. New approx- imate equations are derived for determining both effec- tive length and rigidity of columns using the conven- 316 tional model. A simple and exact equation is given for calculating column deflection in multistory frames, thus facilitating the use of the Grinter substitute frame. ‘The effect of partial bracing on column deflection and stability is also considered in a simple and direct ‘manner. ELASTICALLY RESTRAINED COLUMN WITH EQUAL END RESTRAINT The stability of elastically restrained columns with equal end restraint may be studied using the symmetri cal single-story frame model of Fig. 1(a). Assuming the possibility of a bent configuration un: der the action of vertical load only, each column of Fig. (Ja) will deflect in a sine curve (Fig. 1(b)]. The result- ing bending moments will, therefore, produce a sinu- soidal curvature load on the conjugate beam of Fig. 1(@), defined by Moh EI, sina w where = modulus of elasticity height of column moment of inertia of column effective length factor bending moment at joints /2k = unit curvature load at Point x, measured from the origin at midheight of column. SR ERN Sm Received May 30, 1986, and reviewed under Intute publication policies Garrept © Se Amenan Cont nee Al ah reved, cain ‘lors Pertinent dscsson wil be published fo the May-June PB6 ACY Suc ‘ral Journal i feeived by Jan, 1.1988, ACI Structural Journal J July-August 1987 ‘Bekele Mekonnen iste desgn department head of Consraction Design Ener rte, Adis Abbe, Ehiop. He reccied a BSc degre from Ade Abobo ‘Unies in 1967 and an MS dere from URI n 1969. He she erty of the Ethiopian National Earthquake Commitee onda senor member of he Ethiopian Associaion of Eines and Arches. For the assumed deflection pattern, the beams are in double curvature and rotate by M ” EK, @ where K, = stiffness of beams. In Fig. 1(¢), the rotations and unknown deflection are shown as reactions and moment load, respectively. Taking moments about midspan of the conjugate beam (by integration of pxdx, for the curvature load), the column deflection is obtained as Mh a-a een @ 2) Mk pyle where ¥ = K\/K, = column to beam stiffness ratio. At incipient buckling P=P.= ae ® but 2M pi © ° P Therefore, combining Eq. (3), (4), and (5) F acota) i or a where x? (1 — acota) p= = Mi acote) 8) DR In Eq, (8), B varies between 1.00 at « = x/2 or k = 1.0 and 0.822 at a = O or k = &. But for large values of ¥ ice., large values of k, Eq, (7) shows that the ef- feet of B on k is insignificant. In the interest of simplic- ity, therefore, 8 may be taken as 1.0 and the term m°/ 12’ = 0,822 in Eq. (7) may be rounded down to 0.8, giving kes TF OR © Eq. (9) gives results within 2 percent of the exact value given by Eq. (7). UNEQUAL END RESTRAINT ‘The exact stability analysis of an elastically re- strained column with unequal end restraint is given in Reference 2. The resulting effective length factor equa- tions are transcendental equations which are best solved by the use of alignment charts. However, the following Fig. 1—Buckling of frame of equal end restraint: (a) stability model; (b) deflected shape; (c} conjugate beam ACI Structural Journal | July-August 1987 317 \ (b) BE (¢) Fig. 2—Buckling of frame of unequal end restraint: (a) stability model; (b) de- flected shape; (c) conjugate beam Fig. 3—Bending moments under lateral shear analysis shows that an approximate equation with a maximum error of +2 percent can again be derived. In Fig. 2, a straight-line curvature diagram is used as 4 first approximation for calculating the deflection with the conjugate beam method. This is equivalent to using a lateral shear V as in Fig. 3. As before, the end slopes are obtained for the restraining beams assumed to be subjected to antisymmetrical bending moments at each end. Thus M, 6K © and My a= ae ay ‘Taking moments about the column midheight, the deflection is obtained directly as +i) 6 Or, substituting the values of the end slopes from Eq. (10) and (1) into Eq. (12), and simplifying 318 h A= + YM ++ YIM pee 3) where , and ¥; are the column-to-beam stiffness ratios at Points 1 and 2, respectively. From Fig. 2(c), considering horizontal equilibrium of the conjugate beam M4, Mh 04 soe = + a 1 or Mo MM Magy SEK, * DEK ~ EK, * DEK, From Eq. (15), the end moments may be obtained in terms of the story moment as G and = GAG) ae Note from Eq. (16) and (17) that the inflection point is always within the column height for the assumed model. This fact explains the inappropriateness of the ‘model for use in multistory frames with very flexible beams, because in such cases inflection points may not exist in several top and bottom stories. Substituting values of M, and M, from Eq. (16) and (17) into Eq. (13) and simplifying an a Qs Ege Oth om ACI