You are on page 1of 7

The Effective Length of Columns in Unbraced Frames with more than one story and Pinned Bases.

(Revised including AISC comments)


RAUL CANLE STRUCTURAL ENGINEER FLUOR DANIEL S.A., ASTURIAS, SPAIN.

INTRODUCTION
The AISC specification, both in ASD and LRFD codes, and the EUROCODE 3, includes alignment charts to determine the K factor to be applied in the calculation of the buckling for a particular compressed member in a frame. Both Codes provide similar approach in the calculation of effective length of a compressed members, AISC provisions are based in the studies from Julian and Lawrence (1959), EUROCODE 3 alignment chart was developed by Wood (1974) with similar results to those provided by Julian and Lawrence methodology. In the referenced case where a frame with more than one story and pinned base both codes provides similar results, both unconservative with an error in some cases (see attached examples) close to 80% that is high enough to deserve consideration in the day to day engineering calculations. Several methods has been developed to solve the referenced problem, the most simple methods are the unit load method or Sahmel method that can be found in the reference [1] and the asymmetric method (ASM), that can be found in reference [2].

Some selected examples will be tested with asymmetric method both in AISC and EUROCODE 3 specifications. The presented methods were compared with results from modern FE linear buckling software MARC DESIGNER3. The ASM is easily codified to computer modern languages or customized spreadsheets with obvious advantage against hand calculation. The use of unconservative K factors, lamentably, is widely use in engineering practice specially when the assumptions on which it is based the alignment chart are violated. The referenced type of unbraced frame with pinned connections in the foundation is amply used in the structural designs (e.g. pipe racks, industrial buildings, conveyor support frames, etc) to take in consideration performing special analysis for K buckling factors calculation, specially when tremendous differences exists with the standard procedure.

ASSUMPTIONS
The proposed procedure is based in the same assumptions of idealized conditions that are expound in the AISC and EUROCODE 3 specifications for the alignment chart: 1. Behavior is purely elastic 2. All members have constant section. 3. All joints are rigid.

Sahmel method is based in the similarity of the deflections in a frame, when the buckling load is exceed in a small quantity, and the deflection due to a horizontal unit load acting in the girder. The Sahmel method is not as exact as the asymmetric method and is only mentioned as reference.

cross

Page 1 of 7

4. Rotations at the far ends of the restraining members are equal in magnitude and opposite in sign, producing reverse-curvature bending. P 5. The column stiffness parameter L must I be identical for all columns within the story. 6. Joint restraint is distributed to the column above and below the joint in proportion to I/L of the two columns. 7. All columns in the frame become unstable simultaneously. 8. The frame is subjected to vertical loads applied only at the joints. 9. No significant axial compression force exists in the beams. 10. The frame being analyzed is symmetrical. 11. The loads in the frame are symmetrical. The EUROCODE 3 includes in Annex E several effective stiffness coefficients for beams with rotational restraints, concrete floor slabs frames and axial compression that are not included in the presented examples in order to provide a consensus comparison with AISC Code. Although these limitations are clearly restrictive to the number of cases where can be applied in practice, some of the mentioned unbraced frame designs in industrial practice are often close to those restrictions. The method has application to plane frames where is assumed that there is no buckling normal to the structure plane. In final calculations the investigation of buckling in both planes has to be performed.

NOMENCLATURE
E = Structural Steel Modulus of Elasticity K = Effective length factor I = Moment of Inertia of a section. L = Member length. Pcolumn = Compression load in a column. Pcrit = Euler buckling load LF = Load Factor for a particular member, is the factor that multiplied by the column compression load gives the buckling load for that column. Ccrit = Minimum Load Factor in the whole analyzed model.

