Download as pdf or txt
Download as pdf or txt
You are on page 1of 33

1 Effective Buckling Length of Frames with Tapered Columns and Partially Tapered Beams

2 Sherif M. Ibrahim
3 Structural Engineering Department, Ain Shams University, Cairo, Egypt
4 Email: Sherif_ibraim@eng.asu.edu.eg
5

6 Abstract

7 This paper presents stability analysis of steel frames composed of linearly tapered columns

8 and partially tapered restraining beams. In the first part of the stability analysis, closed-form

9 equations for the bending stiffness of the symmetrical and unsymmetrical partially tapered

10 restraining beam are derived and presented in form of design charts. Subsequently, slope deflection

11 equations of linearly tapered column subjected to axial compressive load are then assembled along

12 with the bending stiffness equations of partially tapered beams to determine the elastic critical load

13 and the effective buckling length factor for non-sway and sway steel frames. Numerical examples

14 and efficient design charts are presented for several braced and unbraced steel frames. Moreover, the

15 effect of gravity loads acting on leaning columns connected to sway frames is investigated. Based on

16 this investigation it is concluded that correction of effective buckling length factor based on story

17 buckling concept can be extended to cover steel frames with tapered column as well.

18 Keywords: stability, effective buckling length, steel frames, tapered columns, partially tapered

19 beams

20 1. Introduction

21 Effective buckling length of steel frames is still one of the main design parameters for

22 column design of steel frames in several design specifications such as AISC [1]. Single-story steel

23 frames composed of fully and partially tapered members result in weight saving for many structural

24 and loading situations. Lee et al. [2] conducted one of the earliest elastic buckling analysis studies on

25 individual tapered columns and frames with tapered columns and prismatic beams. For individual

26 linearly web-tapered column with length (L), they introduced a length factor (g) which enabled the

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
27 calculations of elastic critical load (Pcr) of tapered column based on an equivalent prismatic column

28 with the smallest moment of inertia (I1) and a buckling length (gL) as follows:

𝜋 2 𝐸𝐼1
29 𝑃𝑐𝑟 = (𝑔𝐿)2
(1)

30 𝑔 = 1 − 0.375𝛾 + 0.08𝛾 2 (1 − 0.0775𝛾) (2)

ℎ2
31 In which 𝛾 = −1 (3)
ℎ1

32 Where h2 and h1 are the cross-section depth at the largest and the smallest ends, respectively. For

33 steel frames, Lee et al. [2], provided design charts for the effective buckling length factor of tapered

34 columns restrained by prismatic girders using slope deflection equations. Their research was the base

35 for the provisions of tapered members design included in AISC supplement No. 3 [3]. Lee and

36 Morrell [4] showed that the use of AISC supplement No. 3 can be extended to determine the

37 effective buckling length of frames with linearly fully tapered columns and linearly fully tapered

38 beams as well. Hirt and Crisinel [5] presented an expression for the elastic critical load for

39 symmetrical web-tapered pin-ended column about its major axis based on a concept of prismatic

40 column with the same length but with equivalent moment of inertia (Ieq) as follows:

𝜋 2 𝐸𝐼𝑒𝑞
41 𝑃𝑐𝑟 = (4)
𝐿2

42 where,

𝐼
43 𝐼𝑒𝑞 = (0.08 + 0.92 √𝐼1 ) 𝐼2 (5)
2

44 In which I1 and I2 are the smallest and largest moment of inertia of the linearly tapered column,

45 respectively.

46 Marques et al. [6] derived an Ayrton–Perry formulation for columns with non-uniform cross sections

47 and proposed new design rules for tapered columns. They proposed an amendment to column design

48 equations in Eurocode [7] by introducing additional factor that accounts for the second order effects

49 of tapered columns that was later verified by experimental work of Tankova et al. [8]. Nuguyen et al.

50 [9] used artificial neural network (ANN) to predict the critical buckling load of web-tapered

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
51 columns. They used finite element method to generate a wide database set of 269 specimens to

52 develop ANN model. An ANN-based formula was proposed to estimate the critical buckling load of

53 the web-tapered columns considering various input variables.

54 Effective buckling length of columns in steel frames can be determined by performing stability

55 analysis. Formulation of the stability functions based on slope deflection equations for linearly web

56 tapered I-member subjected to axial compressive loading is well documented in the literature [10-

57 13]. Ermopoulos [14] used these equations to develop design charts for the effective buckling length

58 of frames with tapered columns and prismatic beams. El-Sadder [15] derived exact expression for the

59 stability functions of non-prismatic members with solid rectangular, solid circular and I-cross

60 sections. He used power series approach to solve the fourth-order ordinary differential equation with

61 variable coefficients. Raftoyiannis and Ermopoulos [16] studied the elastic stability of web tapered

62 and stepped columns with initial imperfections of parabolic shape according to EC3 [7]. Their

63 formulation was based on the exact solution of the governing differential equations for columns with

64 web-tapered cross sections. King et al. [17] derived a stiffness modification factors for linearly or

65 symmetrical partially tapered solid rectangular girders restraining prismatic columns in braced or

66 unbraced frames. They used these factors along with AISC alignment charts to calculate the effective

67 buckling length for frames restrained by partially tapered girders with solid rectangular cross section.

68 Bazeos and Karabalis [18] introduced an approximate method to calculate the elastic buckling load

69 for tapered columns with different end conditions and restrained by prismatic beams using simple

70 design charts. Their design charts were based on best fitting curves of wide range of web-tapered

71 columns solved by exact formulation that was presented by Karabalis and Bazeos [19]. From these

72 design charts, the location of a critical cross section of the web-tapered column was determined

73 based on the tapering ratio and the column’s end conditions. The critical buckling load was

74 determined based on the moment of inertia of this critical cross section and the well-known buckling

75 length factors. Saffari et al. [20] presented design-oriented charts to calculate the effective buckling

76 length of columns of gabled frames in which both columns and rafters are linearly web-tapered

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
77 members. Their charts were based on stability analysis and utilizing stability functions introduced by

78 Ermopoulos [10]. Serna et al. [21] proposed an approximate procedure to transform tapered column

79 subjected to non-uniform compressive loading into equivalent prismatic column with non-uniform

80 loading. Their approximate procedure has shown its effectiveness for different tapering ratios with

81 errors in the critical buckling load not exceeding 11% of the exact solution. Lu and Meng [22]

82 considered the effect of shear deformations on the stability analysis of steel frames with tapered

83 columns. They derived the exact elements of the stiffness matrix of tapered member from the

84 governing differential equations considering both axial force and shear deformations. They

85 concluded that shear deformation has minimal effect on columns with slenderness ratios greater than

86 50. Salem [23], used the lateral stiffness of unbraced frames with tapered columns to map them to

87 equivalent prismatic columns through closed-form equations. However, his method was limited to

88 simple portal frames with hinged bases.

89 Many researchers [19] and [24-27] developed finite element (FE) formulations to determine the

90 elastic buckling loads for tapered columns with general non-prismatic cross sections. These studies

91 cover various aspects of tapered columns boundary conditions, type of cross section and flexibility of

92 connections. Kucukler and Gardner [28-29] developed a stiffness reduction function for axial and

93 bending strength which can be incorporated into finite element analysis. This stiffness reduction

94 approach is able to consider the spread of plasticity, initial imperfections, and residual stresses.

95 Although these FE formulations are accurate, they are not quick tool for design oriented engineers.

96 Single story steel frames are usually constructed from welded I-sections that have constant flanges

97 and variable web depth. To optimize the frame design, columns are linearly tapered while beams are

98 partially tapered as shown in Fig. 1. A partially tapered beam is a beam that has web-tapered

99 segments connected to another prismatic segments. The available literatures [17] and [30] only

100 address the modification of bending stiffness of partially tapered girders with solid rectangular cross

101 section. Therefore, in this paper, stability analysis of braced and unbraced steel frames composed of

102 linearly tapered columns and partially tapered beams made from I-section is presented. Calculations

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
103 of the bending stiffness of partially tapered symmetrical and unsymmetrical beams are presented in

104 form of closed-form equations. Afterward, this bending stiffness is incorporated into stability

105 analysis of steel frames with linearly tapered columns to develop efficient and simple design charts

106 for the effective buckling length factor of these columns. Moreover, these design charts can be

107 applied to unbraced frames with leaning columns by applying the story buckling approach. The

108 effective buckling length is an important factor that is used to evaluate column available flexural

109 buckling strength using AISC [1] effective length method provided that all the limitations of this

110 method are satisfied.

