See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/275956168

Method to determine in a simple way the critical load, considers possibility of

variable cross-section under any type of load (including thermal loading)

Conference Paper · May 2010

CITATIONS READS

2 1,094

4 authors, including:

Mariano Cacho-Perez Antolín Lorenzana

Universidad de Valladolid Universidad de Valladolid
48 PUBLICATIONS   27 CITATIONS    144 PUBLICATIONS   289 CITATIONS   

SEE PROFILE SEE PROFILE

Pablo M López-Reyes
Universidad de Valladolid
12 PUBLICATIONS   12 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Automatic Control and Civil Structures View project

PROGNOSIS Y ANALISIS INTEGRADO DE LAS VIBRACIONES INDUCIDAS POR EL HOMBRE EN ESTRUCTURAS View project

All content following this page was uploaded by Mariano Cacho-Perez on 07 May 2015.

The user has requested enhancement of the downloaded file.

 ECCM 2010
IV European Conference on Computational Mechanics
Palais des Congrès, Paris, France, May 16-21, 2010

Method to determine in a simple way the critical buckling load of

frames with variable cross-section members under any type of load
M. Cacho1 , A. Lorenzana1,2 , P.M. López-Reyes2 , R. Sáiz2

1 Structures and Solid Mechanics Group. E.I.I., University of Valladolid, Spain. {cacho,ali}@eis.uva.es
2 CARTIF Centro Tecnológico, Spain. {pablop,rossai}@cartif.es

Abstract

Ensuring that flexural buckling should not take place in a loaded frame is in most cases the most
critical design condition. We present a general approach for the systematic calculation of the critical
buckling load and the buckling mode of any frame. The Navier-Bernoulli beam model is considered,
having the possibility of variable cross-section under any type of load (including any distributed
load and thermal loading). With this purpose, consider the equilibrium equations of each beam,
the compatibility of displacements and the equilibrium conditions in the deformed configuration,
under the hypothesis of infinitesimal strains and displacements, the so-called Second-Order Theory,
which results in a system of differential equations with variable coefficients for each element. In
order to obtain the nonlinear response of the frame, it is necessary to impose the compatibility of
displacements and the equilibrium of forces and moments in each beam-end, also in the deformed
configuration. The solution is obtained by requiring that the total variation of potential energy be
zero at the instant of buckling. The objective of this work is to develop a systematic method to
determine the critical buckling load and the buckling mode of any frame, without using the common
simplifications usually assumed in matrix analysis or finite element approaches. This allows us to
obtain precise results regardless of the discretization done.

Keywords:

Critical load; Buckling mode; Geometrical nonlinearity; Variable inertia; Thermal loading.

1 Introduction

The study of the evolution of straight prismatic elements under compression has been a source of theo-
retical contributions and practical results of great interest for Structural Engineering. Its origin is closely
linked to the new vision of physics that was provided by the well known work of Galileo “Discorsi e
dimostrazioni matematiche intorno à due nuoue scienze attenenti alla mecanica & i movimenti locali” in
1638 (1). The progress in the study of the behavior of beams took place thanks to the development of the
Mathematical Physics in the seventeenth and eighteenth centuries. The contributions from L. Euler, and
J. and D. Bernoulli (2) gave rise to the Strength of Materials as a science, specially through the analysis
of elastic prismatic elements, including the beam-column.

The phenomenon of instability was not completely explained until, among other things, the neutral
axis was correctly positioned in 1826 by L.C. Navier (1) together with beam deflection and bending
theory.

1
 Once the isolated beam behaviour was understood, it was necessary to develop accurate models
for the analysis of the structure as a whole. For linear problems, apart from traditional eigenvalues
methods for determining the critical buckling load, have also appeared others as Southwell and Timo-
shenko methods, that seek bifurcation points and stability limits for a given strain rate, the method of
Engesser-Vianello for determining the post-buckling deformed shape, or the Koiter method (3) for de-
termining the buckling loads even for variable-section columns. On the other hand, there are also well
known approximated methods (2), based on series expansions (4) or energy methods (Rayleigh-Ritz and
Stodola-Vianello (5; 6)).