CALCULATION PROCEDURE
Calculation procedure following minimum steps: requires the

1.- Initial data: Model geometric dimensions, Moment of Inertia values and column loads in consistent units. 2. K factors calculation as per AISC or EUROCODE 3 alignment charts. 3. - Calculation of Euler buckling load for each column of the model. 2 * E * In Pcrit n = 2 2 Ln * K n 4. Calculation of the model Load Factor associated to each column. LFn = Pcrit n Pcolumnn

ASYMMETRIC METHOD
The method included in this paper has been modified to eliminate the inclusion of preliminary correction factors to the column with minimum load factor per story. The decision of such change is due to the small influence of this preliminary adjustment in the final result, taking into account the differences with FE approach that can be considered as reference. 5. - Selection of the minimum Load Factor as Critical Load that produces the overall buckling for the model. Ccrit = Min LFcolumn1 ....LFcolumnn The column associated with this Ccrit will be called the reference column.

6. - Calculation of the Modified K factors per column, applying the Eulers equation taking in consideration the Critical Load calculated in previous step.

Page 2 of 7

EXAMPLE 1 (AISC - MKS Unit System) * E * In


2

Ln * K n

= Pn * Ccrit K mod =

* E * In
2

Ln * Ccrit * Pn

1.- GEOMETRY.
900

Substituting Ccrit in this last equation: K mod = 2 * E * In Pcrit _ ref * Pn L2 n * Pcolumn _ ref 2 * E * I n * Pcolumn _ ref = 2 * E * I ref L2 n * 2 * Pn 2 Lref * K ref

Finally, K mod
600

K mod = K ref *

Lref Ln

I n * Pcolumn _ ref I ref * Pn Member data.


Mem. 1 2 3 4 5 6 Pcol Kg 56600 56600 13600 13600 0 0 E Kg/cm2 2100000 2100000 2100000 2100000 2100000 2100000 I Cm4 38040 38040 27521 27521 74610 23850 L Cm 600 600 365 365 900 900 I/L 63,4 63,4 75,4 75,4 82.9 26,5

In the following paragraph will be analyzed a set of examples, applied to AISC and EUROCODE 3, and the solution for the presented method. All those examples were verified with modern FE linear buckling software package MARC DESIGNER3.

EXAMPLES
2.- AISC K factors calculation. Three different examples are included both with AISC and EUROCODE 3 specifications, the first one in MKS Unit System, second with SI unit system and the last one in U.S. Customary Units. In all the examples the AISC inelastic K-factor procedure was not included due to the fact that the reduction factor is for all the examples equal to 1.0, in other cases the inelastic procedure should be applied. Column 1 and 2.
GA = 10 ; Pin connection in foundation (AISC assumed pinned connection) Ic Lc 75,4 + 63,4 GB = = = 1,67 Ig 82,9 Lg From AISC alignment chart (sidesway uninhibited): K1=K2= 2,05

Column 3 and 4.

GA =

Lc Lg
Ig

Ic

75,4 + 63,4 = 1,67 82,9

Page 3 of 7

965

GB =

Lc Lg
Ig

Ic

75,4 = 2,84 26,5

Final comparison table


Member Ccrit CcritMARC K AISC Kmod KMARC Error Kmod Error AISC 1 9,16 11,47 2,05 2,05 1,84 11%C* 11%C 2 9,16 11,47 2,05 2,05 1,84 11%C 11%C 3 9,16 11,47 1,67 5,86 5,24 12%C 68%U 4 9,16 11,47 1,67 5,86 5,24 12%C 68%U

From AISC alignment uninhibited): K3=K4= 1,65

chart

(sidesway

3. - Calculation of Euler buckling load for each column of the model. 2 * E * I1 2 = 518523 Kg 2 2 L1 2 * K1 2 2 * E * I 3 4 L3 4 * K 3 4
2 2

* U= Unconservative error

C= Conservative error

Pcrit1 2 =

From this example it can be observed that the error in the final K factor with the proposed simplified method, compared with MARC software is about 12%. Regarding method of differences compared software. differences with traditional the AISC alignment chart the are unconservative in 68 % with MARC DESIGNER

Pcrit 3 4 =

= 1550043 Kg

4. Calculation of the model Load Factor associated to each column. Pcrit1 2 518523 = = 9,16 56600 Pcolumn1 2 Pcrit 3 4 1550043 = = 113,97 LF3 4 = 13600 Pcolumn3 4 LF1 2 = 5. - Selection of the minimum Load Factor as Critical Load that produces the overall buckling for the model. Ccrit = Min LFcolumn1 ....LFcolumn4 = 9,16 The reference columns are 1 and 2. 6. - Calculation of the Modified K factors. Column 1 and 2 K mod 1 2 = 2,05 Does not change
K mod34 = 2 * E * I 34 L34 *Ccrit* P34
2

Slenderness ratios for columns 3 and 4 are: K AISC K mod L 365 = 1,67 * = 44,8 rx 13,6 L 365 = 5,86 * = 157,2 rx 13,6

In this case the column slenderness, even with a 68% variation complies with codes.

