Design of Diagonal Cross Bracings Part 1: Theoretical Study

A. PICARD and D. BEAULIEU

Diagonal bracing is commonly used in steel structures to resist horizontal loads. In current practice, the design of this type of bracing system is based on the assumption the compression diagonal has negligible capacity and the tension diagonal resists the total load (Fig. 1). If the diagonals are connected at their intersection point (usual practice), this design procedure is conservative because the effect of this connection on the out-of-plane buckling capacity of the compression diagonal is ignored.1'2,3'4 The restraint provided to the compression diagonal by the loaded tension diagonal is generally sufficient to consider that the effective length of the compression diagonal is 0.50 times the diagonal length (KL = 0.5L) for out-of-plane buckling as for in-plane buckling. Analytical and experimental results1'2 have also shown the ultimate horizontal load on the bracing system is much higher than the horizontal component of the yielding strength of the tension member, because of load sharing between the diagonals. The assumption that the compression diagonal has negligible capacity usually results in overdesign. This paper presents the results of a theoretical study aimed at the determination of the transverse stiffness offered by the tension diagonal in cross-bracing systems and at the evaluation of the effect of this stiffness on the out-of-plane buckling resistance of the compression diagonal. The theory is supported by seven transverse stiffness tests and 15 buckling tests. The test results are reported in the second part of the paper. BUCKLING OF THE COMPRESSION DIAGONAL In double diagonal cross bracing, the tension diagonal acts as an elastic spring at the point of intersection of the compression diagonal as shown in Fig. 2a, where a is the spring stiffness (kips/in. or kN/mm). If a = 0, then K = 1.0 (Fig. 2b) and the elastic critical load Cce is equal to:

Hypotheses:- Connections with negligible moment capacity - Diagonal in compression ignored

Fig. 1.

Usual design

A. Picard is Professor of Civil Engineering, Laval University, Quebec, Canada. D. Beau lieu is Associate Professor of Civil Engineering, Laval University.

Fig. 2.

Buckling modes for compression diagonal

122

ENGINEERING JOURNAL / AMERICAN INSTITUTE OF STEEL CONSTRUCTION

The limit spring stiffness is given by:5 aiim = 7T^ (4) The variation of K is plotted in Fig.1'2 The transverse stiffness or spring stiffness provided by the tension diagonal is obtained from the solution of the differential equation and is given by:1'2 a=m\ \ L ) (v 3 tank v) (9) 4> Exact solution (Ref. 2c). Ic is the moment of inertia of the compression diagonal considering out-of-plane buckling. Using Eqs. the tension diagonal is still behaving elastically.5 and the elastic critical load is equal to: r = 4TT2EIC 4C„ (2) The variation of the dimensionless elastic buckling load is shown in Fig. is also shown in the figure. 3. as shown in Fig. 4 against the dimensionless spring stiffness. 7 is slightly conservative.1/ 2VEI C 1 \ ' // 'J/ // AS Jnear re ation (Eq .— " F j ^ Cce Ce where i fS v ^ .5 (8) Therefore. 5 and 6. It can be seen that Eq. Relationship between effective length factor and nondimensional spring stiffness THIRD QUARTER/1987 123 . In other words. If a = °°. 3 and 7. given in Refs. The exact solution. we obtain: Cr„ = | l + — I C < \CP (7) To determine the parameters a and 7. TRANSVERSE STIFFNESS OF THE TENSION DIAGONAL Lr Let us define the non-dimensional spring stiffness as: = oL 7 " C (5) From Eqs. For the problem studied here. assuming elastic behavior. 5 ) . 1. When the compression diagonal buckles.In this equation. since both diagonals are identical. The exact solution obtained by Timoshenko and Gere is also shown in the figure. 5 is classical. Q represents the transverse force transmitted to the tension diagonal by the compression diagonal at buckling. we will consider a prismatic member subjected to a transversely applied point load Q and a concentrically applied tensile force T. The general solution of the differential equation obtained from equilibrium of the tension member shown in Fig. 4. the spring does not have to be infinitely stiff to obtain K = 0. 3 in terms of the dimensionless spring stiffness.\ . the limit value of the nondimensional spring stiffness is equal to: 7lim = CXlim — = 1 6 (6) If we assume a linear relationship between Cce and 7. the effective length factor becomes: K= V l 6 + 37 0.s i n u + 4 u cos u ( 7 . The elastic behavior assumption is easily satisfied for X-bracing. 5. 7 ) 4/ S/S SS \y\ 8 10 Y 12 1 4 1 6 18 20 Fig. the effective length factor is given by: K= \ l ^ r (3) Timoshenko and Gere 5 demonstrated there is a limit spring stiffness above which the elastic critical load is 4Ce (Fig.5. 4 and 5. then K = 0. Relationship between elastic buckling load and nondimensional spring stiffness Fig.