In the stability analysis of elastic frames, various techniques have been developed: matrix analysis,
finite elements, line elements and generalized splines (3; 7; 8; 9). The matrix analysis was first used in
the aviation industry in the middle of last century and its main instigator was Levy (10), which introduced
the direct stiffness method (11). These techniques determine the stiffness matrices of prismatic elements
subjected to compression, called stability functions (11). The birth of the Finite Element Method (12)
is usually associated to matrix analysis. In the last few years there have been several efforts to improve
the accuracy of the displacement-based finite element methods, for the analysis of the beam-column
when geometric and material nonlinearity are taken into account since, in particular, the variation of
the curvature along the piece is not always modeled accurately (13). Another type of approach, the
derivation of stability functions and its implementation in structure analysis programs, has demonstrated
a high degree of accuracy even using a single element per member (14).

In general, closed analytical solutions are restricted to a few particular cases. For this reason, the
numerical results prevail and therefore, the approximate nature of the solutions. In the study of nonlinear
problems there are two lines of work: one is through mathematics, in which traditional methods, such as
the continuation or homotopy methods (15; 16; 17; 18) have been used. The other is through continuum
mechanics and structures, where several techniques and solving strategies have been proposed: matrix
methods, displacement methods, compatibility, etc. Thus, the incremental methods were the first used in
the nonlinear field, using the concept of tangent stiffness matrix. They evolved following the Newton-
Raphson methods, which were completed using different control techniques to avoid non convergence
situations or error accumulation.

When designing, it is always interesting to assure that under any possible combination of loads,
bending instability does not take place. Therefore it is interesting to have an analysis method to determine
easily and accurately the maximum permissible load, usually known as the critical buckling load. The
aim of this paper is to show a new systematic method to determine the load level and the buckling mode
of any 2D-frame without the need for the simplifications usually assumed in matrix or finite element
approaches (2; 19).

To reach this target, the paper has been organized as follows. Firstly, after this introduction, we
describe the nonlinear behaviour of any beam. Then, we extend this formulation to the analysis of any
2D-frame subjected to any load and boundary condition, in order to determine the critical load and the
buckling mode. After that, we present the numerical results for three different examples. Lastly, we
resume the main conclusions obtained in this work.

2 Nonlinear beam behaviour

The set of equations that describes the nonlinear behaviour of an isolated beam will depend on the
hypothesis taken into account. In this work, we consider usual 2D-frames (2; 5; 19; 20), made of straight
and slender (with the possibility of variable cross section) beams, subjected to any load and boundary
condition under the Navier-Bernoulli model for elastic beams.

2
2.1 Equilibrium

The equilibrium conditions for every differential portion ds must be posed in the deformed configura-
tion (Fig. 1). The coordinate system to which the magnitudes are referred will be denoted by (X,Y, Z).

qY (s)

V (L)

H(L)
qX (s)
V (0)
T1
M (L)
H(0) s
T2

M (0) s ds

Y
V (s) V + dV
H + dH
H(s)

X M + dM
M (s)

θ(s)

Figure 1: Equilibrium

Considering the equilibrium of forces along the axes (X,Y ) and the equilibrium of moments along
Z−axis, we obtain the following system of equations, in terms of Piola-Kirchhoff stresses (13; 21):

 H (s) + qX (s) = 0
 0

V 0 (s) + qY (s) = 0 (1)
M (s) − H(s)θ(s) +V (s) = 0
 0


where ()0 means derivatives with respect to coordinate s. Using the hypothesis of small displacements,
sin(θ) has been approximated to θ and cos(θ) to 1.

2.2 Compatibility and constitutive equations

The equations that relate internal forces and displacements (u, v, θ) are:
 h α i
 H(s) = EA(s) u 0
(s) − (T1 + T2 )
2



θ(s) = v (s)
0

(2)
  
 α
 M(s) = EIz (s) θ0 (s) + (T1 − T2 )


h(s)

where E is Young’s modulus, and Iz (s) and A(s) are the inertia and the area of the cross-section of the
beam, respectively. The thermal effect is included, supposing a linear variation of the temperature along
the beam depth h(s) between T1 and T2 , being α the coefficient of thermal expansion.

The previous equations (Eq. 1), together with the compatibility and constitutive equations (Eq. 2),
lead to a system of linear differential equations with variable coefficients which, in general, has not an
explicit solution. The solution can be obtained for each case by means of a numerical analysis (22; 23).
Once the response of each beam is known, the response of the whole frame will be determined by
imposing the compatibility and constitutive equations between all the beam-ends of the structure.