111 Lb Lb
I2c I2c I2c
112 I2b I2b I2b
I1b I1b
113 a Lb a Lb a Lb

114
Lc
Lc

115
I1c I1c
I1c
116

117 a- Frame with symmetrical partially tapered beam b- Frame with unsymmetrical partially tapered beam

118 Fig.1. Single story frames with partially tapered beams and tapered columns

119 2. Bending stiffness of partially tapered I-beams

120 This section presents the derivation of the bending stiffness of partially tapered I-beam that provide

121 restraining effect to columns in frames permitted to sway or prevented from sway. The ratio between

122 the bending stiffness of partially tapered I-beam and that of prismatic I-beam with the smallest

123 moment of inertia represents a modification factor that can be utilized in the stability analysis of steel

124 frames composed of partially tapered I-beam restraining girders.

125 2.1 Symmetrical partially tapered beams

126 Symmetrical partially tapered I-beam is generally used in girders with rigid ends. Thus, for sway

127 frames, the beam is bent in double curvature while for no sway frames the beam is bent in single

128 curvature [31]. The variation of the moment of inertia of linearly fully tapered I-beam, shown in Fig.

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
129 2 with respect to x-axis origin at virtual point of zero cross section is provided in [10,14 and 32] as

130 follows:

𝑥 2 𝑥 2
131 𝐼𝑥 = 𝐼1𝑏 [𝐴] = 𝐼1𝑏 [(𝑏−1)𝐿 ] (6)
𝑏

𝐼
132 Where, A is the distance from the smallest end to x-axis origin, 𝑏 = √ 2𝑏⁄𝐼 and 𝐼1𝑏 & 𝐼2𝑏 are
1𝑏

133 moment of inertia of the tapered beam at the smallest and the largest ends of the beam, respectively.

134 Similar equations to Eq. (6) can be used for the tapered parts of a symmetrical partially tapered beam

135 but considering the shift in the partially tapered beam x-axis origin shown in Fig. 3 as follows:

𝑥 2
136 𝐹𝑜𝑟 0 ≤ 𝑥 ≤ 𝛼𝐿𝑏 ∶ 𝐼𝑥 = 𝐼1𝑏 [𝑏 + (1 − 𝑏) 𝛼𝐿 ] (7)
𝑏

137 𝐹𝑜𝑟 𝛼𝐿𝑏 ≤ 𝑥 ≤ (1 − 𝛼)𝐿𝑏 : 𝐼𝑥 = 𝐼1𝑏 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 (8)

(1−𝑏) 𝑥 2
138 𝐹𝑜𝑟 (1 − 𝛼)𝐿𝑏 ≤ 𝑥 ≤ 𝐿𝑏 ∶ 𝐼𝑥 = 𝐼1𝑏 [1 − (𝐿 − (1 − 𝛼))] (9)
𝛼 𝑏 y
x
139
x Ix o
140
Lb I1b A
I2b
141

142 Fig. 2. Linearly fully tapered beam coordinate system.

143
y
Lb
144
x
I1b o
145 I2b I2b
a Lb a Lb
146

147 Fig. 3. Symmetrical partially tapered beam coordinate system.

148 2.1.1 Symmetrical partially tapered beams in sway Frames

149 In sway frames both ends of the restraining beam are subjected to equal end moments in the same

150 direction [31] as shown in Fig. 4. The slope moment-rotation equation in this case can be written as

151 follows:

1 𝐿 𝑀 𝐿 𝑀
152 𝜃= (∫0 𝑏 𝐸𝐼𝑥1 𝑥 𝑑𝑥 − ∫0 𝑏 𝐸𝐼𝑥2 𝑥 𝑑𝑥) (10)
𝐿𝑏 𝑥 𝑥

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
153 where, 𝜃 is the beam end rotation.

154 The variation of the bending moments along the beam span can be expressed as:
𝑀 𝑀
155 𝑀𝑥1 = 𝐿 𝑥 𝑎𝑛𝑑 𝑀𝑥2 = (𝐿𝑏 − 𝑥) (11)
𝑏 𝐿𝑏

156 Using Eq. (11) and considering the three segments of the symmetrical partially tapered beam, Eq.

157 (10) can be rewritten as:

𝑀 𝛼𝐿𝑏 𝑥2 (1−𝛼)𝐿𝑏 𝑥2 𝑏 𝐿 𝑥2 𝛼𝐿𝑏 (𝐿𝑏 −𝑥) 𝑥


158 𝜃= (∫0 𝑑𝑥 + ∫𝛼𝐿 𝑑𝑥 + ∫(1−𝛼)𝐿 𝑑𝑥 − ∫0 𝑑𝑥 −
𝐿𝑏 𝐸𝐿𝑏 𝐼𝑥 𝑏 𝐸𝐿𝑏 𝐼1𝑏 𝑏 𝐸𝐿𝑏 𝐼𝑥 𝐸𝐿𝑏 𝐼𝑥

(1−𝛼)𝐿𝑏 (𝐿𝑏 −𝑥) 𝑥 𝑏𝐿 (𝐿𝑏 −𝑥) 𝑥


159 ∫𝛼𝐿 𝑑𝑥 − ∫(1−𝛼)𝐿 𝑑𝑥 ) (12)
𝑏 𝐸𝐿𝑏 𝐼1𝑏 𝑏 𝐸𝐿𝑏 𝐼𝑥

160 Substituting Eq. (7) and (9) into Eq. (12) and performing the integration, the moment-rotation

161 relationship can be expressed as follows:


6 𝐸𝐼1𝑏
162 𝑀= 𝛽1𝑠 𝜃 (13)
𝐿𝑏

1 12𝛼3 𝛼+(1−𝛼)(1−𝑏) {𝛼+(1−𝛼)(1−𝑏)}2


163 where, 𝛽 = {(1−𝑏)3 [2(1 − 𝑏) + (2𝑏 + 2 ) ln(𝑏) + (1 − 𝑏)(𝑏 + ]+
1𝑠 𝛼 𝑏𝛼2

6𝛼2 𝛼+(1−𝛼)(1−𝑏)
164 (1−𝑏)
[1 − ] + 1 − 6𝛼 + 12𝛼 2 − 8 𝛼 3 } (14)
𝑏𝛼

165 Lb

166 M

167 M
x y
168
M
169 x Mx2
Mx1
170 M

171 Fig. 4. Symmetrical partially tapered beam moment-rotation diagram in sway frames.

172 The term 𝛽1𝑠 , represents a modification factor that can be multiplied by the bending stiffness of

173 similar prismatic beam having the smallest moment of inertia (I1b) to obtain the actual bending

174 stiffness of symmetrical partially tapered I-beam which is required for stability analysis of frames

175 permitted to sway. The value of this modification factor is plotted in Fig. 5 for different tapered

176 length ratios ( 𝛼) and different (I2b / I1b) values . The obtained bending stiffness of symmetrical

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
177 partially tapered restraining beam can be used along with the ASIC [1] alignment charts to determine

178 the effective buckling length of prismatic columns in sway permitted frames.

6.5
M Lb
6.0
I2b I1b I2b
5.5
a Lb a Lb
5.0 𝟔𝑬𝑰𝟏𝒃
M= 𝜷𝟏𝒔 𝜽 M
𝑳𝒃
4.5
4.0
β1s

3.5
3.0
2.5
2.0
1.5
1.0
0 0.1 0.2 0.3 0.4 0.5
179
a

180 Fig. 5. Bending stiffness modification factor for symmetrical partially tapered I-beam restraining sway frames.
181
182 2.1.2 Symmetrical partially tapered beams in non-sway frames

183 For frames prevented from sway, the restraining beam is subjected to uniform and constant bending

184 moment (M) along its span as it is bent in single curvature [31]. The moment-rotation equation in

185 this case can be written as follows:

1 𝐿 𝑀 𝑀 𝛼𝐿𝑏 𝑥 (1−𝛼)𝐿𝑏 𝑥 𝐿 𝑥
186 𝜃= (∫0 𝑏 𝐸𝐼 𝑥 𝑑𝑥 ) = 𝐿 (∫0 𝑑𝑥 + ∫𝛼𝐿 𝑏
𝑑𝑥 + ∫(1−𝛼)𝐿 𝑑𝑥 ) (15)
𝐿𝑏 𝑥 𝑏 𝐸𝐼𝑥 𝑏 𝐸𝐼1𝑏 𝑏 𝐸𝐼𝑥

187 The integration in Eq. (15) can be solved considering that (M) is constant and substituting with

188 moment of inertia of the beam from Eqs. (7) to (9) for the three parts of the beam. The moment-

189 rotation relationship can be expressed as follows:


2 𝐸𝐼1𝑏
190 𝑀= 𝛽1𝑛𝑠 𝜃 (16)
𝐿𝑏

1 2𝛼2 𝛼+(1−𝛼)(1−𝑏)
191 where, = {1 − 2𝛼 + (1−𝑏) [ − 1]} (17)
𝛽1𝑛𝑠 𝑏𝛼

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
192 Fig. 6 illustrates the variation of the bending stiffness modification factor for symmetrical partially

193 tapered restraining beam for frames prevented from sway in terms of the tapered length ratio (𝛼) and

194 the ratio of the largest to smallest moment of inertia (I2b / I1b) of the beam.