EXAMPLE 1 (EUROCODE 3 - MKS Unit System) EUROCODE 3, K factors calculation. Column 1 and 2.
GA = 1 ; Pin connection in foundation (no rigidity ) Ic Lc 75,4 + 63,4 = 0,626 GB = = Ig Ic 82,9 + 75,4 + 63,4 Lg + Lc

2 * E * 27521 = 5,862 3652 * 9,16*13600

or,

K mod 3 4 = K12 *

I *P L12 * 34 1 2 L3 4 I 12 * P3 4

K mod 3 4

600 27521 * 56600 = 2,05 * * = 5,86 365 38040 * 13600

From EUROCODE alignment chart (sidesway uninhibited): K1=K2= 2,83

Page 4 of 7

Column 3 and 4.
75,4 + 63,4 GA = = = 0,626 Ig Ic 82,9 + 75,4 + 63,4 Lg + Lc

Lc

Ic

Member data.
Mem. 1 2 3 4 5 6 7 8 9 Pcol KN 69 69 44 44 25 25 0 0 0 E MPa 205800 205800 205800 205800 205800 205800 205800 205800 205800 I m4 5,696E-5 5,696E-5 5,696E-5 5,696E-5 5,696E-5 5,696E-5 1,94E-5 1,94E-5 1,94E-5 L m 4,8 4,8 1,2 1,2 1,4 1,4 3 3 3 I/L *10-6 11,87 11,87 47,47 47,47 40,69 40,69 6,47 6,47 6,47

GB =

Lc Lg + Lc
Ig Ic

Ic

75,4 = 0,74 26,5 + 75,4

From EUROCODE 3 alignment chart (sidesway uninhibited): K1=K2= 1,90 Final comparison table
Member Ccrit CcritMARC K EUROCODE 3 Kmod KMARC Error Kmod Error EC3 1 4,83 7,79 2,83 2,83 2,23 27%C 27%C 2 4,83 7,79 2,83 2,83 2,23 27%C 27%C 3 4,83 7,79 1,90 8,07 6,36 24%C 70%U 4 4,83 7,79 1,90 8,07 6,36 24%C 70%U

2. - Calculation table.
Member GA GB KAISC PCRIT(KN) LF CCRIT KMOD 1-2 10 9,17 2,94 580 8,45 8,45 2,94 3-4 9,172 13,62 3,14 8151 184,8 8,45 14,68 5-6 13,62 6,29 2,86 7216 294,5 8,45 16,88

Eurocode 3 gives conservative error of 27% compared with MARC Software in columns 1-2 and unconservative error of 70% in the columns 3-4. EXAMPLE 2 AISC (SI Unit System) 1.- GEOMETRY.

Final comparison table


Member Ccrit CcritMARC K AISC Kmod KMARC Error Kmod Error AISC 1-2 8,45 19,47 2,94 2,94 1,94 51%C 51%C 3-4 8,45 19,47 3,14 14,68 9,67 51%C 67%U 5-6 8,45 19,47 2,86 16,88 11,12 51%C 74%U

In this case the conservative error of Asymmetric method is about 51% versus MARC Software, AISC standard procedure gives an unconservative error 67% in members 3-4 and 74% in members 5-6.

Page 5 of 7

EXAMPLE 2 EUROCODE 3 (SI Unit System) Calculation table.