3
2.3 Stability equations

As it is well known, the nonlinear response of each beam depends on the resolution of the system of
differential equations Eq. 1 and Eq. 2 as the loading factor λ increases, considering the boundary con-
ditions. This solution is stable and well defined. Nevertheless, there is a value of the loading factor for
which indeterminate transverse displacements suddenly appear (instant of buckling).

A method for determining the critical value λcri that causes this phenomenon consists in introducing
a small perturbation in its equilibrium, supposing an infinitesimal flexural deformation ∆u with respect
to the stable state. The structure in this new state must also satisfy Eq. 1 and Eq. 2, so the incremental
displacements (∆u, ∆v, ∆θ) and the incremental forces (∆H, ∆V, ∆M) must satisfy the following stability
equations(5):

 ∆H (s) = 0
0


∆V 0 (s) = 0 (3)
∆M (s) − H(s)∆θ(s) − θ(s)∆H(s) + ∆V (s) = 0
0



 ∆H(s) = EA(s)∆u (s)

0


∆θ(s) = ∆v0 (s) (4)
∆M(s) = EIz (s)∆θ (s)
0



Eq. 3 and Eq. 4, together with the boundary conditions of the original problem with no external
loadings, allow us to obtain λcri .

3 Frame buckling analysis. Equilibrium and compatibility

When analyzing frames, firstly it is necessary to set a common reference system for all its beams, called
global reference system (Xg ,Yg , Zg ).

Once the response of each beam is known, the next step is to discretize the frame into its beam
elements. Due to the formulation used in this work, with the assumed hypothesis, it is enough to take the
minimum number of elements, that is, one per member (the straight parts between nodes).

The magnitudes of each beam have to be expressed in the global reference system, depending on the
initial angle αb for each beam (Fig. 2), using standard matrix procedures:
     
H cos αb − sin αb 0 H
 V  =  sin αb cos αb 0 ·  V  (5)
M (Xg ,Yg ,Zg ) 0 0 1 M (X,Y ,Z)

Similarly, conditions on displacements between beam-ends at any node must be imposed, together
with the equilibrium conditions, including external forces and moments. All these external loads are
multiplied by the unique loading parameter λ, assuming proportional loading.

To determine the buckling mode we solve the stability equations (Eq. 3 and Eq. 4) according to the
loading factor λ, and then we replace one equilibrium equation by an additional condition on displace-
ments. That is, we impose an arbitrary value for one of the degrees of freedom of the frame.

Magnitudes of the pre-buckling state appear in the stability equations. Therefore, it is necessary
to solve simultaneously the constitutive equations of each beam and the compatibility and equilibrium

4
 Deformed configuration

Y Yg

X
αb Initial configuration
Xg

Figure 2: Initial orientation of the beam

equations in the nodes. These equations, together with the boundary conditions, lead a system of linear
differential equations with variable coefficients. Its solution can be found using numerical techniques,
like the Shooting Method (22; 23), as most of the times there is not a close or explicit mathematical
expression for this solution.

It is known that, in the instant of buckling, the deformed shape of the frame will correspond to the
buckling mode, so the critical load value will be the one that cancels the variation of the potential energy
functional:
1 b Li n
Z o
∆W (λ) = ∑ ∆Mi (s)∆θ0i (s) + Hi (s)[∆θi (s)]2 ds (6)
2 i=1 0
where b is the total number of members of the structure.

Therefore, the condition (∆W (λ) = 0) allows us to calculate the critical load value numerically using,
for example, the Secant Algorithm or the Newton-Raphson Algorithm (22; 23).

4 Numeric results

We present here some examples of increasing complexity, to show the methodology and the reliability
of the procedure.

4.1 Beam under compression (Validation)

Figure 3: Constant cross-section beam (case a)

Case a It is well known that the solution for this case (Fig. 3) corresponds to the problem of the fixed-
free beam, subjected to compression P. In this case, the boundary conditions of the nonlinear problem
are: (
H(0) = −P
(7)
V (0) = M(0) = u(L) = v(L) = θ(L) = 0
Integrating the system of differential equations of Eq. 1 and Eq. 4:
(
H(s) = −P
(8)
V (s) = M(s) = u(s) = v(s) = θ(s) = 0

5
On the other side, the boundary conditions of the stability equations are:
(
∆v(0) = v0
(9)
∆H(0) = ∆V (0) = ∆M(0) = ∆u(L) = ∆θ(L) = 0

where v0 is an arbitrary sufficiently small value.