Structural Journal / July-August 1987 or, as a function of the deflection of a column with ends fully restrained against rotation (M+ My) j, (vie a (DEI) "© (@)I2EK. as) Where the constant a, which may be termed the rigidity coefficient, is given by — Dive + 6 +6 (20) The absolute rigidity or lateral stiffness K, is then ob- tained as 12. P cr) v x = ¥ = ax, ( ‘The term in brackets generally being a constant, the relative rigidity or lateral stiffness is given by aK,, and is always less than the flexural stiffness K.. Ac incipient stability failure, in the absence of lateral load, and under the action of vertical load, the de- flected shape will again be a sine curve; but the preced- ing result for deflection under lateral load may be ex- pected to give useful results, with certain modifica- tions. Thus, from Fig. 2(6) @2) and @3) r, substituting the approximate value of from Eq. (19) into Eq. (23) fe 4) where the rigidity coefficient @ is given by Eq. (20). For vy = Yr = 0, Eq. (24) gives k = 23/12, a result with a 10 percent error since the exact solution from Eq. (7) gives k = 1.0. Now let \ be the correction factor needed to obtain the exact value of k as defined by @s) ‘Then, for yy = ¥; = 0,4 = 12/x*, The deflected shape being a complete sine curve, this case represents the ‘maximum departure from the assumed straight line and hence the maximum value of d. For Yi = vz = 2, = 1.0, that is, no correction is needed since for such a case the deflected shape would actually be a straight line as assumed. ACI Structural Journal / July-August 1987 Thus, for the two extremes, the corrected equations for kare +4W, +) +6 With +6 +0 26) a) 2s + 4, + 4) +6 rd) Vit het 6 a era en Observing the dominance of the product term in Eq. (27) and making empirical adjustments to it analogous to those made above in obtaining Eq. (9), the following. generally applicable approximate equation for & is fi- nally obtained urvi + au! + wD +6 ke Vit +6 cD where vi = 0.80, and Yi = 0.84 To allow for continuity, the definition of y is gener- alized asthe ratio of the sum of the column stiffness to the sum of the beam stiffnesses meeting ata joint.” Values of & obtained from a computer solution of the previously mentioned transcendental equations are given in Tables 1 and 2. Comparison of the corre- Sponding table computed using Eg. (28) (not included herein), shows that the error in Eg, (28) never exceeds + 2 percent, proving that Eq, (28) may be used in lieu of current alignment chars. Reference 4 gives tables for both braced and un- braced cases, but considerable errors have been found, especially in the table for braced columns. COMPARISON BETWEEN ABOVE TWO MODELS In the special case of equal end restraint yy = v2 = ¥. Eq. (20) and (28) (based on the second model) re- duce, respectively, 10 9) and k= IFO G0) Eq. (30) is identical to Eq. (9), based on the first model, Similarly, if the average end restraint y, is used in- stead of individual values, the following equations are obtained ik en and k= TF OR, @2 319 Table 1 — Effective length factor k: Columns not braced against sidesway zo [or [o2 [os [oa [os [os [o7 [os [oo | osi[io nr [pXfo fort] 02s] 0.43 | 0.67 | 1.00] 1.50 | 2.33 | 4.00 | 9.00 | 10.0 | = © | 0 | 00] 102 | vor | 107 [att | a6 [22 [rai | nas [165 | 68 | 2.00 ot [ot [1.02 | 1.08 | 1.06 | 1.09 | 1.3 | vas | 1.24 | 1.33 | 1.47 | 167 | 1.69 | 2.04 02 [0.25 [1.08 [1.06 | 1.08 | 1.11 | 1.18 | 1.20 [1.27 [a6 | 149 [a.m [1.73 | 208 03 | 0.43 | nor | 1.09 | vir | 1.4 {118 | 1.23 [1.30 | 1.39 | 1.53 [1.75 21a 04 | 067 [ast [aaa [vis] ae | 122] 127 [a4 ) 143 | 157 | nao] 1.83 | 222 os | 1.00 [1.16 | 1.18 [1.20 [1.23 [127] 132 [1.39 [149 [0 [rar] 190 [233 06 | 1.50 | 122 | 1.24 [1.27 [120 [134] 139 [146 [156 [1.72 [198] 2.01 [2.48 or [2.33 | 1.31 | 1.39 | 1.36 | 1.39 | 1.43 | 1.99 | 1.56 | 1.67 | vga | 213 | 207 | 29 08 [4.00 | 145 | 147 | 149 [ss [157 163 | 1.72 [ 188 | 208 [238] 2.43 [3.8 09 | 9.00 | 165 | 1.67 [1.71 [1.75 | a0] 17 | 198 | 243 | 2.8 | 207] 295 | 4.27 1 | 10.0 | 1.68 | 169 | 1.73 [1.77 | vas | 1.90 | 201 [2.17 | 243 | 295] 3.00 | 246 10 | = [2.00] 208] 208 | 214 [222] 233 [240273 | 3.18 [427] oas[ = 2 1 A comparison of these approximations with the more exact equations for individual end restraint is made in graphical form in Fig. 4. Thus, results obtained using the generally applicable Eq. (28) are plotted in Fig. 4(b) together with those obtained using Eq. (32), which plot as straight lines tangent thereto. Also shown in Fig, 4(b) are two straight lines representing the limits ¥,/¥s < 3and ¥\/¥: > %, within which the discrepancy be- tween the more conservative linear equations and more accurate curves may be considered minor. In Fig. 4(a), the curves representing these limits are plotted