3.5
Lb

3.0 I2b I1b


I2b
a Lb a Lb
𝟐𝑬𝑰𝟏𝒃
2.5 M M= 𝜷𝟏𝒏𝒔 𝜽 M
𝑳𝒃
b1ns

2.0

1.5

1.0
0 0.1 0.2 0.3 0.4 0.5
a
195

196 Fig. 6. Bending stiffness modification factor for symmetrical partially tapered I-beam restraining no sway frames.
197
198 2.2 Unsymmetrical partially tapered beams with hinged far end

199 Unsymmetrical partially tapered I-beam is generally used in girders with rigid connection at the

200 largest end and hinged connection at the smallest end. Whether used as restraining beam in sway

201 permitted or sway prevented frames, unsymmetrical partially tapered I-beam is bent with single

202 curvature [31]. The variation of the moment of inertia of unsymmetrical partially tapered I-beam

203 along the x-axis shown in Fig. 7 can be written as follows:

204 𝐹𝑜𝑟 0 ≤ 𝑥 ≤ (1 − 𝛼)𝐿𝑏 : 𝐼𝑥 = 𝐼1𝑏 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 (18)

(1−𝑏) 𝑥 2
205 𝐹𝑜𝑟 (1 − 𝛼)𝐿𝑏 ≤ 𝑥 ≤ 𝐿𝑏 ∶ 𝐼𝑥 = 𝐼1𝑏 [1 − (𝐿 − (1 − 𝛼))] (19)
𝛼 𝑏

206 The variation of the bending moment along the beam span with respect to x-axis is given by:
𝑀
207 𝑀𝑥 = 𝐿 𝑥 (20)
𝑏

208 The rotation at the rigid beam end can be determined from the following relationship:
Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
1 (1−𝛼)𝐿𝑏 𝑀 𝐿 𝑀
209 𝜃= (∫0 𝑥 2 𝑑𝑥 + ∫(1−𝛼)𝐿
𝑏
𝑥 2 𝑑𝑥 ) (21)
𝐿𝑏 𝐸𝐼1𝑏 𝑏 𝐸𝐼𝑥

210 The above integration can be carried out by substituting the moment of inertia for the tapered

211 segment from Eq. (19) and considering that M and I1b are constants. The moment-rotation

212 unsymmetrical partially tapered I-beam with hinged far end at the smallest end can be written as

213 follows:
6 𝐸𝐼1𝑏 2 𝐸𝐼1𝑏
214 𝑀= 𝛽2𝑠 𝜃 = 𝛽2𝑛𝑠 𝜃 (22)
𝐿𝑏 𝐿𝑏

1 3 3𝛼3 𝛼+(1−𝛼)(1−𝑏) {𝛼+(1−𝛼)(1−𝑏)}2


215 where, 𝛽 = = 2 {(1−𝑏)3 [(2 ) ln(𝑏) + (1 − 𝑏) + (1 − 𝑏)] +
2𝑠 𝛽 2𝑛𝑠 𝛼 𝑏𝛼2

216 (1 − 𝛼)3 } (23)

217 y
Lb
218
x
219 I1b o
I2b
Hinged end
220 M a Lb

221

222 Fig. 7. Unsymmetrical partially tapered beam coordinate system.

223 𝛽2𝑠 𝑎𝑛𝑑 𝛽2𝑛𝑠 represent modification factors that can be multiplied by the bending stiffness of

224 similar prismatic beam having the smallest moment of inertia (I1b) to obtain the actual bending

225 stiffness of unsymmetrical partially tapered I-beam with hinged end which is required for stability

226 analysis of frames permitted to sway and prevented from sway, respectively. Fig. 8, shows these

227 modification factors for various tapered length ratio (𝛼) and various values of (I2b / I1b).

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
3.5 Lb
Hinged End
I2b I1b
3.0
a Lb 𝟔𝑬𝑰𝟏𝒃
M= 𝜷𝟐𝒔 𝜽
M 𝑳𝒃
𝟐𝑬𝑰𝟏𝒃
2.5 M= 𝜷𝟐𝒏𝒔 𝜽
b2s = b2ns /3

𝑳𝒃

2.0

1.5

1.0

0.5
0 0.2 0.4 a 0.6 0.8 1
228

229 Fig. 8. Bending stiffness modification factors for unsymmetrical partially tapered I-beam with hinged smaller end
230 restraining sway and no sway frames.
231 2.3 Applications on frames with partially tapered beams and prismatic columns

232 To verify and illustrate the use of the derived bending stiffness modification factors for partially

233 tapered beams, the effective buckling lengths of columns in three steel portal frames shown in Fig. 9

234 are calculated. The cross-section dimensions of beams and columns are listed in Table 1. The

235 calculations are performed using the basic equations of alignment charts instead of reading values

236 from the charts for accuracy reasons. The basic equations to determine the effective buckling length

237 of column in frames permitted to sway or prevented from sway are given in [31] and [33]. These

238 equations can be simplified in case of columns with hinged and fixed bases by substituting 𝐺𝐴 equals

239 to infinity and zero, respectively. Thus, the alignment charts equations can be rewritten for these

240 special cases as follows:

𝜋 𝜋 6
241 For sway frames with hinged base: (𝑘 ) tan (𝑘 ) − =0 (24)
𝐺𝐵

𝜋 𝜋
242 For sway frames with fixed base: (𝑘 ) 𝐺𝐵 + 6 tan (𝑘 ) = 0 (25)

𝜋 2 𝜋 𝜋
243 For no sway frames with hinged base: 0.5 (𝑘 ) 𝐺𝐵 + 1 − (𝑘 ) / tan (𝑘 ) = 0 (26)

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
𝐼𝑐 /𝐿𝑐
244 where, k is the effective buckling length factor of column and 𝐺𝐵 = in which Ic is the
𝛽𝑠 𝐼1𝑏 /𝐿𝑏

245 column moment of inertia, Lc is the column length, 𝛽𝑠 represents the relevant bending stiffness

246 modification factor obtained from Eqs. (14), (17) & (23) for partially tapered beam, I1b is the

247 smallest moment of inertia of the beam and Lb is the beam length. Eqs. (24) to (26) can be solved to

248 determine the value of k using iteration technique for any given value of 𝐺𝐵 . The three frames shown

249 in Fig. 9 are solved for different tapered length ratio, 𝛼, for the restraining beam and results are

250 compared with those obtained from linear buckling analysis (LBA) using STAAD.Pro software [34].

251 A numerical model for each of the three portal frames shown in Fig. 9 was constructed using

252 prismatic beam element for columns and the uniform part of the beam while a web-tapered beam

253 element is used for the tapered portion of the beam. An initial gravity load is applied on each

254 column. STAAD.Pro software [34] performs eigenvalue calculations to get buckling factor (BF)

255 which is the amount by which the initially applied loads need to be multiplied by to just cause

256 buckling (i.e. elastic buckling load = Pcr = BF × initial load). Eigenvalue procedure can be

257 summarized in the following steps:

258 1- First, the primary deflections are calculated by linear static analysis based on the provided

259 initial loading.

260 2- Primary deflections are used to calculate member axial forces. These forces are used to

261 calculate geometric stiffness matrix [Kg]. Both the large deformations effects and the small

262 deformations effects for members are calculated.

263 3- An eigenvalue problem is formed, according to the following equation:

264 |[𝐾] − 𝐵𝐹 × [𝐾𝑔 ]| = 0 , where [K] is the global stiffness matrix.