Member GA GB KEC3 PCRIT(KN) LF CCRIT KMOD 1-2 1 0,902 5,1 193 2,81 2,81 5,1 3-4 9,172 13,62 3,14 5638 127,85 2,81 25,46 5-6 13,62 6,29 2,86 5065 206,76 2,81 29,28

3 4 5 6 7 8 9 10 11

88 88 132 209 44 0 0 0 0

29000 29000 29000 29000 29000 29000 29000 29000 29000

350 350 612 1330 350 3540 3540 1600 1600

234 234 234 234 234 198 234 198 198

1,496 1,496 2,615 5,684 1,496 17,879 15,128 8,081 8,081

Final comparison table


Member
Ccrit CcritMAR K AISC Kmod KMARC Error Kmod Error AISC

Final comparison table


Member Ccrit CcritMARC K EC3 Kmod KMARC Error Kmod Error EC3 1-2 2,81 13,03 5,1 5,1 2,37 115%C 115%C 3-4 2,81 13,03 3,77 25,46 11,82 115%C 68%U 5-6 2,81 13,03 3,41 29,28 13,59 115%C 75%U

1-2 8,25 10,5 1,07 1,70 1,5 13%C 28%U

3-4 8,25 10,5 1,08 1,58 1,4 13%C 23%U

5 8,25 10,5 1,71 1,71 1,5 14%C 14%C

6 8,25 10,5 1,71 2,0 1,8 11%C 5%U

7 8,25 10,5 1,67 2,24 2,0 12%C 16%U

For this example with EC3 the Kmod is much more conservative than in the previous case with AISC alignment charts since the Ccrit factor is lower than the AISC factor. EXAMPLE 3 AISC (US Customary Unit System) 1.- GEOMETRY.

EXAMPLE 3 EC3 (US Customary Unit System) Final comparison table


Member
Ccrit CcritMARC K EC3 Kmod KMARC Error Kmod Error EC3

1-2 5,32 7,89 1,1 2,12 1,73 22%C 36%U

3-4 5,32 7,89 1,12 1,98 1,65 20%C 32%U

5 5,32 7,89 2,13 2,13 1,75 22%C 22%C

6 5,32 7,89 2,13 2,50 2,1 20%C 2%C

7 5,32 7,89 2,06 2,80 2,33 20%C 12%U

Although this example does not comply with the assumption of column stiffness parameter L*(P/I)^0,5 identical for all columns within the story, the modified K factor gives good approximation to real values. On the other hand, it can be observed that the EC3 gives higher errors in the Kmod calculation since the initial K factors calculated are more conservative than AISC approach. Member data.
Mem. 1 2 Pcol Kip 44 44 E Ksi 29000 29000 I In4 144 144 L In 198 198 I/L In3 0,727 0,727

In any case the modified K factor is always conservative.

Page 6 of 7

CONCLUSIONS
The asymmetric method gives a reasonable approximation to the K factor to perform final code check analysis for the referenced case. Although is not as accurate as FE solutions the approximation is close enough, and much more conservative than the direct application of alignment charts included in EC3 and AISC Codes. In general EC3 gives more conservative results than AISC Code. For more accurate results it is recommended the calculation with FE linear buckling software due its versatility and no limitation to boundary conditions or geometry complexity.

REFERENCES
1. Paul Sahmel, Duisburg-Neudorf, Der Stahlbau April 1955 2. C.V.G. Siderurgica del Orinoco (SIDOR) .Manual de Proyectos de Estructuras de Acero. 1982 3. MARC Analysis Research CorporationEurope. 2715CA Zoetermeer, The Netherlands. MARC Designer Software. 4. LeMessurier,W.J. A Practical Method of Second Order Analysis, Part 2-Rigid Frames AISC Engineering Journal 2nd Qtr.,1977 . 5. Lui, E.M. A Novel Approach for K Factor Determination AISC Engineering Journal 4th Qtr.,1992 6. American Institute of Steel Construction. Specification for Structural Steel Buildings Allowable Stress Design. Chicago IL 1989 7. American Institute of Steel Construction, Load and Resistance Factor Design Specification for Steel Buildings, Chicago IL 1993 8. CEN (1993) ENV 1993-1-1 Eurocode 3, Design of Steel Structures, European Committee for Standardization, Brussels.

Page 7 of 7