in terms of y (see Curves 1 and 3). In the alter- native plot for Curves 1 and 3, in terms of the mean value Yq, the two limits coincide and the single broken line is obtained (see Curve 4). The plot of Eq. (32), for the case of equal end restraint, is also shown as Curve 2. The difference between the latter two curves consti- tutes the error in Eq. (32), with respect to Eq. (28). Analytic comparison shows that this discrepancy at- tains its maximum value of 8 percent at Yu = 10, the upper limit subsequently proposed. ‘At this point, it must be recalled that this difference pertains strictly to the single story models of Fig. 1(a) and 2(a), and, as demonstrated by the subsequent prac- -al examples, does not reflect the true error in Eq. (32), since Eq. (28) itself errs on the unsafe side when applied to multistory frames. Fig. 5 shows the corre- sponding plots in terms of rigidity, the maximum dis- crepancy in this case being 18 percent. RANGE OF APPLICATION As stated earlier, the foregoing equations for effec- tive length and rigidity, and by extension current align- ‘ment charts, cannot be expected to apply in those cases where column inflection points are outside the story under consideration. In this connection, the detailed study in Reference $ on location of column inflection points in multistory frames gives relevant information, 320 from which it can be concluded that for regular frames, column inflection points will remain within each col- ‘umn provided ¥,, is less than about 10. For irregular frames with variations in beam stiff- ness and story height, the preceding limit should be re- duced to obtain inflection points within columns. It is also interesting that in Fig. 4(a) and S(a), which show the variation of effective length and rigidity factors re- spectively, with y, for the preceding single-story models, all the curves are almost straight when Yn €X- ceeds 10 to 15. In view of the findings in Reference 5, this would suggest that limiting ¥,. to 10, rather than the mean value Yq. is a necessary and sufficient condition for the application of the proposed equations and current de- sign aids, This conclusion is also supported by the sub- sequent analysis of several frames. SPECIAL CASES For cases of special interest, the general equations given for effective length factor (Eq. (28)] and rigidity coef- ficient [Eq. (20)] may be simplified to give the follow- ing results. Fixed lower support: ¥; = 0 Ee Gs) a 04) Hinged lower support: Ys = 2 k= ees 5) G 06) Med ACI Structural Journal / July-August 1987 Table 2—Effective length factor k: Columns braced against sidesway = [0 01 oz [os [oa [os [os [a7 [os [09 | oa [io a 0 for | 0.25 | 043 | 0467 | 1.00 | 1.30 | 2.33 | 5.00 | .00| 10.0 | o ° 0.30 | 0.83 | 0.56 | 0.38 | 061 | 043 | 06s | 0.66 | 0.68 | 0.0 | 0.69 | 0-70 om 0.53 [os | 0.8 | 0.61 [ 0463 | 046 | 0.68 | 0.69 | 0.71 [ 072 | 072 | 074 02 0.56 | 0.38 | 0.61 | 0.64 [067 | 049 [on [0.73 | 0.74 | 06 | 0.76 | 0.77 03, 0.58 | 0461 | 0.64 | 0.67 | 0.70 | 0.72 [ 074 | 0.76] 0.78 | 0.79] 0.80 | 0.81 os [067 | 061 | 0.463 | 0.67 | 0.20 | 0.72 | 075 [0.77 [0.79 Joa [0.83 | 083 | 0.84 05 [1.00 [0.43 | 0.66 | 0.69 | 072 | 075 | 0.77 | 0.40 | 042 | 0.84 | 086| 0.86 [0.88 06 [1.30 [04s | os | 0.71 | 0.78 | 0.77 | 0.80 | 092 | 0.85 | 087 | 0.89] 0.89 [0.90 07 [233 [0.66 | 0.09 | 0.73 | 0.76 [ 079 | oz | os | oa7 | oso | os | om | 093 08 | 400 [0.68 [on | 0.74 | 0.78 | ot | 08s | 07 | 0.89 | 092 | 094] 094| 0.96 0.9 | 9.00 [0.69 | o7 [0.76 0.79 [os | 06 | 089 | 0.91 | 0.94 | 096 | 096 | 0.98 0.91 | 10.00 | 0.69 [072 | 0.76 | 0.80 | 0.3 | 0.86 | 0.89 | 0.1 | 0.94 | 096 | 0.96 | 0.98, 1.00 |= [0.70 | 0.74 | 0.77 [0.81 | 0.84 | 0.88 | 0.90 | 0.93 | 0.96 | 0.98 | 0.98 | 1.00 Subscripts 1 and 2 designate ends of column, THE | ioe 40) (b) = Fig. 4—Effective length factor k ACI Structural Journal | July-August 1987 321 (b) Fig. S—Rigidity coefficient Some of these results, or close approximations thereof, have been obtained previously.” ADJUSTMENTS FOR TOP AND BOTTOM FLOORS The need for special treatment of top and bottom floors is well known.** Such treatment, however, has to be consistent with the need for simplicity in practical application. Accordingly, the following simple empiri- cal modifications, which will give reasonable results, are proposed, For the top floor, the value y; for the top joint should be taken as twice the actual value, with the floor then being treated as any intermediate floor. For the bottom floor of multistory frames with fixed supports, an average value y., equal {0 one-half the top restraint value y, should be used. The floor should then bbe treated as having equal end restraint y,, at each end. Eq. (33) and (34) should, therefore, be used only for single-story frames, for which they give exact results. No modifications are needed for hinged end condi- tions. 