EVALUATION OF THE ELASTIC CRITICAL

LOAD OF TAPERED COLUMNS WITH

CIRCULAR OR SQUARE HOLLOW SECTIONS

A. M. Baptista1, J.-P. Muzeau2 and N. Boissonnade2

1
National Civil Engineering Laboratory (LNEC),
Av. Brasil nº101, 1700-066 Lisboa, Portugal
2
LERMES/CUST, Blaise PASCAL University,
BP 206, 63174 Aubière Cedex, France

ABSTRACT
This paper presents a method of evaluation of the elastic critical load of tapered members with circular and
square cross-sections, under simple compression. The cross-sections of these members may be full solid or
hollow, with constant wall thickness, and their external dimensions linearly vary between the column ends.
This method is an adaptation of Euler’s equation, by means of the introduction of a coefficient calibrated
after the results of a large parametrical study. The results obtained with this method show a very good
agreement with those obtained by other analytical and numerical methods. The main advantage of this
method lies in the simplicity of its application, allowing a very fast calculation when compared with
numerical methods, based on the finite element method for example, or with analytical solutions requiring the
solving of complicated mathematical expressions.

KEYWORDS
Elastic stability, critical load, tapered columns, hollow sections, circular sections, square sections.

INTRODUCTION
Tapered structural members represent a valuable solution for some types of steel structures. Nevertheless,
their design may present some difficulties, such as the evaluation of their elastic critical loads. One
possible solution for this problem lies in the use of numerical methods, such as the finite element method
for example. However, computer programs that include special formulation for tapered finite elements, and
allow the evaluation of the structural member critical loads, are not usually available for most of designers.

On the other hand, the use of finite elements with uniform cross-sections to solve this problem requires a
very fine discretisation of the tapered members, which makes this method very tedious due to the large
amount of data and results involved in the computer calculations.

Besides numerical calculations, some analytical methods are available, which usually depend on
coefficients quantified by tables or diagrams. The use of these methods is liable to calculation errors,
resulting from the need of interpolating between tabulated values.
This paper proposes an adaptation of the Euler relationship to the evaluation of the elastic critical load of
tapered members with circular or square cross-sections, solid or hollow, under simple compression. The
diameter of the circular cross-sections or the width of the square sections linearly varies between two
extreme values, located at the member ends. The adaptation is made by means of a coefficient, whose
value depends on several parameters that characterise the influence of the ratios between the extreme
values of the cross-section dimensions and between these values and the thickness of the cross-section
wall. The method proposed for evaluating this coefficient is based on simple analytical expressions. It is
founded on the results of a parametrical study, which has been carried out from more than one hundred
different tapered columns.

The results given by the proposed method have been compared with those obtained by means of different
computer programs. Their differences were very satisfactory, usually smaller than 0,3%.

GENERAL FORMULATION OF THE ANALYTICAL METHOD

The evaluation of the critical load of tapered columns is a classical problem of the elastic stability theories.
Several solutions have been proposed for this problem, as in Timoshenko (1936) and Gere & Carter (1963)
for instance, but they are usually given by means of tables or charts that cover only a small range of cases,
as far as the cross-sections geometry and the members support conditions are concerned. Therefore, there is
usually the need of interpolation between the given values, which may introduce some additional numerical
errors in the final value of the columns critical load.

The solutions found in literature about this specific problem are often written under the form of Eqn. 1 or
Eqn. 2, where Iref represents the moment of inertia (second moment of area) of a reference cross-section
(smallest or largest cross-section, for example) of the tapered column.

Pcr.var = m E I ref l 2 (1)

Pcr.var = P * π 2 E I ref l 2 = P * Pcr.ref (2)
Pcr.var = ρvar Pcr.min (3)

The coefficients m or P*, which depend on the geometrical characteristics of the column cross-sections,
allow the evaluation of the tapered column critical load Pcr.var after the value of the critical load Pcr.ref of a
reference column, with a constant cross-section and the same support conditions of the tapered column. The
solution proposed in this paper for the tapered column critical load Pcr.var is written under the form of Eqn. 3,
where Pcr.min represents the critical load of the reference column, which has the same length and support
conditions, and a constant cross-section equal to the smallest cross-section of the tapered column.

TAPERED COLUMNS WITH SOLID CIRCULAR OR SQUARE CROSS-SECTIONS

The first solution presented in this paper concerns the evaluation of the critical load of cantilever tapered
columns with solid circular or square cross-sections, fixed at its large end and free at the other end. The
diameter (or the width) of the cross-sections presents a linear variation along the column length, between a
maximum value D, at the fixed end, and a minimum value d, at the free end. This solution corresponds to
the case study identified by a coefficient n=4 in Timoshenko (1936) and Gere & Carter (1963).