265 In LBA, the output of the analysis is the elastic critical load 𝑃𝑐𝑟 , which can be used to determine the

266 effective length factor as follows:

𝑃
267 𝑘 = √𝑃 𝑒 (27)
𝑐𝑟

𝜋 2 𝐸𝐼𝑐
268 In which 𝑃𝑒 = 𝐸𝑢𝑙𝑒𝑟 𝑙𝑜𝑎𝑑 = , where E is the modulus of elasticity.
𝐿2𝑐
Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
269 Lb = 20 m Lb = 20 m
Ic
Ic Ic
I2b I2b I2b
270
I1b I1b
a Lb a Lb a Lb
271

Lc = 8 m
Lc = 8 m

272

273 Ic Ic
Ic
274

275 a- Portal frame 1 b- Portal frame 2

276 Lb = 20 m
Ic Ic
277 I2b I2b
I1b
a Lb a Lb
278
Lc = 8 m

279

280
Ic Ic
281
282 c- Portal frame 3

283 Fig. 9. Dimensions of steel portal frames (1 to 3) with prismatic columns and partially tapered beams.

284 The results from alignment charts equations based on the modified bending stiffness of partially

285 tapered beam and those from LBA for the column effective length factor of the three steel portal

286 frames are provided in Tables 2 to 4. It is evident from these tables that the proposed modification

287 factors for partially tapered beam along with alignment charts equations provide accurate estimate

288 for the effective length factor compared to the results from LBA as the error percent is less than 1.7

289 % for case of frames permitted to sway and 2.4 % for frames prevented from sway. In order to

290 investigate the effect of partial tapering of beams on the effective length factor, the ratio between k-

291 factor at any given tapered length ratio (𝛼) to k-factor at tapered length ratio = 0, is indicated in

292 Tables 2 to 4. Frames with prismatic beams having the smallest cross section properties (i.e., tapered

293 length ratio = 𝛼 = 0) represents the ideal case where effective buckling length factors can be obtained

294 directly from AISC alignment charts. It is evident from Tables 2 to 4 that partial tapering in beam

295 has significant effect on reducing the effective buckling length factor for case of frames permitted to
Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
296 sway (e.g. frames 1 and 2 for which maximum reduction in k-factor can reach 33% compared to the

297 similar frame with prismatic beam having the smallest cross section property, I1b) while this effect is

298 minimal for braced frames (e.g. frame 3 for which maximum reduction in k-factor is only 4%

299 compared to similar frame with prismatic beam).

300 Table 1
301 Dimensions and properties of portal frames 1 to 3 cross sections
Cross section Web dim. Flanges dim. Moment of inertia E (MPa)

(mm) (mm) (mm4)

Beam cross section 1 300 × 6 150 × 10 𝐼1𝑏 = 85.6 × 106 2.05 × 105

Beam cross section 2 700 × 6 150 × 10 𝐼2𝑏 = 549.6 × 106 2.05 × 105

Column cross section 400 × 6 200 × 10 𝐼𝑐 = 200.13 × 106 2.05 × 105

302
303 Table 2
304 k-factor for frame 1 (symmetrical partially tapered beam)- sway permitted-hinged base
𝛼 𝛽1𝑠 𝐺𝐵 K LBA Error 𝑘𝛼 /𝑘𝛼=0

(Eq. 14) (Eq. 24) (STAAD.Pro) percent (Eq. 24)

Pcr (kN) k

0 1 5.84 3.62 481 3.63 -0.27 1

0.2 2.08 2.81 2.81 745 2.91 -1.66 0.78

0.4 3.69 1.58 2.51 972 2.55 -1.62 0.69

0.5 4.30 1.36 2.44 1033 2.470 -1.34 0.67

305
306 Table 3
307 k-factor for frame 2 (unsymmetrical partially tapered beam) – sway permitted-fixed base
𝛼 𝛽2𝑠 𝐺𝐵 K LBA Error 𝑘𝛼 /𝑘𝛼=0

(Eq. 23) (Eq. 25) (STAAD.Pro) percent (Eq. 25)

Pcr (kN) k

0 0.5 11.69 1.71 2172 1.71 -0.10 1

0.2 0.72 8.09 1.62 2379 1.63 -0.53 0.95

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
0.4 1.04 5.62 1.53 2639 1.55 -1.07 0.89

0.6 1.44 4.06 1.45 2926 1.47 -1.51 0.85

0.8 1.85 3.17 1.39 3188 1.41 -1.64 0.81

1.0 2.15 2.72 1.34 3367 1.37 -1.61 0.78

308
309
310 Table 4
311 k-factor for frame 3 (symmetrical partially tapered beam)- sway prevented-hinged base
𝛼 𝛽1𝑛𝑠 𝐺𝐵 K LBA Error 𝑘𝛼 /𝑘𝛼=0

(Eq. 17) (Eq. 26) (STAAD.Pro) percent (Eq. 26)

Pcr (kN) k

0 1 5.84 0.969 6519 0.985 -1.68 1

0.2 1.32 4.43 0.960 6625 0.977 -1.79 0.99

0.4 1.94 3.01 0.944 6813 0.963 -2.03 0.97

0.5 2.53 2.31 0.930 6963 0.954 -2.38 0.96

312

313 3. Slope deflection equations of linearly tapered columns

314 The variation of the moment of inertia for a linearly tapered column can be written using equation

315 analogous to Eq. (6), but with the column notations as follows:

𝑥 2 𝑥 2
316 𝐼𝑥 = 𝐼1𝑐 [𝐴] = 𝐼1𝑐 [(𝑐−1)𝐿 ] (28)
𝑐

𝐼
317 Where, A is the distance from the smallest end to x-axis origin, 𝑐 = √ 2𝑐⁄𝐼 and 𝐼1𝑐 & 𝐼2𝑐 are
1𝑐

318 moment of inertia of the tapered column at the smallest and the largest ends, respectively.

319 The slope deflection equations for linearly tapered member shown in Fig. 10 in the presence of axil

320 compressive force are well documented in [13-14] and can be written in terms of the smallest

321 moment of inertia (I1c) as follows:

2𝐸𝐼1𝑐 𝛿
322 𝑀𝑎 = [𝛼1 𝜃𝑎 + 𝛼2 𝜃𝑏 − (𝛼1 + 𝛼2 ) 𝐿 ] (29)
𝐿𝑐 𝑐

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
2𝐸𝐼1𝑐 𝛿
323 𝑀𝑏 = [𝛼3 𝜃𝑎 + 𝛼4 𝜃𝑏 − (𝛼3 + 𝛼4 ) 𝐿 ] (30)
𝐿𝑐 𝑐

2𝐸𝐼1𝑐 𝛿
324 𝑉 𝐿𝑐 = [−(𝛼1 + 𝛼3 ) 𝜃𝑎 − (𝛼2 + 𝛼4 ) 𝜃𝑏 + (𝛼1 + 𝛼2 + 𝛼3 + 𝛼4 ) 𝐿 ] − 𝑃 𝛿 (31)
𝐿𝑐 𝑐

325 The last term in Eq. (31) represents the (𝑃 − 𝛿) effect on columns in sway frames. For completeness,

326 the other coefficient in Eqs. (29) to (31) are given by the following equations [14]:

(𝑓2 −𝑓4 )(𝜇 2 +0.25) 2


327 𝛼1 = ℓ (32)
2𝑓0

𝑓4 (𝜇 2 +0.25) 2
328 𝛼2 = 𝛼3 = ℓ (33)
2𝑓0

{−𝑓4 cos[𝜇 ln(1+ℓ)]+𝑓3 𝑠𝑖𝑛[𝜇 ln(1+ℓ)]}(𝜇 2 +0.25) 2


329 𝛼4 = ℓ √1 + ℓ (34)
2𝑓0

330 where


331 𝑓1 = {0.5 cos[𝜇 ln(1 + ℓ)] − 𝜇 sin[𝜇 ln(1 + ℓ)]} − 0.5 ℓ (35)
√1+ℓ


332 𝑓2 = {𝜇 cos[𝜇 ln(1 + ℓ)] − 0.5 sin[𝜇 ln(1 + ℓ)]} − 𝜇 ℓ (36)
√1+ℓ

333 𝑓3 = √1 + ℓ cos[𝜇 ln(1 + ℓ)] − 0.5ℓ − 1 (37)

334 𝑓4 = √1 + ℓ sin[𝜇 ln(1 + ℓ)] − 𝜇ℓ (38)

335 𝑓0 = 𝑓1 𝑓4 − 𝑓2 𝑓3 (39)