The solution of the stability equations (Eq. 3 and Eq. 4) subjected to the mentioned boundary condi-
tions is:
∆H(s) = 0



EIz K 3 v0 cos (KL)



(s) =

 ∆V
KL cos (KL) − sin (KL)





EIz K 2 v0 sin (Ks)


=

−

∆M(s)
KL cos (KL) − sin (KL)





∆u(s) = 0

(10)
Kv0 [cos (KL) − cos (Ks)]


∆θ(s) =


sin (KL) − KL cos (KL)





e−iK(L+s) v0

 n
= ieiKL − ieiL(L+2s) +


 ∆v(s)
2 [KL cos (KL) − sin (KL)]




 o
+ eiKs [K(L − s) − i] + eiK(2L+s) [i + K(L − s)]



r
P
where K =
EIz

The potential energy variation functional is:

EIz K 3 v20 cos (KL)
Z Ln
1 o
∆W (P) = ∆M(s)∆θ0 (s) − P [∆θ(s)]2 ds = (11)
2 0 sin (KL) − KL cos (KL)
so the critical buckling load is:
π2 EIz
Pcri = (12)
(2L)2
because it is the value that cancels the functional. This result is the expected (2) (classical buckling
problem).

λ · q0

L = 4m Iz = 25166 · 10−8 m4
6
q0 = 10 N/m E = 2.1 · 1011 P a

Figure 4: Variable axial force (case b)

Case b In this case (Fig. 4), the beam is subjected to a load distribution that causes a linear axial force
(for example, a column subjected to its dead weight). In this problem, the resulting differential equation
has variable coefficients and it can be solved using the classical approach (roots of the determinant of the
stiffness matrix). Nevertheless, the computational effort is lower with the energy approach proposed: it
only requires the calculation of one integral. After that, the resulting critical buckling load is:

λcri = 6.47176 (13)

If we compare this value with the result provided by the FEM analysis tool (24) (based on the lineari-
zation using the geometrial stiffness matrix), we can conclude that at least 10 elements are necessary to
have results with an error less than 0.5% (Table 1).

6
 Table 1: Compressed beam. Linear axial force. Results using standard FEM code

Number of elements 2 10 50 100 200

λcri 5.81257 6.44510 6.47070 6.47151 6.47170
εr (%) 10.19 0.4119 0.01638 0.003863 0.0007726

λ · P0

L = 4m Iz (s) = π · [r(s)]4 /4
E = 2.1 · 1011 P a Iz (0) = 25166 · 10−8 m4
P0 = 106 N r(L) = 1.2 · r(0)

Figure 5: Variable cross-section beam (case c)

Case c A more interesting case, from a practical point of view, appears when some of the dimensions
of the beam are variable along the longitudinal axis (Fig. 5). The differential equations in this case are
linear, but they have variable coefficients, so the energy approach is clearly advantageous against the
matrix approach, specially when the structure is made of a large number of beams. The resulting critical
buckling load is:
λcri = 13.5635 (14)
As it can be seen in Table 2, we need very fine discretizations with the matrix approach to obtain similar
numerical values.
Table 2: Compressed beam. Variable cross-section beam. Standard FEM results

Number of elements 2 10 50 100 200

λcri 10.9825 13.0753 13.4672 13.5154 13.5395
εr (%) 19.03 3.599 0.7080 0.3546 0.1769

Case d In the last validation example we consider the instability by thermal effect of a fixed-fixed
beam (Fig. 6). First we check (Table 3) that the result obtained for the beam with constant dimensions
(with Iz (s) = Izmax ) coincides with the theoretical one:
(2π)2 EIz
Ncri = = EAαλcri ∆T (15)
L2
λcri = 46.0077 (16)

For the variable cross-section beam, as specified in Fig. 6, the solution is:
λcri = 35.9735 (17)

In Table 3, we compare the solution of both problems with the one obtained using the standard FEM
code. The buckling mode is indicated in each figure in a thick line.