The value of Pcr.var is found by means of Eqn. 4, identical to Eqn. 3:

Pcr.var = ρcirc Pcr.min (4)

where: Pcr.min = π 2 E I min l 2 (5)
The reference value (non-dimensional) of the tapered column slenderness, λ var , may be evaluated by means
of Eqn. 6, Baptista (1998):
2 _
E Amin ε y 1  εy 
 = λ min
2
_ Py Py 1 Amin l 2
λ var = = = = ε = λ
2
(6)
Pcr .var ρ circ Pcr .min ρ circ π 2 EI min l 2 ρ circ I min π 2
y
ρ circ  min
π  ρ circ
 

where l is the column buckling length, Amin and Imin , the area and the moment of inertia of its smallest
cross-section, εy , the yielding strain of the material (εy=σy/E), and Py the yielding load of the column in
simple compression, which depends on the area Amin. The ρcirc factor may be evaluated by means of Eqn. 7:

ρ circ = A0 + A1 (D d ) + A2 (D d ) + A3 (D d ) + A4 (D d ) + A5 (D d )
2 3 4 5
(7)

The values for the A0 to A5 coefficients to be used in the evaluation of the ρcirc factor are given in Table 1.

Table 1
Coefficients for the evaluation of the ρcirc factor
A0 A1 A2 A3 A4 A5
0,29502151 -0,8287695 0,9206092 0,6931470 -0,0826502 0,0040933

The critical load Pcr.var may also be evaluated, in a simplified manner, by means of charts similar to those
shown in Figures 1 and 2, which have been drawn using Eqns. 4 to 7. Figure 1 gives the value of the
reduced critical load Pcr.var/Py as a function of the D/d ratio and the reference reduced slenderness λ min .

Figure 2 gives the variation of the ρcirc factor as a function of the D/d ratio. This factor represents the
increase of the critical load of the tapered column when the value of diameter D of the largest section
grows. However, this growth requires an increase of the column volume, which may be quantified by the
ratio Vvar/Vmin, between the volumes of the tapered and the reference columns. This ratio is given by Eqn. 8.

Vvar 1 (D d ) − 1 1
( )
3
= = (D d ) + D d + 1
2
(8)
Vmin 3 (D d ) − 1 3

So, the increase of resistance per mass unit of the column, resulting from its tapered shape, may be
obtained dividing the ρcirc factor by the ratio Vvar/Vmin. The evolution of these benefits is also represented in
Figure 2, as a function of the D/d ratio. It may be seen that it grows faster for small values of this ratio and
that it becomes practically constant for D/d ratios greater than 4,5.

The results obtained with the proposed method have been compared with those given by two other
methods, i.e. Timoshenko (1936) and Gere & Carter (1963) methods.

The method presented by Gere & Carter (1963) is based on Eqn. 9, which may be written under the form of
Eqn. 10, resulting from Eqns. 4, 5 and 9.

d P 2   d d Pcr .var L2
 cr . var L +
tan  =0 (9)
D EI min   D−d  D EI min
 
d π   d d π
tan ρ circ +  ρ circ = 0 (10)
D 2   D−d  D 2
 Tapered column Tapered column
Pcr.var/Py
with full circular cross-section with full circular cross-section
20 20
D/d=1
19 D/d=1,5 18
18 D/d=2 ρcirc
17 D/d=2,5 16
16 D/d=3
15 D/d=3,5 14
14 D/d=4
13 D/d=4,5 12
12 D/d=5
11 10
10
9 8
8 ρcirc * Vmin / Vvar
7 6
6
5 4
4
3 2
2
1 0
0 λmin 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0
0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8 2,0 D/d

Figure 1 – Variation of the critical load versus the Figure 2 – Increase of the critical load versus
reference reduced slenderness λ min the diameter ratio D/d

The relative differences of the ρcirc factor values given by Eqn. 10 to those given by Eqn. 7 are shown in
Table 2 for some particular values of the D/d ratio. The absolute values of these differences vary between
0,001% and 0,091%; if this comparison is restricted to D/d values from 1,5 to 5,0, than the absolute values
of their relative differences do not exceed 0,02%, showing thus an excellent agreement between the Gere &
Carter (1963) approach and the one presented in the present work. Nevertheless, the second one has an
important advantage over the first one, consisting in the great simplicity of application of Eqn. 7 when
compared to the resolution of Eqn. 10.
The method presented by Timoshenko is based on Eqn. 11. The value of the m coefficient may be obtained
by interpolation between the closest m values given in a table for ten different Imin/Imax ratios, within the
limits: 0,1 ≤ Imin/Imax ≤ 1 (which correspond to the following D/d ratios: 1≤ D/d ≤1,77828). The value of
the ρcirc factor may be evaluated, in this case, according to Eqn. 12, resulting from Eqns. 4, 5 and 11.