𝐿𝑐
336 ℓ= =𝑐−1 (40)
𝐴

𝑃𝐴2 𝑃 𝐿2
337 𝜇 = √𝐸 𝐼 − 0.25 = √(𝑐−1)2𝑐𝐸 𝐼 − 0.25 > 0 (41)
1𝑐 1𝑐

338 Rearranging Eq. (41), axial load P can be written as:

(𝜇 2 +0.25)(𝑐−1)2 𝐸𝐼1𝑐
339 𝑃= (42)
𝐿2𝑐

340 Substituting Eq. (42) into Eq. (31) results in the following equation:

2𝐸𝐼1𝑐 (𝜇 2 +0.25)(𝑐−1)2 𝛿
341 𝑉 𝐿𝑐 = [−(𝛼1 + 𝛼3 ) 𝜃𝑎 − (𝛼2 + 𝛼4 ) 𝜃𝑏 + (𝛼1 + 𝛼2 + 𝛼3 + 𝛼4 − )𝐿 ] (43)
𝐿𝑐 2 𝑐

342

343

344

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
345

346 x y

347 b Ix a
x o
I2c I1c
348 Lc A

349 a- Geometry of linearly tapered column


Mb
350 P b
V
351 𝜽𝒃
𝜹 𝜽𝒂
x V P
352 a
Ma
353 b- Deformed shape of linearly tapered column
354 Fig. 10. Geometry and deformed shape of linearly tapered column
355
356
357
358 4. Stability analysis of steel frames

359 A number of assumptions, similar to some of those used to develop AISC [1] alignment charts, are

360 adopted in the current frame stability analysis and they are listed as follows:

361 1- Behavior is purely elastic.

𝑃
362 2- The tapered columns stiffness parameter 𝐿𝑐 √𝐸𝐼 is equal for all columns.
1𝑐

363 3- In frames prevented from sway, rotations at opposite ends of restraining beams are equal in

364 magnitude and opposite in direction.

365 4- In frames permitted to sway, rotations at opposite ends of restraining beams are equal in

366 magnitude and direction.

367 5- All columns buckle simultaneously.

368 6- Axial compression force in restraining beams is negligible.

369 7- Shear deformations are neglected.

370 The other assumptions adopted in AISC alignment charts of constant members (for beams and

371 columns) cross sections and rigid end connections are not applicable since the current analysis takes

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
372 into considerations: a) variation in both columns and beams cross sections, and b) the effect of

373 hinged connection at far end of unsymmetrical partially tapered beam.

374 Based on the above assumptions, a sub-assemblage consists of the tapered column and the partially

375 tapered beams at both ends of the column can be analyzed and the equilibrium equations can be

376 written as illustrated in the following sections.

377 4.1 No sway frames

378 In frames prevented from sway, columns are restrained at both ends by restraining beams or by the

379 base support. In case that the restraining beam is partially tapered, the bending moment at the beam

380 end connected to the column can be obtained from Eqs. (16) and (22) according to the beam type.

381 For column prevented from sway, only Eqs. (29) and (30) are used to calculate the sum of bending

𝛿
382 moment at each end of the column and the term (𝐿 ) equals zero. Thus, for a column prevented from
𝑐

383 sway the sum of bending moments at both ends, (a and b), of the column can be written as follows:

2𝐸𝐼1𝑐 (∑ 𝛽𝑛𝑠 𝐼1𝑏 /𝐿𝑏 )𝑎 2𝐸𝐼1𝑐 1


384 ∑ 𝑀𝑎 = [(𝛼1 + ) 𝜃𝑎 + 𝛼2 𝜃𝑏 ] = [(𝛼1 + 𝐺 ) 𝜃𝑎 + 𝛼2 𝜃𝑏 ] = 0 (44)
𝐿𝑐 𝐸𝐼1𝑐 /𝐿𝑐 𝐿𝑐 𝐴

2𝐸𝐼1𝑐 (∑ 𝛽𝑛𝑠 𝐼1𝑏 /𝐿𝑏 )𝑏 2𝐸𝐼1𝑐 1


385 ∑ 𝑀𝑏 = [ 𝛼3 𝜃𝑎 + (𝛼4 + ) 𝜃𝑏 ] = [ 𝛼3 𝜃𝑎 + (𝛼4 + 𝐺 ) 𝜃𝑏 ] = 0 (45)
𝐿𝑐 𝐸𝐼1𝑐 /𝐿𝑐 𝐿𝑐 𝐵

𝐸𝐼1𝑐 /𝐿𝑐 𝐸𝐼1𝑐 /𝐿𝑐


386 where, 𝐺𝐴 = (∑ 𝛽𝑛𝑠 𝐼1𝑏 /𝐿𝑏 )𝑎
, 𝐺𝐵 = (∑ 𝛽𝑛𝑠 𝐼1𝑏 /𝐿𝑏 )𝑏
and (∑ 𝛽𝑛𝑠 𝐼1𝑏 /𝐿𝑏 )𝑎 & (∑ 𝛽𝑛𝑠 𝐼1𝑏 /𝐿𝑏 )𝑏 are half

387 the sum of the bending stiffness of partially tapered beams connected to column ends (a) and (b),

388 respectively. 𝛽𝑛𝑠 depends on the type of the partially tapered beam and is determined from Eqs. (17)

389 and (23).

390 Column buckling occurs when the following condition is satisfied:

1
(𝛼1 + 𝐺 ) 𝛼2
𝐴
391 | 1
| =0 (46)
𝛼3 (𝛼4 + 𝐺 )
𝐵

392 Eq. (46) can be furtherly simplified in case that the column has hinged or fixed base at point (a). For
1
393 a column with hinged base the term 𝐺 = 0 and Eq. (46) can be rewritten as:
𝐴

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
𝛼1 𝛼2
394 |𝛼 (𝛼4 +
1
)| = 0 (47)
3 𝐺𝐵

395 For a column with fixed base, 𝜃𝑎 = 0, and only Eq. (45) is used to determine the buckling conditions

396 as follows:

1
397 (𝛼4 + 𝐺 ) = 0 (48)
𝐵

398 Different values of parameter (𝜇) are iterated till the results of Eqs. (46) to (48) are equal to zero for

399 given values of parameters (c) and (𝐺𝐵 ) which account for the tapering in the column and partial

400 tapering in restraining beams, respectively. Once the critical value of parameter (𝜇) is obtained, the

401 critical buckling load is determined from Eq. (42) as follows:


2 +0.25)(𝑐−1)2
(𝜇𝑐𝑟 𝜋 2 𝐸𝐼1𝑐
402 𝑃𝑐𝑟 = (49)
𝜋2 𝐿2𝑐

403 where 𝜇𝑐𝑟 is the value of parameter (𝜇) which satisfies Eqs. (46) to (48).

404 Therefore, effective buckling length factor, k, which can be used along with the smallest moment of

405 inertia of the column is calculated as follows:


𝜋
406 𝑘= (50)
2 +0.25)
(𝑐−1) √(𝜇𝑐𝑟

407 4.2 Sway frames

408 For linearly tapered column permitted to sway, Eqs. (29), (30) and (43) are used along with the Eqs.

409 (13) and (22) of the partially tapered restraining beams to write the equilibrium equations. The

410 general equation for the critical buckling load of column in sway frame can be written as follows:

3
(𝛼1 + 𝐺 ) 𝛼2 −(𝛼1 + 𝛼2 )
𝐴
| 3 |
411 𝛼3 (𝛼4 + 𝐺 ) −(𝛼3 + 𝛼4 ) =0 (51)
| 𝐵 |
(𝜇 2 +0.25)(𝑐−1)2
−(𝛼1 + 𝛼2 ) −(𝛼3 + 𝛼4 ) (𝛼1 + 𝛼2 + 𝛼3 + 𝛼4 − )
2

𝐸𝐼1𝑐 /𝐿𝑐 𝐸𝐼1𝑐 /𝐿𝑐


412 where 𝐺𝐴 = (∑ 𝛽𝑠 𝐼1𝑏 /𝐿𝑏 )𝑎
, 𝐺𝐵 = (∑ 𝛽𝑠 𝐼1𝑏 /𝐿𝑏 )𝑏
and (∑ 𝛽𝑠 𝐼1𝑏 /𝐿𝑏 )𝑎 & (∑ 𝛽𝑠 𝐼1𝑏 /𝐿𝑏 )𝑏 are one sixth