322 Remember that these modifications are needed be- cause the basic equations that are derived for single- story frames will otherwise give significant errors when applied to multistory frames, due to the boundary ef- fects. PARTIAL BRACING Consideration of stability and rigidity in the case of partial bracing by partitions and other so called non- structural components has been introduced by Wood." The European Convention for Constructional Steel- work have endorsed Wood's recommendations and corporated them in their steel code," giving several charts for deflection calculation of partially braced frames. Alternatively, and perhaps more simply, the rigidity approach reviewed may be used as follows. In the pres~ cence of stiffening due to cladding in any story, the to- tal story stiffness may be obtained by direct addition of the individual stiffmesses of the frames and the clad- ding. Thus, for bracing stiffness of a story = Sy, a bracing rigidity coefficient a, may be defined such that ACI Structural Journal / July-August 1987 a. = TREK, G7 where EK, is the sum of the flexural stiffnesses of the columns in the story. Total story stiffness is then obtained as K, sere (ex an : 9) Stability effects may be considered using this paper’s Eq. (47). Reference 8 gives recommendations on maxi- mum permissible values for the bracing stiffness § 12a,. GRINTER SUBSTITUTE FRAME As will be shown by subsequent examples, using the local single-story model for predicting either the stabil- ity or the rigidity of columns in multistory frames can- not be expected to give acceptable results in all cases of practical interest. In fact, gross errors, generally on the unsafe side, may be expected when the preceding ap- proximate equations or current alignment charts are used outside the above proposed range of application. It is therefore necessary in such cases to use a model that more accurately predicts the actual deflection of multistory frames under lateral load. Fortunately, Grinter" has discovered a simple substitute frame which gives almost exact results in all but the most unusual cases. The same substitute frame was independently discovered by Kloucek,"* who made an exhaustive study of frame deflections and introduced certain modifica- tions to the Grinter frame to make it applicable even to unusual cases Grinter’s discovery is equivalent to the observation that the deflection under lateral load of any frame may be expected to be the same as that of a substitute sym- metrical single bay frame with member stiffnesses ob- tained as the sum of the stiffnesses of individual mem- bers of the original frame (see Fig. 6). For such a frame Naylor!” has introduced a simple ‘moment distribution technique for finding the bending ‘moments due to lateral loads. This consists, basically, of using a substitute beam-column frame with the sum of the column stiffnesses and six times the sum of the beam stiffnesses (Fig. 6(c)] to find the conventional moment distribution factor at each levels using a carry- over factor of 1.0 instead of 0.5; and obtaining the fixed end moments (of the same sign) at each end of ‘each column, as one-half the story moment. After per- forming an otherwise conventional moment distribu- ACI Structural Journal | July-August 1987 (a) (b) tc) Fig. 6—(a) Real frame; (b) substitute symmetrical frame; (c) substitute beam-column frame + 4g, 0% Mei ‘ a a, Ke Kez LL Fig, 7—Definition of terms in Eq. (39) tion, as for a continuous beam, the final bending mo. ments are obtained at the ends of columns and re- straining beams. Having obtained the column and beam moments for the substitute frame, using Naylor’s or any other con- venient method, the interstory deflection for each col ‘umn may be obtained using the following simple equa- tion, which is easily derived from the slope-deflection equation B= (M+ M, + My + Ma) yep 40) where a and ay are defined as the ratio of the stiffness of a column under consideration to that of one re- straining beam at each end carrying the bending mo- ments Mz, and Mya, respectively. Fig. 7 clarifies any possible confusion between the various definitions of a and ¥. Eq. (40) could, of course, be generalized to ap- ply to any loading condition. However, this simple form will suffice for lateral loads. MOMENT MAGNIFIER According to ACI 318-83,’ the moment magnification factor 6, for frames not braced against sidesway is given by 323 1 = ae? 10 «n OEP. where EP, and EP, are the summations of factored ax- ial load and critical buckling load for all columns in a story, and 6 is the strength reduction factor. Combining Eq. (21) and (25) and simplifying, the lateral stiffness K, of