Combining 1.09v2) (13) EI.5 L -0. 200 40 0 0 0 A (14) In Eqs. Since the diagonals are connected at their intersection.. 13 very closely approximates Eq. 5 and 9. 13 60 40 AS AS 20 /y 4 1^4. It can be seen Eq. should not exceed 200. the compression diagonal could be stiffer. However. If SI units are used E . 6.38 is in (ksi) -1 . Comparison of Eqs.e.where v is defined as: The following approximation for Eq.tank v) (12) The maximum force in the tension diagonal being AFy.( 3 + 1.L i n * ar relatio n (Eq. Eq. then 0 < v 2 < 80. both diagonals are identical and It = Ic = I. 12 and 13 are compared in Fig. 10 together with Eq. 12 when 0 < v 2 < 80. we obtain the following equation for the evaluation of the dimensionless spring stiffness: \ i r 2 / \ / c / ( v . considering out-of-plane bending of the X-bracing. both diagonals. Eqs.5 L- 160 ss\ 140 120 f Exact solution (Eq. 12 is proposed: Hfr (10) 7 = ^ . Tension member subjected to transverse force at mid-span to 0. Let us first assume the out-of-plane slenderness ratio of the compression diagonal. In the previous equation. Since this paper deals with double diagonal bracings.tank v) (11) In X-bracing systems. in X-trusses supporting loads which always act in the same direction. 0.86 10 20 3 0 4 0 5 0 6 0 7 0 8 0 9 0 Fig.20 (MPa)~\ Then Fy is in MPa. the constant 1. 6. Eq. 12 and 13 124 ENGINEERING JOURNAL / AMERICAN INSTITUTE OF STEEL CONSTRUCTION . 11 can be rewritten as: 7 = TT / (v . i. E is the elastic modulus of the material and It is the moment of inertia of the tension diagonal. 14 and E = 29 000 ksi leads to: l\AFy 'A EI T — — ' — ffflfffo Considering a maximum value for Fy of 58 ksi. we get: L/2 r LY 2.l2K 100 80 4 4y V $7 ^ . 5.200 000 MPa and the constant is equal Fig. 9 and 10. This practical range for parameter v 2 was evaluated as follows.

47Ce (16) Introducing Eq.72 (see Fig. As seen.Eq.( E q .4 0.2 1 . Eq.7 min = 4.523-^Uo.60 0. the rotational restraint was relatively important. 19 can be used to determine the effective length factor (Cce = Ccr where Ccr is the measured buckling load). 8. CIT ratio 6 2. from Eq.86 + 4.85 times the half diagonal length.8 1. (18) = 0. If the compression diagonal buckles in the elastic range.e.. and test results suggest use of an effective length factor of 0. the effective length factor increases. combining Eqs.3 In the derivations the rotational restraint of the connec- 0.6 1 0. These equations are valid for continuous diagonals which are connected at their intersection point.6 C/T 5 K = 0. One is obtained from Eq. Therefore. 2 4 = 1. when the CIT ratio is larger than 1. will be used in the theoretical analysis.50 CrJT (19) K2Cr.72 • ! 0. from Eq. 3. the spring stiffness is equal to the flexural stiffness. 1.2 IQ 0.55 During the buckling tests reported in the second part of the paper. 17 can be rewritten as: y = 4. COMPRESSION-TENSION RATIO Since the behavior of the braced frame shown in Fig. |J L vC . 19. 1 is elastic up to buckling of the compression diagonal. 19 and the other from the following equation: Cce = TT2 EII(KE)2. 22) THIRD QUARTER/1987 125 . 4 \J Fig. For inelastic buckling. 4).4 1 (see fi rT n•° ^•^•^i _60^_ 1.425 if the total length of the diagonal is considered.75. 8.0 (a min = 48 EIIL3. Eq.36 — (17) KMQ. Effective length factor vs. 18.3-7) 1. 5 or Eq. 16 into Eq.36 Combining 8 and 18 gives: K= \/0. Eq. 10 can be rewritten as: TTL T 2 J T:2EI K= = 2. Ce = K2CC( Therefore. J iO) 0. K = 0. 7. Therefore a is never equal to zero and K is always smaller than 1. It reaches its maximum value when the CI T ratio is equal to 0. It is interesting to note the first term in Eq.65 ^ . the value of T is known. Kmax = 0.70 0. In Ref. If one diagonal is interrupted at the intersection point. Since Ic = It = /.523--^->0. 19 can thus be rewritten as: \/0. 4. two values of Cce can be computed. 7.8 A/A b 1. 13 gives: y = 4.0 1.6.6 8 1 0 1 2 1 4 C/T Fig. i. Variation in CIT ratio with AIAb ratio (Eq. 1 r ~~' t From Eq.50. However. 6). Two final comments should be made concerning the previous equations. Then T = 0.86. The results of the transverse stiffness tests reported in the second part of the paper demonstrate the validity of Eqs. Eq. 19 is not valid. Eq. 10 and 13 gives: a = Q ..50 (20) CIT Eq. = 48|/ + E L 3 4 3 6 r L tions was ignored.. the measured inelastic buckling load can be introduced into AISC equations7 (safety factor removed) or into CSA equations8 (resistance factor removed) to determine the effective length factor. 20 is plotted in Fig. the force in the tension diagonal was kept constant. constant stiffness is not maintained and the connection at this point may become a weak link when the interrupted diagonal is compressed. These two values are compared to evaluate the accuracy of Eq. being an acceptable approximation (Fig. 13 and 15.0 y 1. 5.86 + 4. 15 represents the flexural stiffness of the tension diagonal. 13. With this value of K.2 r^\ 0. the ratio CceIT is equal to the ratio of the force in the compression diagonal C to that in the tension diagonal T obtained from an elastic analysis of the frame.