413 of the sum of the bending stiffness of partially tapered beams connected to column ends (a) and (b)

414 respectively. 𝛽𝑠 depends on the type of the partially tapered beam and is determined from Eqs. (14)

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
1
415 and (23). For case of a column hinged at end (a), the term 𝐺 = 0 and Eq. (51) can be simplified as
𝐴

416 follows:

𝛼1 𝛼2 −(𝛼1 + 𝛼2 )
3
417 || 𝛼3 (𝛼4 + 𝐺 ) −(𝛼3 + 𝛼4 ) || = 0 (52)
𝐵
(𝜇 2 +0.25)(𝑐−1)2
−(𝛼1 + 𝛼2 ) −(𝛼3 + 𝛼4 ) (𝛼1 + 𝛼2 + 𝛼3 + 𝛼4 − )
2

418 For case of a column fixed at end (a), 𝜃𝑎 = 0 , Eq. (51) can be simplified as follows:

3
(𝛼4 + 𝐺 ) −(𝛼3 + 𝛼4 )
𝐵
419 | (𝜇 2 +0.25)(𝑐−1)2
|=0 (53)
−(𝛼3 + 𝛼4 ) (𝛼1 + 𝛼2 + 𝛼3 + 𝛼4 − )
2

420 Depending on the boundary conditions and the stiffness of the restraining beams at both ends of

421 column (GA & GB) and the tapering column parameter (c) Eqs. (51) to (53) can be solved using

422 iteration technique to obtain the critical buckling load and the effective buckling length factor using

423 Eqs. (49) and (50), respectively.

424 4.3 Examples and Design Charts

425 4.3.1 Example 1

426 Three portal frames with linearly tapered columns and partially tapered beams as shown in Fig. 11

427 are analyzed by stability analysis presented in the previous sections and results are compared with

428 LBA using STAAD.Pro software [34]. A numerical model for each of the three portal frames shown

429 in Fig. 11 was constructed using prismatic beam element for the constant part of the beam while a

430 web-tapered beam element is used for the columns and the tapered portion of the beam. Eigenvalue

431 analysis previously described in Section 2.3 is carried out to determine the elastic critical load for

432 each frame. The cross-section dimensions of beams and columns are listed in Table 5. It is evident

433 Tables 6 to 8 that the proposed modification factor for partially tapered beam along with stability

434 analysis of linearly tapered columns provide accurate estimate for the effective length factor

435 compared to the results from LBA with error percent less than 2.7 % for case of frames permitted to

436 sway and 0.34 % for frames prevented from sway. The partial tapering of beam has substantial effect

437 on the effective length factor for the case of frames permitted to sway with tapered columns (e.g.,
Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
438 frames 4 and 5) as the k-factor can be as low as 60 % of the k-factor of similar frames with prismatic

439 beam having the smallest cross section properties as indicated in Tables 6 and 7. On the other hand,

440 as evident from Table 8 the effect of partial tapering of beam on the k-factor for braced frame 6 has a

441 marginal reduction on k-factor by 2.3 % at most compared to similar frame with prismatic beam

442 having the smallest cross section properties.

443

444 Lb = 20 m Lb = 20 m
I2c
I2c I2c
445 I2b I2b
I2b
I1b
a Lb I1b
446 a Lb
a Lb
Lc = 8 m

Lc = 8 m
447

448
I1c I1c
I1c
449

450 a- Portal frame 4 b- Portal frame 5

451 Lb = 20 m
I2c
I2c
452 I2b I2b
I1b
453 a Lb a Lb
Lc = 8 m

454

455
I1c
I1c
456

457 c- Portal frame 6

458 Fig. 11. Dimensions and properties of steel portal frames with partially tapered beams and tapered columns.

459 Table 5
460 Dimensions and properties of portal frames 4 to 6 cross sections
Cross section Web dim. Flanges dim. Moment of inertia E (MPa)

(mm) (mm) (mm4)

Beam cross section 1 300 × 6 150 × 10 𝐼1𝑏 = 85.6 × 106 2.05 × 105

Beam cross section 2 700 × 6 150 × 10 𝐼2𝑏 = 549.6 × 106 2.05 × 105

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
Column cross section 1 400 × 6 200 × 10 𝐼1𝑐 = 200.13 × 106 2.05 × 105

Column cross section 2 600 × 6 200 × 10 𝐼2𝑐 = 480.13 × 106 2.05 × 105

461
462
463
464 Table 6
465 k-factor for frame 4 (symmetrical partially tapered beam)- sway permitted-hinged base
𝛼 𝛽1𝑠 𝐺𝐵 K LBA Error 𝑘𝛼 /𝑘𝛼=0

(Eq. 14) (Eq. 52) (STAAD.Pro) percent (Eq. 52)

Pcr (kN) k

0 1 5.84 3.37 552.7 3.38 -0.39 1

0.2 2.08 2.81 2.54 942.8 2.59 -2.10 0.75

0.4 3.69 1.58 2.11 1353.5 2.16 -2.59 0.63

0.5 4.30 1.36 2.03 1479.7 2.07 -1.86 0.6

466
467 Table 7
468 k-factor for frame 5 (unsymmetrical partially tapered beam) – sway permitted-fixed base
𝛼 𝛽2𝑠 𝐺𝐵 K LBA Error 𝑘𝛼 /𝑘𝛼=0

(Eq.23) (Eq. 53) (STAAD.Pro) percent (Eq. 53)

Pcr (kN) k

0 0.5 11.69 1.55 2564.5 1.57 -1.22 1

0.2 0.72 8.09 1.49 2765.6 1.51 -1.59 0.96

0.4 1.04 5.62 1.41 3033.2 1.44 -2.06 0.91

0.6 1.44 4.06 1.34 3347.6 1.37 -2.47 0.86

0.8 1.85 3.17 1.28 3648.4 1.32 -2.68 0.83

1.0 2.15 2.72 1.24 3871.1 1.28 -2.66 0.8

469
470
471
472
473
474
475

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
476 Table 8
477 k-factor for frame 6 (symmetrical partially tapered beam)- sway prevented-hinged base
𝛼 𝛽1𝑛𝑠 𝐺𝐵 K LBA Error 𝑘𝛼 /𝑘𝛼=0

(Eq. 17) (Eq. 47) (STAAD.Pro) percent (Eq. 47)

Pcr (kN) k

0 1 5.84 0.782 10275 0.785 -0.36 1

0.2 1.32 4.43 0.778 10381.3 0.781 -0.36 0.995

0.4 1.94 3.01 0.771 10575 0.773 -0.38 0.986

0.5 2.53 2.31 0.764 10750 0.767 -0.43 0.977

478

479 4.3.2 Example 2

480 The gabled steel frame shown in Fig. 12 was analyzed by Saffari et al. [20]. They calculated the

481 column buckling length factor (km) based on the moment of inertia of a cross section located at the

482 middle of the column height (Im) for special case of symmetrical partially tapered beam when 𝛼 =

483 0.5. This value of (km) can be transferred to a buckling length factor (k) based on the moment of

484 inertia of the smallest cross section from the following equation:

𝜋 2 𝐸𝐼𝑚 𝜋 2 𝐸𝐼𝑚 𝐼 𝜋 2 𝐸𝐼1𝑐 𝜋 2 𝐸𝐼1𝑐


485 𝑃𝑐𝑟 = = × 𝐼1𝑐 = 𝐼 = (54)
(𝑘𝑚 𝐿𝑐 )2 (𝑘𝑚 𝐿𝑐 )2 1𝑐 𝑘𝑚 × 1𝑐 ×𝐿2𝑐
2 𝑘 2 ×𝐿2𝑐
𝐼𝑚

𝐼 𝑘
486 Therefore, 𝑘 = √ 𝐼1𝑐 𝑘𝑚 = [1+(𝑐−1)/2]
𝑚
(55)
𝑚

487 The length of the restraining beam of the gabled frame is taken as the actual total inclined length of

488 the rafter. The bending stiffness modification factor is calculated from Eq. (14) and considering the

489 tapered length ratio 𝛼 = 0.5. Table 9 shows comparison between the buckling length factor obtained

490 by the current analysis and results from Saffari et al. [20]. It is concluded that the current proposed

491 analysis result for the buckling length factor is almost identical to those of Saffari et al. [20] results

492 for hinged and fixed base conditions.