m EI max
Pcr .var = (11)
L2
= 2 max = 2 (D d )
m I m
ρ circ
4
(12)
π I min π

The relative differences of the ρcirc factor values given by Eqn. 12 to those given by Eqn. 7 are also shown
in Table 2. The absolute values of these differences vary between 0,09% and 1,36% in a non-uniform way,
which is probably related to the numerical errors resulting from the interpolation process.
For D/d values greater than 1,77828, out of the range covered by Timoshenko (1936), the ρcirc factor has
been calculated by extrapolation. However, the large differences found between these values and those
calculated according to Eqn. 7 show that this method is not suitable. Therefore, the application of the
Timoshenko method must be restrained to the above-mentioned range of D/d values.
 Table 2
Relative differences of the ρcirc factor values evaluated according to Timoshenko (1936) and
Gere & Carter (1963) to the values of the ρcirc factor according to the present work
D/d Baptista & Muzeau Gere & Carter Dif(%) Timoshenko Dif(%)
1,1 1,305475 1,306120 0,049 1,303272 -0,169
1,2 1,662735 1,664251 0,091 1,657088 -0,340
1,3 2,075436 2,077029 0,077 2,059233 -0,781
1,4 2,545639 2,546850 0,048 2,514558 -1,221
1,5 3,075276 3,075889 0,020 3,072565 -0,088
1,6 3,666145 3,666120 -0,001 3,616453 -1,355
1,8 5,038160 5,037158 -0,020 5,055052 0,335
2,0 6,673678 6,672378 -0,019 7,062086 5,820
2,5 11,978540 11,978693 0,001 15,474278 29,183
3,0 19,109171 19,110307 0,006 30,771243 61,029
3,5 28,137651 28,137495 -0,001 55,962223 98,887
4,0 39,104187 39,103082 -0,003 94,603589 141,927
4,5 52,032458 52,033263 0,002 150,798851 189,817
5,0 66,944967 66,944106 -0,001 229,198650 242,369

TAPERED COLUMNS WITH CIRCULAR OR SQUARE HOLLOW SECTIONS

Evaluation of the critical load
The second solution presented in this paper concerns the evaluation of the critical load of a cantilever
tapered column with a hollow circular or square cross-section, fixed at one end and free at the other end.
The external diameter (or width) of the cross-sections presents a linear variation along the column length,
between a maximum value D, at the fixed end, and a minimum value d, at the free end. The thickness t of
the cross-sections wall is constant along the column length.

The value of Pcr.var is found by means of Eqn. 13, identical to Eqn. 3. The critical load Pcr.min of the
reference column is also defined by Eqn. 4.
Pcr.var = ρhol Pcr.min (13)

The ρhol factor may be obtained by means of Eqn. 14:

ρ hol = B0 + B1 (t d ) + B2 (t d ) + B3 (t d )
2 3
(14)

The Bi coefficients (i=0 to 3), used in the evaluation of the ρhol factor, are calculated according to Eqn. 15.
The values for the Bij coefficients needed for these calculations are given in Table 3.
Bi = Bi 0 + Bi 1 ( D d ) + Bi 2 ( D d ) + Bi 3 ( D d )
2 3
(15)

Table 3
Coefficients for the evaluation of the ρhol factor
Bi Bi0 Bi1 Bi2 Bi3
B0 (i=0) 0,166951172 -0,378556380 1,199683641 0,015695978
B1 (i=1) -0,098259968 -1,956605350 1,940427622 0,098235933
B2 (i=2) 8,211617455 -15,168446798 7,746830598 -0,538262701
B3 (i=3) -12,045269681 23,839166595 -13,852721015 1,485784260
The critical load Pcr.var may also be evaluated, in a simplified manner, by means of charts similar to those
shown in Figures 3 and 4, which have been drawn from Eqns. 14 and 15.