any column is obtained as Kao 42) Fe (2) ‘The lateral stiffness for any story therefore is given by Eq. (43) as the summation of stiffnesses of all columns in the story IK, 3) I, The critical load for any story may be obtained in terms of the story stiffness as ohh tp, = 5% (44) Substituting this value of EP, into Eq. (41), the mo- ‘ment magnifier may be obtained in terms of the story stiffness as =-—yr 3) 1 ork In applying Eq. (45) in lieu of Eq. (41), the story stiff- ness EX; may be obtained, using Eq. (21), as the story shear over the interstory deflection or, alternatively, as the summation of lateral column stiffnesses obtained with the rigidity coefficient method described above. ‘The factor \ in the preceding equations varies be- tween 1.0 and 1.22 and may be obtained from Eq. (25) as A= Gar (46) However, for practical application, taking X = 1.1 for all cases will give satisfactory results. ‘Accordingly, the following approximate relation for the moment magnification factor 6, is finally obtained as an alternative to the ACI approximate equation —— 7 4 1.1 EP, an 1 EKA, ECCENTRIC LATERAL LOADS When lateral loads are eccentric to the center of tig ity at any level, the following approximate consider- ation using the concept of story rigidity may be ap- plied. 324 As stated in Reference 14, a rotation magnifier 6, ‘may be derived to take into account the effect of ec- centricity. The first-order story rotation about the cen- ter of rigidity 6, is magnified by the factor 6, deter- ‘mined from (as) Eq. (48) may also be presented in terms of the rota- tional story stiffness / as a9) where the total rotational stiffness / about the center of, jgidity is analogous to a polar moment of inertia and is obtained as T= 11 = EK OF + EK YP (50) ‘The overall magnification considering both translation and rotation will then be different for each frame in each story as given by 3.4 46, T+ ony where lateral load due to story shear Tateral load due to torsion The ratio -y for the respective orthogonal directions of the lateral load is given by* = Reh forthe direction (62) and y= FE for tne ycirection (53) where e = eccentricity of lateral load K,.and K,, are the rigidities in the x- and y~ ,J are the centroidal distances for each frame in the x- and y-direetions, respectively FULLY BRACED COLUMNS The following approximate formula for the effective length factor for braced columns is given by Newmark (as quoted in Reference 15) +08) 54) (U, + 08), + 0.8) oD ACI Structural Journal | July-August 1987 k= . . ny 5,600) (ana) (a) (b) Fig. 8—(a) Five-story concrete frame under gravity and lateral loads (from Refer- ence 16); (b) substitute frame Eq. (54) gives results within 2 percent of alignment charts and may be simplified for special cases roa wa 07 JPEEE tnedease (55) toa b= (PEI hinged case 86) v+o4 , k= TADS equalend fixity (57) Plots of Eq. (54) are given in Reference 15. The corre: sponding plots of Eq. (57) would be tangential to those of Eq. (54), which, as shown in Reference 15, are sim- ilar to the charts in Fig. 4(b), representing unbraced frames. ‘As before, Eq. (57) may be used in general by re- placing individual end fixities by the average value. In this case, the range of application need not be re stricted as k varies between much narrower limits. NUMERICAL EXAMPLES FOR SWAY FRAMES Example 1: Five-story three-bay frame The frame shown in Fig. 8(a) is selected to demon- strate the use of the story stiffness approach presented. ‘This same frame has been used in Reference 16 10 com: pare the method proposed therein with other methods Fig. 8(b) is the Grinter substitute frame for the frame of Fig. 8(a). Table 3 shows the loadings and member properties of the substitute frame taken directly from Reference 16. Member properties for the complete frame were taken from Reference 17 for the analysis of the same. ‘The analysis of the substitute frame using Naylor's method is shown next and summarized in Table 4, ACI Structural Journal / July-August 1987 Table 3—Data for frame of Fig. 8(a) Sifiness x 10, fey Loading, kips_ Column | Beam —| Story shear _| Axial Story [Ek EL A EP. s ma | 11535 3 22s 3 joa | 11835 93 63 3 Bs | sas 13 302 2 ase | sas | ana 308 i boa | mss ns 179 Properties of beam-column frame column stiffnesses: same as in Table 3 beam stiffness: six times Table 3 values = 6 x 115.25 = 691.5 Moment distribution factors e.g., fifth story: top column: 71.1/(71.1+ 691.5) = 0.093 bottom column: 71.1/(71.1 +79.2 4691.5) = 0.085 Values for other floors are similarly computed and shown in Table 4, Lateral load on beam-column frame is half the load on the Grinter substitute frame. Fixed end moments (FEM) in beam-column frame is lateral load times half the story height. For example, for the