The most important aspect of this approach is its generality, because it considers any type of loading,
the effect of dead weight, thermal effects and variable depth sections in the same way, with no additional
effort. Now we apply the proposed method to some classical examples.

4.2 Lee frame (Application 1)

The structure of Fig. 7, called ‘Lee frame’ (25), shows in a clear and easy way the objective of this work
and the possibilities of generalization of the numerical technique used.

7
 T1

T2

L = 4m T1 = 40 ◦ C A(s) = π · [r(s)]2 r(s) = a + b · s + c · s2

E = 2.1 · 1011 P a T2 = 0 ◦ C Iz (s) = π · [r(s)]4 /4 r(L/2) = 0.8 · r(0)
α = 1.2 · 10−5 ◦ C−1 ∆T = (T1 − T2 ) = 40 ◦ C Iz (0) = 25166 · 10−8 m4 r(L) = r(0)

Figure 6: Temperature (case d)

Table 3: Fixed-fixed beam. Thermal effect. Standard FEM results

Number of elements 2 10 50 100 200

λcri 38.5335 36.0012 35.9883 35.9881 35.9881
Constant cross-section
εr (%) 7.116 0.07700 0.04114 0.04059 0.04059
λcri 46.0547 46.0431 46.0272 46.0261 46.0242
Variable cross-section
εr (%) 0.1022 0.07694 0.04238 0.03999 0.03586

We consider, for simplicity, that both beams of the frame have the same lenght (L), material (E)
and section (h(s), A(s), Iz (s)), and we suppose a rigid joint in section b. Applying the method for the
numerical values shown, we obtain the following critical load factor for constant cross-section (of Iz =
Izmax ):
λcri = 47.5063 (18)
Whereas for the frame with variable cross-section shown in Fig. 7, the critical loading factor is:

λcri = 14.5092 (19)

To achieve accurate results (error less than 0.5%) using standard FEM codes, at least 50 elements per
member would be necessary in the first case, and 200 in the second.

4.3 Four beam frame (Application 2)

Finally, we apply the proposed method to the frame of Fig. 8. With a four-element discretization
(elements coinciding only with beam members) and assuming a proportional loading state (except the
temperature and dead weight, which remain constant), the critical value for a constant cross-section
(Iz (s) = Izmax ) is:
λcri = 29.1906 (20)
and for the case of Fig. 8, with a variable cross-section, the critical buckling load is:

λcri = 9.39849 (21)

Again, a large number of elements would be necessary to achieve similar results using standard FEM
codes.

5 Conclusions

The main conclusions of this work can be summarized as follows:

Firstly, for the mathematical model, the starting point has been Euler’s buckling theory. It assumes
small strains and displacements, and leads to the concepts of critical load and buckling mode, as a
consequence of the mathematical approach. Therefore, it is not possible to know the displacements for
the buckling mode or to determine post-buckling behaviour.

8
 λ · q0

λ · P0

T1

T2

L = 4m
P0 = 106 N
q0 = 103 N/m r(s) = a + b · s
T1 T2 E = 2.1 · 1011 P a rmin = 0.6 · rmax
α = 1.2 · 10−5 ◦ C−1 A(s) = π · [r(s)]2
ρ = 7850kg/m 3
Iz (s) = π · [r(s)]4 /4
g = 9.81m/s2 Iz (0) = 25166 · 10−8 m4
T1 = 40 ◦ C
T2 = 20 ◦ C

Figure 7: Lee frame. Variable inertia beams

λ · P2

T1
T1
λ · P1 λ · P1
T2
λ·F T2 β

F = 2.4 · 105 N E = 2.11 · 1011 P a

P1 = 3.0 · 105 N α = 1.2 · 10−5 ◦ C−1

λ·q P2 = 2 · P1 ρ = 7850kg/m3 L λ·q

T1 T2 q = 1.0 · 103 N/m g = 9.81m/s2 T1 T2
T1 = 40 ◦ C L = 4m
T2 = 20 ◦ C β = 15 ◦

L L
2
A(s) = π · [r(s)] r(s) = a + b · s
Iz (s) = π · [r(s)]4 /4 rmin = 0.6 · rmax
Izmax = 25166 · 10−8 m4

Figure 8: Four beam frame. Variable inertia beams

Secondly, with regard to the analysis tool that we have developed, we can conclude that it is a
systematic and easy tool which allows us to study more general cases, such as variable section beams,
distributed loads, 3D-structures, etc. Although the computer implementation has been developed within
a symbolic manipulation software, it is not possible, in general, to obtain analytical solutions according
to all the parameters, but only numerical solutions.