ρ hol Tapered column

with hollow circular
ρ hol* V min /V var
52 13 t/d=0,3
50 t/d=0,3 cross-section t/d=0,28
48 t/d=0,28 12 t/d=0,26
46
44 t/d=0,26 t/d=0,24 t/d=0,30
11
42 t/d=0,24 t/d=0,22
40 t/d=0,22 10
t/d=0,2
38 t/d=0,30 t/d=0,18
36 t/d=0,2
t/d=0,16
34 t/d=0,18 9
t/d=0,14
32 t/d=0,16
30 8 t/d=0,12
28 t/d=0,14 t/d=0,1
26 t/d=0,12 7 t/d=0,08
24 t/d=0,1 t/d=0,06
22 6 t/d=0,04
20 t/d=0,08
t/d=0,02 t/d=0,02
18 t/d=0,06 5
16 t/d=0,04
14 t/d=0,02
12 t/d=0,02 4 Tapered column
10 with hollow circular
8 3
6 cross-section
4 2
2 D/d D/d
0 1
1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0

Figure 3 – Increase of the critical load as a Figure 4 – Increase of the critical load per
function of the D/d and t/d ratios volume unit of the tapered column, as a
function of the D/d and t/d ratios

Figure 3 gives the variation of the ρhol factor as a function of the D/d and t/d ratios. This factor represents
the increase of the critical load of the tapered column when the values of the wall thickness t or the
diameter D of its largest section are enlarged.

Once again, the enlargement of the cross-section dimensions implies an augmentation of the column
volume, which may be quantified by the ratio Vvar/Vmin, between the volumes of the tapered and the
reference columns. In the case of tapered columns with hollow circular cross-sections, this ratio is given by
Eqn. 16.
Vvar 1  (D d ) − 2 * t d + 1 
=   (16)
Vmin 2  1− t d 
The increase of resistance per mass unit of the column, as a function of the D/d and t/d ratios, is
represented in Figure 4. It may be seen that it is always increasable with both ratios, D/d and t/d. Figure 4
shows also the advantages of using tapered columns with hollow circular sections compared to those of
tapered members made of full circular sections, since the increase of resistance per mass unit is always
growing with the D/d ratio, in the first case, and it tends to stabilise in the second case (Figure 2).

This calculation method, using either Eqns. 14 and 15 or Figure 3, may also be applied to evaluate the
critical load of tapered columns made of hollow square sections, with a constant wall thickness t and an
external width varying linearly in both directions between two extreme values, d and D, at the ends of
column. These cantilever columns are supposed to have the same support conditions, fixed at the basis
(corresponding to the largest cross-section) and free at the top.
In the meanwhile, it should be noticed that the calculation method presented above has been developed
under the hypothesis that all the cross-sections of the tapered column belong to Classes 1, 2 or 3, according
to Eurocode 3 (1992). So, an additional verification is needed for the effects of local buckling, in the case
of very slender (Class 4) cross-sections.

Application example
The following application example concerns a tapered column with hollow circular section, with an
external diameter varying linearly between two extreme values d and D, at the ends of column. The column
is fixed at its basis, where D=1035 mm, and free at the top, where d=300 mm. The wall thickness t of the
cross-sections is constant and equal to 36 mm. The Young modulus of the material is 210 GPa and its yield
strength fy is taken equal to 235 MPa.

Before the calculation of the critical load Pcr.var, it is necessary to evaluate the following ratios:
D t = 1035 36 = 28,8 ; D d = 1035 300 = 3,45 ; t d = 36 300 = 0,12

Since D/t is smaller than 50, all the cross-sections may be classified as Class 1, according to Eurocode 3,
and no checking is required for local buckling. The value of the critical load of this tapered column may be
evaluated according to Eqn. 13. The ρhol factor is determined by means of Eqn. 14, the Bi coefficients being
calculated by means of Eqn. 15 and Table 3:
B = 0,166951172-0,378556380 × 3,45 + 1,199683641× 3,52 + 0,015695978 × 3,453 = 13,78469995
0
B = - 0,098259968 - 1,956605350 × 3,45 + 1,940427622 × 3,452 + 0,098235933 × 3,453 = 20,28131486
1
B = 8,211617455 - 15,168446798 × 3,45 + 7,746830598 × 3,452 - 0,538262701 × 3,453 = 25,98410949
2
B = -12,045269681 + 23,839166595 × 3,45 - 13,852721015 × 3,452 + 1,85784260 × 3,453 = -33,67046913
3
ρ = 13,78469995 + 20,28131486 × 0,12 + 25,98410949 × 0,122 - 33,67046913 × 0,123 = 16,53444634
hol

The critical load Pcr.min of the reference column (with constant cross-sections equal to the smallest
cross-section of the tapered column) is evaluated by Eqn. 5, where the buckling length l is
l = 2 L = 2 × 15300 mm = 30600 mm and the moment of inertia Imin of the reference cross-section is
(
I min = π d 4 − (d − 2t )
4
) 64 = 264957199 mm 4
. Therefore, the value of Pcr.min is:

π 2 EI min π 2 × 210000 × 264957199

Pcr . min = = = 586478 N ,
l2 30600
and the critical load Pcr.var of the tapered column, calculated by means of Eqn. 14, is:
P =ρ P = 16,53444634 × 586478 N = 9697 kN .
cr. var hol cr. min

The value of ρhol factor could also be obtained, in a simplified manner, by means of Figure 3. In the present
case, for D d = 3,45 and t d = 0,12 , the value of ρhol has been estimated as ρ hol = 16,5 and, therefore, we
have P = 16,5 × 586478 N = 9677 kN , which is only 0,2% different from the previous one.
cr. var
However, this difference depends on the errors associated to the reading of ρhol from Figure 3.

Nevertheless, it should be emphasised that the calculation of the tapered column critical load value may
become very easy if a simple computer calculation sheet is used to store the Bij coefficients from
Table 3 and do all the calculations in order to get the value of Pcr.var.
This particular problem is not usually covered in the literature concerning the stability of tapered columns.
So, in order to compare the previous results, the critical load of this tapered column has also been
evaluated by means of a numerical procedure, using a computer program able to take into account tapered
finite elements, Baptista (1994). The solution given by this program was Pcr.var= 9695 kN.

It is possible to see that the value of Pcr.var= 9697 kN, obtained by means of the calculation method
proposed in this paper, is very close (only 0,02% larger) to the result of the numerical computation, which
means that the first method may represent a good alternative to the second one. A similar conclusion has
been obtained in several other calculation examples used to test the first method, the largest differences
between the results given by these two methods being smaller than 0,3%.

CONCLUSIONS
This paper proposes an evaluation method for of the elastic critical loads of tapered members with solid or
hollow circular and square cross-sections, under simple compression. The external dimensions of the
cross-sections vary linearly between the column ends. This method is an adaptation of Euler’s equation, by
introducing a coefficient depending on the main parameters that characterise the tapered form of the
column. The evaluation of this coefficient is made by means of simple analytical expressions, calibrated
from the results of a large parametrical study of this type of tapered columns.

The advantages of the method are the accuracy of its results, which are very close to those given by other
methods, and the simplicity of its application, allowing a very fast calculation by means of a common
computer worksheet or a hand calculator. By this way, it is not necessary to solve complicated mathematical
expressions or to spend a lot of time preparing data for numerical methods, as in the finite element method for
example. The proposed method allows also the user to develop charts or tables, for a fast determination of the
critical load, according to his specific needs, reducing the inevitable errors associated to the reading from a
chart or interpolation between tabulated values.

This method improves other methods proposed by the authors for the evaluation of the critical load,
Baptista & Muzeau (2001), or the ultimate elastic-plastic load, Baptista (1995, 1997), of rectangular, I-
shaped or built-up tapered columns, with several combinations of support conditions at their ends.

REFERENCES
Baptista A.M. (1994). Modèle non linéaire géométrique et matériel fondé sur l'analyse des déformations
globales des sections. PhD Thesis, Blaise Pascal University, Clermont-Ferrand, France, 480 p.
Baptista A.M. (1995). Resistência ao varejamento de pilares de inércia variável em aço. Nota Técnica
nº27/95-NCE, LNEC, Lisboa, Dezembro 1995, 234 p.
Baptista A.M. (1997). Estudo da resistência de pilares de inércia variável em aço em regime elastoplástico.
Nota Técnica nº13/97-NCE, LNEC, Lisboa, Dezembro 1997, 88 p.
Baptista A.M. (1998). Utilização das curvas de encurvadura do Eurocódigo 3 na verificação da segurança
de elementos estruturais de aço com inércia variável sujeitos a compressão simples. Jornadas Portuguesas
de Engenharia de Estruturas, LNEC, 301-310.
Baptista A.M. and Muzeau J:P. (2001). Determinação da carga crítica de pilares com secção de altura
variável sujeitos a compressão simples. III Encontro CMM, Aveiro, Portugal, 595-604.
Gere J.M. and Carter W.O. (1963). Critical Buckling Loads for Tapered Columns. Transactions of ASCE,
128:2, 736-754.
Timoshenko S. (1936). Theory of Elastic Stability, 1st Edition McGraw-Hill, New York, USA, 518 p.
ENV1993-1-1 (1992) – Eurocode 3: Design of steel structures. Part 1.1: General rules and rules for
buildings. European Committee for Standardisation (CEN), 344 p.