493

494
Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
495

H = 2.31 m
I1b
496 I2c
I2b I2b I2c
497
Relative values of moment of inertia
498
Lc = 4 m

𝐼1𝑐 = 1, 𝐼2𝑐 = 10, 𝐼1𝑏 = 0.8 , 𝐼2𝑏 = 6


499
Column base is either hinged or fixed.

500 I1c I1c

501 L=8m

502 Fig. 12. Gabled frame 7: Dimensions and relative cross section properties [20].

503 Table 9
504 k-factor for frame 7 (symmetrical partially tapered beam)- sway permitted-hinged base
𝛼 𝛽1𝑠 𝐺𝐵 Column K Saffari et al. [20] Error

(Eq. 14) base (Eq. 52) km K percent

[20] (Eq.55)

0.5 4.886 0.591 hinged 1.255 2.610 1.254 0.04

0.5 4.886 0.591 fixed 0.778 1.619 0.778 0.001

505

506 4.3.3 Design Charts

507 To provide efficient and simple design tool to evaluate the buckling length factor of steel frames

508 composed of linearly tapered columns and partially tapered beams a set of design charts are

509 presented in Fig. 13 to 16. The design charts show the buckling length factor in terms of two main

510 parameters which are the ratio of the largest to smallest moment of inertia of the column and the ratio

511 GB. The effect of the partial tapering in the restraining beams is accounted for by using the proposed

512 bending stiffness modification factors (𝛽1𝑠 , 𝛽2𝑠 , 𝛽1𝑛𝑠 𝑎𝑛𝑑 𝛽2𝑛𝑠 ) which depends on the shape and

513 end conditions of the partially tapered beam. It is evident from these charts that the buckling length

514 factor decreases by increasing the ratio of moment of inertia of the largest to the smallest end of the

515 column.

516

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
517

518

519

520

521

522

523

524

525

526

527

528

529 Fig. 13. Buckling factor for frames prevented from sway with hinged base.

530

531

532

533

534

535

536

537

538

539

540

541

542 Fig. 14. Buckling factor for frames prevented from sway with Fixed base.

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
543

544

545

546

547

548

549

550

551

552

553

554

555

556

557

558

559 Fig. 15. Buckling factor for frames permitted to sway with hinged base.

560

561

562

563

564

565

566

567

568

569

570

571
Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
572 Fig. 16. Buckling factor for frames permitted to sway with Fixed base.

573 4.4 Effect of leaning columns in sway frames

574 Leaning columns in sway frames have no resistance to base shear and P-𝛿 effect of the portion of

575 loads on these columns must be resisted by other rigidly connected columns. Fig. 17 shows a steel

576 portal frame with two rigidly connected columns at the middle and two leaning columns at the two

577 edges. In this figure, the two leaning columns can be thought as a representation of all leaning

578 columns attached to the frame. The load ratio (n) is defined as the ratio between the sum of load

579 acting on all columns (leaning and rigidly connected) to the sum of load acting on the rigidly

580 connected columns only and it can be expressed by the following equation:

∑ 𝑃+𝑃𝑙𝑒𝑎𝑛𝑖𝑛𝑔
581 𝑛= ∑𝑃
(56)

582 The minimum value of (n) is unity which represents a frame without leaning columns. Frames with

583 (n) greater than unity always have leaning columns. The P-𝛿 effect resulting from leaning columns

584 must be considered in shear equilibrium Eqs. (31) or (43) by adjusting them as follows:

2𝐸𝐼1𝑐 𝛿
585 𝑉 𝐿𝑐 = [−(𝛼1 + 𝛼3 ) 𝜃𝑎 − (𝛼2 + 𝛼4 ) 𝜃𝑏 + (𝛼1 + 𝛼2 + 𝛼3 + 𝛼4 ) ]−𝑛𝑃𝛿 (57)
𝐿𝑐 𝐿𝑐

2𝐸𝐼1𝑐 (𝜇 2 +0.25)(𝑐−1)2 𝛿
586 𝑉 𝐿𝑐 = [−(𝛼1 + 𝛼3 ) 𝜃𝑎 − (𝛼2 + 𝛼4 ) 𝜃𝑏 + (𝛼1 + 𝛼2 + 𝛼3 + 𝛼4 − 𝑛 ) 𝐿 ] (58)
𝐿𝑐 2 𝑐

587 Therefore, the buckling conditions for sway frames with leaning columns can be written as follows:

3
(𝛼1 + 𝐺 ) 𝛼2 −(𝛼1 + 𝛼2 )
𝐴
| 3 |
588 𝛼3 (𝛼4 + 𝐺 ) −(𝛼3 + 𝛼4 ) =0 (59)
| 𝐵 |
(𝜇 2 +0.25)(𝑐−1)2
−(𝛼1 + 𝛼2 ) −(𝛼3 + 𝛼4 ) (𝛼1 + 𝛼2 + 𝛼3 + 𝛼4 − 𝑛 )
2

589
P P
(n-1) P (n-1) P
590 I2c I2c

591 I1b I1b I1b


592 𝑷+𝑷𝒍𝒆𝒂𝒏𝒊𝒏𝒈
Lc n= ≥𝟏
𝑷

593
I1c I1c
Lb Lb Lb
594

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
595 Fig. 17. Steel portal frame with leaning columns.

596 Eq. (59) is solved by iteration technique for different values of n, GB and I1c/I2c. The effect of the

597 load ratio (n) on the buckling length factor of tapered columns in sway frames with hinged and fixed

598 bases is shown on Figs. 18 and 19, respectively. The curves in these figures are plotted for certain

599 values of GB. It is obvious from these figures that the buckling length factor, for any given value of

600 GB, increases with the increase of load ratio (n) resulting from additional P-𝛿 effect on leaning

601 columns. Moreover, for a specific value of GB , the buckling length factor curves for different ratios

602 (n) are parallel. This implies that the buckling length factor (kn>1) at specific value of GB, and at

603 specific value of n > 1 for all I2c / I1c , is just a constant multiplier of the buckling length factor (kn=1)

604 at n=1. To determine this multiplier, the relationship between the ratio (kn>1)/ (kn=1) and the ratio (I2c

605 / I1c ) is plotted in Figs. 20 and 21 for hinged and fixed base columns, respectively. For the case of

606 sway frames with hinged bases the value GB is varied from zero to 10 while for the case of fixed

607 bases GB is varied from zero to infinity. As expected, Figs. 20 and 21 show that the ratio (kn>1)/ (kn=1)

608 is almost constant for similar values of GB, irrespective of the value of I2c / I1c. Moreover, it is clear

609 from these figures that there is an upper bound value for the ratio (kn>1)/ (kn=1) for each value of n.

610 The value of this upper bound can be taken conservatively equals to √𝑛 which is indicated by the

611 dashed lines in Figs. 20 and 21.

612 Therefore, the buckling length factor (kn>1) in case of sway frames with leaning columns can be

613 determined by multiplying the buckling length factor (kn=1) in case of no leaning columns by the

614 upper bound value as follows:

615 𝑘𝑛>1 = √𝑛 𝑘𝑛=1 (60)

616 Eq. (60) is similar to the story buckling approach discussed by Yura [35] and included in AISC [1],

617 which means that the story buckling approach is valid to sway frames composed of rigidly connected

618 tapered columns and leaning column as well.

619
Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
620

621

622

623

624

625

626

627

628

629

630

631

632

633 Fig. 18. Effect of loading on leaning columns on the effective buckling length of hinged base sway frames.

634

635

636

637

638

639

640

641

642

643

644

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
645

646 Fig. 19. Effect of loading on leaning columns on the effective buckling length of fixed base sway frames.

647

648

649

650

651

652

653

654

655

656

657

658

659 Fig. 20. Ratio of k-factor for sway frames with leaning columns to k-factor without leaning columns-hinged base case.

660

661

662

663

664

665

666

667

668

669

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
670

671

672 Fig. 21. Ratio of k-factor for sway frames with leaning columns to k-factor without leaning columns-fixed base case.

673 5. Conclusions

674 The current paper developed closed form equations and design charts for bending stiffness

675 modification factors for two common types of partially tapered beams that can be used in stability

676 analysis of steel frames with prismatic or linearly tapered columns. For steel frames with prismatic

677 columns, these modifications factors can be used along with AISC alignment charts and extend the

678 applicability of this alignment charts to include partially tapered restraining beams. For steel frames

679 with linearly tapered columns, the proposed modifications factors are incorporated in the stability

680 analysis and simple design charts are presented for case of sway or no sway frames with different

681 boundary conditions. Results from the proposed approach agrees well with numerical results from

682 LBA and published results in previous literature. The results of the investigated cases showed that

683 partial tapering of restraining beam significantly reduces effective buckling length of frames

684 permitted to sway compared to that of similar frames with prismatic beams having the smallest cross

685 section while that effect is marginal in case of braced frames. Moreover, the additional P-𝛿 effect,

686 resulting from leaning columns in sway frames, on the effective buckling length was investigated. It

687 was proved that story buckling approach commonly used in prismatic steel frames can be extended

688 to steel frames composed of tapered columns and partially tapered beams.