fifth story pen =i 31 x Bac tetnaip With the preceding data, Naylor's moment distribution method may be used to find all support moments in the The particular moment distribution technique adopted in Table 4 takes the unbalanced moment for ‘each cycle of distribution as the sum of the initial un- balanced moments at each joint plus the carry-over moments from neighboring joints (carry-over factor 1.0). This approach insures automatic error elimina- jon in successive distributions. The iteration is ter- 325 Table 4—Moments and story drift: Frame of ‘Column stiffness, > woe [SR 792 7a 10 inskip Beam stiffness, os ous wus wus wus 10 in-kip Distribution factors 0.108 0.059 0.100| 0,096 0.097 | 0.093 0.096) o.08s 0.093 FEM, in-kip 990990) 700 |ssi ssi | 381331 | 112 Unbalanced 1960 ua ie a moment in -kip First dsteibtion = 174 ~150) ~148 ~si| as as Second distribution 189-161 | 155 53] 47-15 Third distribution = 200) = 190162 | = 155 csi] “a7 o15 FEM 90990] 770 770] st mil a2 Distributed moment = 200) - 190 ~162 | ~ 155 “33) -a7 Carryover moment |200 — | 162 190| 106 tol] 15, moment, [1190790] 742798] 02 sofa ta a - a 078 on 0.09 0m column sffness | ‘beam sifness aM, FaMy hugs 2096 ism 926 a Mt Me 1980, sao | to 62 | oe 2M. ass 1036 2674 188 98 RELA 35 130 “es 6600 sms a= DME, 02 ost 039 028 0.10 A tiiaty 0 ost 039 024 0.09 coeticent method), A (computer solution | 0.83, 052 039 028 0.10 ful frame, in minated when successive distributed moments for each = [379 + 283 + 0.69(459 + 883)]/6600 = 0.241 in Joint converge approximately to the same respective value. The final column moments are obtained by summing up the fixed end moments, the support end distribu- tions, and the carry-over moments. The beam moments are equal in absolute value to the sum of the column ‘moments meeting at each joint. Interstory deflection is calculated next using Eq. (40) and the final moments obtained in Table 4. The details of the calculation are given for the fourth floor only. ‘The column-to-beam stiffness ratios e are first deter- mined for the Grinter substitute frame a = 79,2/115.25 = 0.69 The beam moments are obtained as the sum of adjoin- ing column moments My, = 80 + 379 (59 in.-kip My = 183 + 600 = 883 in.-kip ‘The column moments for the fourth floor are M, = 319 inckip Mz = 283 inkip TEI Ji = 12 x 19.2 x 10/144 326 6600 kips ‘The results obtained by this method are identical to the computer results for the substitute frame. Drift calculation using story-stiffness approach The drift for the first floor of the substitute frame is calculated next to demonstrate the use of the story stiffness approach. Results for all floors are shown at the bottom of Table 4 SEK. SEK, 90.4 + 85.8 115.250 Re Using Eq. (34) = 6 + 1.536 + 4 x 1.53) = 0.62 0.62 x 12 x 90.4 x 10° 4 4 kips/in. = SR = 0.4 in, MOMENT MAGNIFICATION FACTORS Table 5 shows moment magnification factors 6, com- puted with six different methods. The results clearly fall into two groups. In the first three the strength reduc- tion factor 6 is considered and in the last three it is not. ACI Structural Journal / July-August 1987 grits tat omy, dl | —1gs475x10% amt — | 4 4 a — 4 ee £2600 Z "wy “ “ mn a n| x " ny f- 3 I + }_9m_. ___4m__| (a) Fig. 9—Frames for Example 2 Table 5—Comparison of 5, values 7 eee] (tam) | Iuwcr ie] els] ie |e THOR ay In the ACI equation, the factor is obviously in tended to cater mainly to the possible presence of un- derstrength materials that would reduce the column stiffness. However, the probability that such a material exists throughout the column is much lower than for a single section. Accordingly, it could be argued that, for consideration of stability, it is more appropriate to use a higher reduction factor than the section capacity re- duction factor, or even to omit it altogether. In any case, whatever value is adopted for , there is no justi- fication for applying it in some methods and omitting others. Therefore, to comply with current ACI re- quirements by the second group of methods, the col- umn stiffness should be taken as $1. With this correc- tion, most of the discrepancies in Example I would dis- appear. Example 2: Seven-story single-bay frame ‘Two seven-story single-bay frames are shown in Fig. 9a) and 9(b). Fig. 9a) shows a typical frame of a re- ACI Structural Journal / July-August 1987 8.21108 mt b> cently completed extremely flexible flat slab building, Yn = 30. ‘The frame of Fig. 9(b) is derived from the first frame by increasing the stiffness of all beams by a factor of about 10, thereby obtaining a frame with more usual end restraints, Y= 3. Other details and loads are given in the figures themselves. The effective length factors k and rigidity coeffi- cients @ are obtained for each frame using the follow: ing five methods, in order of decreasing accuracy. Method 1 — Interstory deflection is calculated using computer program which considers extension of mem- bers. The corresponding rigidity coefficients a and ef- fective length factors & are determined using Eq. (21) and (25), respectively. Method 2 — As in Method 1, except that member extension is neglected. Method 3 — Coefficients a and k are determined us- ing and restraint values for each column and the gen- «ral equations, Eq. (20) and (28), respectively. Method 4 — Coefficients a and k are determined us- ing average end restraints in Eq. (31) and (32), respec- tively. Method 5 — Same as Method 4 but with adjust- ‘ments for top and bottom floors as proposed earlier. Results of analysis with these five methods are pre- sented in Tables 6 and 7. ASSESSMENT OF RESULTS Based on a study of the results obtained for the above frames, the following conclusions may be made. I. The Grinter substitute frame, without modifica- tion, may be used for determining the slenderness ef- fects and lateral stiffness of frames when effects of col- 327 Table 6— Frame of Fig, 9(a) (Example 2): Effective length and rigidity coefficients using different methods S09 T stear¥, | Story di, mm Rigidity coetient a Efstive length factor Method | Sehod | Mtiod | Mao | Method | Method | Method | Method] Mod] Method | Meta | Mod seory| en | Mpe] Meg Mepot| Megod Me | ree Meee se ee etl a, 7 oa [a8 | 002 [oor | oo | oo | cos | or | so | ar | a3 | so [var 5 nis [|u2 | oox | 00% | oon | oon | — | so | so | so | so | — [aa 5 37 | 133 | cas | oo | 0033 [oo | — | a7 | a6 | so | so | — [2a «| [is | 146 | oom [002 [00m | oom | — | 4s | a4 | so | so | — 293 3 | 2 [as | 2 | oom | oo | oon | oom | — | 41 | a1 | so | so 33 2 [2m [ns [us [00s [oow | oma [oo [ — | 36 | 3s | so | so | — |as 1] 20 [3a [38 [om [ons = [oo [30 [29 | 19 | — | «1 [ma Method 2 — Exact Solution neplcting nember extension. Matea 3 — Slsion wee snd Neos Sittion ge Fa. GI aR 3, Table 7— Frame of Fig. 9(b) (Example 2): Effective length and rigidity coetticients using different methods S05 stear¥, | Story it mn Rigidity coefficient a Efecive leah factor & Method | Method | Nieod | Nthod [ Method | Method | Method] Mthod [Method | Method | Method | athod S00 | ae ee ee eee eee eee soe Mee Mego eee eres 7 {#7 | 1s | oss | oss [oz | oss | 032 205 | ie [ne | 16s as © [19 | 206 | 143 | om | ose | 026 | oa | — | 2 | um [1s | os | — [2a s | 16 | 2se [19s [oan | oa | ozs | 026 | — | 20 [19 [18 [is | — 20 «| 7 | 29 | 235 | oats | o2ss | 026 | 026 ig [179 [us [te 20 x [| m [30 | 26 | oan | 0266 | 02 | oa | — | 190 | 17] 18 | 1s | — [20s 2 | 22 [28 | 25 | oss | 02m | oas | oas | — | 17 | 16 | vas | ua | — [29 [ae [a2 fu [ose | oa [om | — | ov | ise | use [139 ist [5.18 ‘od | — Eat lain coving nb xenon Method 3 — Solution sine Ea (20 tn6 GB Metted 4 — Solution uig Ea (31) an G2 Metnod $ — Solon wih proposed asustments for top and botom Hor umn extension are insignificant. When significant, ne- glect of such effects is unconservative, 2. The use of Eq. (28) and design aids like the Jack- son and Mooreland alignment charts give acceptable results within limits. The proposed limit Yu, < 10 may be used as a guide. Outside this range, unconservative results are obtained compared to the more exact solu- tions. 3. Eq. (32), the approximate formula based on the average end restraint, may be used in lieu of Eq. (28) and the alignment charts. This is preferable to using charts or the corresponding equation, as the accuracy of either, compared to the more rigorous solutions, is generally no better, and in many cases worse than the simpler equation. Similarly, the simpler Eq. (31) for rigidity is preferable to Eq. (20). 4. All the preceding approximate equations, as well ‘as existing alignment charts, can be based on the sub- stitute Grinter frame depicted in Fig. 6(b), since there is no need to determine individual values of effective length and rigidity for each column. The resulting ef- fective length factor or rigidity coefficient may be used to obtain the effective story height or story rigidity, re- spectively, saving considerable design time. 328 5. Neglect of column extension is seriously unconser- vative for narrow and rigid frames, e.g., the frame of Fig. 9(b). CONCLUSION Current design aids for calculating effective length and deflection of multistory frames are generally based ‘on a conventional local model of one column and its restraining beams. It is shown that such aids can be used with confidence only under certain conditions, namely for frames that are not 100 flexible (mx < 10) and in which effects of column extension are insigni cant. For such frames, simple, easily remembered ap- proximate equations are derived that