The writers are thankful to Moffet and to the Natural Sciences and Engineering Research Council of Canada for its financial support. 22 is plotted in Fig. III. A. American Institute of Steel Construction. Finch Double Diagonal Cross Bracings in Buildings Research Report No.379-1. Chicago. This is the most critical loading condition for the design of the compression diagonal because the CIT ratio exceeds one. 8. Consequently. C. Vickers. 21 becomes: REFERENCES 1. Australia. Vol.5 times the diagonal length when the diagonals are continuous and attached at the intersection point. Therefore. F. which corresponds to the minimum spring stiffness. Eq. It can be seen that large values of Ab.. cos3^ ^ i + Asin36 A \ Ac J The parameters of this equation are defined in Fig. CONCLUSION The theoretical study reported in this paper shows that in double diagonal bracing systems the effective length of the compression diagonal is 0. A. the term A sin39A4cinEq. G. 2. In practical situations. ST7. Eq.6 and the l v a l u e is equal to 0. Laval University. sin39 is always smaller than one. and D. as shown in Fig. J. Tension-Compression Cross Bracing Using Star Angles Mechanical Engineering Project Report.391). Kitipornchai. November 1984.Y. 8. S. 1. 1984. IstQtr. 21 was derived for a load pushing against one side of the frame as in Fig. and J. S.. 7). 1. April 1986. Chicago. Pellicione Cross-Bracing Design ASCE Journal of the Structural Division. New York.1. Tests were done to demonstrate the validity of the equations used to determine the transverse stiffness or spring 126 ENGINEERING JOURNAL / AMERICAN INSTITUTE OF STEEL CONSTRUCTION . 1961. 4. and J. McGill University. Moreover.e. 21 is small compared to 1 and can be neglected with only a minor loss in accuracy resulting in errors which are on the conservative side. May 1982. Canadian Standards Association Steel Structures for Buildings—Limit States Design Standard CAN3S16. University of Queensland. The results of these tests are reported in the second part of the paper. CE56. El-Tayem. 4145). Moffet. 105. Specification for the Design. 3 In a typical braced frame. P. Timoshenko. the area of the cross section of the columns Ac is usually much greater than the area of the cross section of the diagonals A.infinity. Department of Civil Engineering. (pp. Montreal. i. and D. Rexdale. ACKNOWLEDGMENTS A part of the results presented in the paper are drawn from the Master's thesis of Pierre Moffet. 8. Gere Theory of Elastic Stability McGraw-Hill Book Co. N. D. P. July 1979 (pp. 5. the AIAb ratio is much smaller than one. 70-73). (pp. By considering the elastic deformations of the braced frame shown in Fig. L. equal to the flexural stiffness of the tension diagonal (T = 0 in Eq. and S. Goel Effective Length Factor for the Design of X-bracing Systems AISC Engineering Journal. III. 1986. f. Department of Civil Engineering. 3. 6. the CIT ratio is usually smaller than 1. 1. 17). Fabrication and Erection of Structural Steel for Buildings 1978.5 (Fig. small values of the AIAb ratio. Department of Civil Engineering. T. tend to equalize the forces in the diagonals.1 + (A) cos > e m Eq. 15 or Eq. Ontario. DeWolf. Beaulieu Resistance au flambement de pieces comprimees supportees en divers points par des pieces tendues Report GCT-06-03 (in French). Vickers3 derived the following equation to compute the CIT ratio: C T = 1 + stiffness provided by the tension diagonal and the validity of the equations used to determine the effective length factor. Inc. 1. 1. M.