689 6. References

690 [1] AMERICAN INSTITUTE OF STEEL CONSTRUCTION. Specification for Structural Steel Buildings. 130 East
691 Randolph Street, Suite 2000, Chicago, Illinois 60601, 2016. ANSI/AISC 360-16.

692 [2] Lee, G. C., Morrell, M. L. and Ketter, R. L. Design of tapered members. 1972, Welding Research Bulliten
693 173, pp. 1-32.

694 [3] AMERICAN INSTITUTE of STEEL CONSTRUCTION. Supplement No. 3 to the Specificationfor the Design,
695 Fabrication & Erection ofStrucutral Steel for Building. New York : AISC, 1974.

696 [4] Lee, G.; Morrell, M. Application of AISC Design Provisions for Tapered Members. 1975, Engineering
697 Journal, American Institute of Steel Structures, Vol. 12, issue 1, pp. 1-13.

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
698 [5] Hirt, M.A. and Crisinel, M. Charpentes Métaliques–Conception et Dimensionnement desHalles et
699 Bâtiments.: Traité de Génie Civil, Vol. 11. Press Polytechniques et Universitaires Romandes, Lausanne., 2001.

700 [6] Marques, L.; Taras, A.; Simos da Silva,L.; Grenier, R.; Rebelo,C. Development of a consistent buckling
701 design procedure for tapered columns. 2012, Journal of Constructional Steel Research, Vol. 72, pp. 61-74.

702 [7] Eurocode 3. Design of Steel Strucutres, Part 1-1: General Rules and Rules for Buildings. : BS-EN 1993-1-1,
703 2005, Incorporating Corrigenda February 2006 and April 2009.

704 [8] Tankova, T.; Martins, J.P.; Simões da Silva, L.; Simões, R.; Craveiro, H.D. Experimental buckling
705 behaviour of web tapered I-section steel columns. 2018, Journal of Constructional Steel Research, Vol. 147,
706 pp. 293-312.

707 [9] Nguyen, T.; Tran, N.N; Nguyen, D.D. Prediction of Critical Buckling Load of Web Tapered I-Section Steel
708 Columns Using Artificial Neural Networks. 2021, International Journal of steel structures,
709 https://doi.org/10.1007/s13296-021-00498-7

710 [10] Ermopoulos, J. Ch. Elastic Stability Analysis of Plane Rectangular Frames with Varying Stiffness
711 Members. National Technical University of Athens. Athens: 1984. Ph.D. thesis.

712 [11] Ermopoulos, J. Ch and Kounadis, A. N. Stability of frames with tapered built-up members. 1985, Journal
713 of Strucutral Engineering, ASCE, Vol. 111, issue 9, pp. 1979-1992.

714 [12] Ermopoulos, J. Ch. Buckling of tapered bars under stepped axial loads., 1986, Journal of Structural
715 Engineering, ASCE, Vol. 112, issue 6, pp. 1346-1354.

716 [13] Ermopoulos, J. Ch. Slope-deflection method and bending of tapered bars under stepped loads., 1988,
717 Journal of Contructional Steel Research, Vol. 11, issue 2, pp. 121-141.

718 [14] Ermopoulos, J. Ch. Equivalent Buckling Length of Non-uniform Members. 1997, Journal of
719 Constructional Steel Research, Vol. 42, issue 2, pp. 141-158.

720 [15] Al-Sadder, S.Z. Exact expressions for stability functions of a general non-prismatic beam-column
721 member. 2004, Journal of Constructional Steel Research, Vol. 60, issue 11, pp.1561-1584.

722 [16] Raftoyiannis, I. G.; Ermopoulos, J. Ch. Stability of tapered and stepped steel columns with initial
723 imperfections. 2005, Engineering Structures, Vol. 27, issue 8, pp. 1248-1257.

724 [17] King, W.; Duan, L.; Zhou, R.; Hu, Y.; Chen, W. K-factors of framed columns restrained by tapered girders
725 in US codes. 1993, Engineering Structures, Vol. 15, issue 5, pp. 369-378.

726 [18] Bazeos, N.; Karabalis, D. L. Efficient computation of buckling loads for plane steel frames with tapered
727 members. 2006, Engineering Structures, Vol. 28, issue 5, pp. 771-775.

728 [19] Karabalis, D.L. and Beskos, D.E. Static, dynamic and stability analysis of structures composed of tapered
729 beams. 1983, Computer and Strucutres, Vol. 16, issue 6, pp. 731-748.

730 [20] Saffari, H.; Rahgozar, R.; Jahanshahi, R. An efficient method for computation of effective length factor
731 of columns in a steel gabled frame with tapered members. 2008, Journal of Constructional Steel Research,
732 Vol. 64, issue 4, pp. 400-406.

733 [21] Serna, M. A.; Ibáñez, J. R.; López, A. Elastic flexural buckling of non-uniform members: Closed-form
734 expression and equivalent load approach. 2011, Journal of Constructional Steel Research, Vol. 67, issue 7,
735 pp. 1078-1085.

736 [22] Lu, N.; Meng, L. The element stiffness matrix of a tapered beam with effects of shear deformation and
737 its stability application. 2011, Advanced Materials Research, Vols. 308-310, pp. 1383-1388.
Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993
738 [23] Salem, E. Mapping web-tapered member to a prismatic member for buckling analysis of sway frames-
739 closed form equation. 2013, Structural Stability Research Council Annual Stability Conference, St. Louis,
740 Missouri, April 16-20.

741 [24] Chan, S.L. Buckling analysis of structures composed of tapered members., 1990, Journal of Strucutral
742 Engineering, ASCE, Vol. 116, issue 7, pp. 1893-1906.

743 [25] Kim, M.C., Chang, K.C. and Lee, G.C. Elastic and Inelastic Buckling Analysis of Thin-Walled Tapered
744 members. 1997, Journal of Engineering Mechanics, ASCE, Vol. 123, issue 7,pp. 727-735.

745 [26] Valipour, Hamid R. and Bradford, Mark A. A new shape function for tapered three-dimensional beams
746 with flexible connections. 2012, Journal of Constructional Steel Research, Vol. 70, pp. 43-50.

747 [27] Chiorean, C. G.; Marchis, I. V. A second-order flexibility-based model for steel frames of tapered
748 members. 2017, Journal of Constructional Steel Research, Vol. 132, pp. 43-71.

749 [28] Kucukler, M., Gardner, L. Design of laterally restrained web-tapered steel structures through a

750 stiffness reduction method. 2018, Journal of Constructional Steel Research, Vol. 141, pp. 63-76

751 [29] Kucukler, M., Gardner, L. Design of web-tapered steel beams against lateral-torsional buckling through
752 a stiffness reduction method. 2019, Journal of Constructional Steel Research, Vol. 190, pp. 246-291

753 [30] Duan, L.; Lei, H.; Chen, W. Effective Length Factors of Compression Members, 2014, Bridge Engineering
754 Handbook: Fundamentals, Second Edition, pp. 427-450.

755 [31] Ziemian, Ronald D. Guide to Stability Design Criteria for Metal Structures: Sixth Edition: John Wiley and
756 Sons, 2010.

757 [32] Timoshenko, S. P. and Gere, J. M. Theory of Elastic Stability,. second edition. New York : McGraw- Hill,
758 1961.

759 [33] DUMONTEIL, P. Historical Note on K-factor Equations. 1999, Engineering Journal, AISC, Vol. 36, issue 2,
760 pp. 102-103.

761 [34] STAAD.Pro . SELECT Series 6, 2015, Bently Systems Inc., Exton ., PA, USA.

762 [35] Yura, Joseph. The Effective Length of Columns in Unbraced Frames. April 1971, Engineering Journal, AISC
763 Vol. 8, issue 2, pp. 37-42.

764

Preprint copy of: Effective buckling length of frames with tapered columns and partially tapered
beams, Journal of Constructional Steel Research 187(12)- DOI: 10.1016/j.jcsr.2021.106993

You might also like