would give results with accuracy comparable to, and in some cases better than, current design aids. ‘The Grinter substitute frame is recommended for use with these methods as well as for handling flexible frames. In such cases, bending moments may be ob- tained using Naylor's method and deflections, then cal- culated using Eq. (40). Consideration of stability and drift is treated in a simple and unified manner, including effects of clad- ding and eccentric lateral loads. ACI Structural Journal / July-August 1987 ACKNOWLEDGMENTS “The computer programs fr frame analysis were prepared by Eng. Gedion Worede, The author acknowledges with pratitude his asi tance in analyzing the example frames. The assistance of Mrs. Mix make Abebe and Miss Melesu Chakisa is also gratefully acknow!- ceded, NOTATION ° lateral stifesscoeficient 2, = bracing stiffness coeticent = eccentricity of story shear measured from center of rg vay modulus of elasticity modulus of elasticity of concrete height of column height of story rotational sifness of story about center of rigidity ‘moment of inertia of bear ‘moment of inertia of columa Story rigidity in xirection Story riaiity in y-direction effective length factor Flexural stiffnes of beam Aexural stiffness of column lateral sifeess of column Intra stiffness of column in sdiretion lateral stiffness of eoluma in ydiretion bending moment bending moment at upper end of column bending moment at lower end of column bending moment (absolve value) at beam ends 1 and 2, respectively fal load on column = -rtcal bucking load of column factored aia load on column stability index BK/GK, + Ey) bracing stiftess factor bracing stiffs story torsional moment = story shear factored story shear distance from centr of Fidity to column in xairection tistance from center of rghit t0 column in y-direction 2k ‘ariable function of correction factor due to eccentricity in story shear ‘moment magnifier ‘moment magnifier due to translation ‘moment magnifier due tration imterstory detection factor that depends on shape of curvature diagram of column = tmnit curvature foad on conjugate beam ACI Structural Journal / July-August 1987 angle of rotation Fatio of flexural stifneses of columns and beams = 08 average value fora column n= limiting value for ¥ CONVERSION FACTORS i, 254mm Hoot = 0.3048 Tkip = 4.448KN I kipfin? = 6.895 MPa Hinckip = 0.113 KNm REFERENCES 1, ACI Committee 318, “Building Code Requirements for Rein forced Conerete (ACI 318-83)," American Concrete Institute, De trot, 1983, 11 p. 2. McGite, We, Ste! Structures, Prentice-Hall, Inc, Englewood Clits, 1986, pp. 431-473 3. ACI Committee 318, “Commentary on Building Code Require- ments for Reinforced Concrete (318-83) (ACI SIBR-B3), American ‘Conctte Institute, Detot, 1983, 155 p. “4-CRSI Handbook, sh Edition, Concrete Reinforcing Stel Insti- tute, Chicago, 1980, pp. 2-14 - 2-13 5! Muto, Aseismic Desi of Buildings, Maruzen Co. Limited, Tokyo, 1974, 361 pp. 6. Blume, John A.; Newmark, Nathan M.; and Corning, Leo H., Design of Nulistory Reinforced Concrete Buildings for Earthquake Motions, Portland Cement Association, Skokie, 196i, pp. 71-73. 7. Cheong Sait Moy, Fy "Control of Defesions in Unbraced Stee! Frames,” Proceedings naitution of Civil Enginers (London), Part, 2,81, Dee, 1974, pp. 619-63, '8. European Recommendations for Steel Construction, European Convention for Constuctional tcelwork, Roterdam, 1978, 358 p. '. Wood, R. H., “Effective Lengths of Columns in Muli-Story Buligings," Structural Engineer (London), V. $2, No.7, July 1974, 1p. 238-288, No. 8, Aug. 1974, pp. 298-302. and No. 9, Sept. 1974, bp. 341-346, id, Wood, R. H., and Roberts, E. H., “A Graphical Method of Predicting Sidesway in the Design of Multstory Bulldings," Pro- ‘ceedings, Institution of Civil Engineers (London), Patt 2, V. 89, June 1995, pp. 353-372. 1! Grier, L-E., Theory of Modern Steel Structures, Revised Edition, Macmillan Publishing Co., New York, 1949, V. 2,140 pp 12. Kioueek, C. V., Distbution of Deformation, Arta, Prague, 1949, 512 pp. 1B. Naylor, N., “Side Sway in Symmetrical Building Frames, Structural Engineer (London), V.28, No.4, Ap. 1980, pp. 99-103 Ti MacGregor, James G., and Hage, Sven F, Stabilly Analysis and Design of Concrete Frames,” Procedings, ASCE, V. 103, ST10, (Oct, 1977, pp. 1983-1970, 13, Kavanagh, Thomas C., “Effective Length of Framed Col- umni,” Transactions, ASCE, V. 127, Part 2, 1962, pp. 81-101 16. Hey Cheng-Tan Thomas, “Lateral Displacement for Un- braced Concrete Frame Buildings," ACI Jounwat, Proceedings V. 82, No. 6, Nov-Dec. 1985, pp. 853-862. 17, Rice, Paul F. and Hoffman, Edward S., Structural Design Guide to the ACI Bullding Code, 2d Eaton, Van Nostrand Rein fold Co., New York, 1979, 470 pp. 329

You might also like