Lastly, with regard to the numerical results, we confirm that, thanks to an exact mathematical for-
mulation, we obtain an advantage when calculating: we have accurate results with a minimum number
of elements, and the results are almost the same for finer discretizations, with the computational effort
being significantly smaller.

References
[1] S. P. Timoshenko. History of Strength of Materials. Dover Publ. Inc., New York, USA, 1953. 1

[2] A. Chajes. Principles of Structural Stability Theory. Prentice Hall, New Jersey, USA, 1974. 1, 2, 6

9
 [3] W. T. Koiter. Post-buckling analysis of a simple two bar frame. Recent Prog. in App. Mech. - The
Folke Odqvist Volume, pages 337–354, 1967. 2
[4] S. P. Timoshenko. Strength of Materials. Krieger Publishing Company, Florida, USA, 1976. 2
[5] G. J. Simitses and D. H. Hodges. Fundamentals of Structural Stability. Elsevier Inc., Oxford, UK,
2006. 2, 4
[6] C. H. Popelar. Optimal Design of Beams against Buckling: A Potential Energy Approach. Mecha-
nics Based Design of Structures and Machines, 4(2):181–196, 1976. 2
[7] Z. P. Bazant. Structural Stability. Int. J. Solids and Structures, 37:55–67, 2000. 2
[8] W. H. Wittrick and F. W. Williams. An Algorithm for Computing Critical Buckling Loads of Elastic
Structures. Mechanics Based Design of Structures and Machines, 1(4):497–518, 1973. 2
[9] W. F. Chen and T. Atsuta. Theory of Beam-Columns, volume 2. McGraw-Hill, New York, USA,
1977. 2
[10] W. R. Spillers and R. Levy. Optimal design for plate buckling. J. Struct. Engng., 116:850–858,
1990. 2
[11] J. H. Argyris. Recent Advances in Matrix Methods of Structural Analysis. Pergamon Press, Oxford,
UK, 1964. 2
[12] M. J. Turner, R. W. Clough, H. C. Martin, and L. J. Topp. Stiffness and deflection analysis of
complex structures. Jour. Aerona. Sci., 23:805–823, 1956. 2
[13] M. A. Crisfield. Non-linear Finite Element Analysis of Solids and Structures. Vol.1: Essentials.
John Wiley & Sons, New Jersey, USA, 2000. 2, 3
[14] B. N. Alemdar and D. W. White. Displacement, Flexibility and Mixed Beam-Column Finite Ele-
ment Formulations for Distributed Plasticity Analysis. J. Struct. Engng., 131(12):1811–1819, 2005.
2
[15] D. F. Davidenco. On a New Method of Numerical Solution of Systems of Nonlinear Equations.
Dokl. Akad. Nauk., 88:601–602, 1953. 2
[16] A. Eriksson. Fold Lines for Sensitivity Analyses in Structural Instability. Comput. Methods Appl.
Mech. Engn., 114:77–101, 1994. 2
[17] C. A. Felippa. Introduction to the analysis of nonlinear elastic structures by the Finite Element
Method. University of Colorado at Boulder, 1986. 2
[18] J. W. Vick. Homology theory: an introduction to algebraic topology. Springer-Verlag, New York,
USA, second edition, 1994. 2
[19] S. P. Timoshenko. Theory of Elastic Stability. McGraw-Hill, New York, USA, 1963. 2
[20] D. H. Hodges. Nonlinear beam theory for engineers. American Institute of Aeronautics and Astro-
nautics, Reston, USA, 2006. 2
[21] W. T. Koiter. On the principle of stationary complementary energy in the nonlinear theory of
elasticity. SIAM J. Appl. Math., 25(3):424–434, 1973. 3
[22] R. L. Burden and J. D. Faires. Numerical Analysis. Thomson, Mexico, 1998. 3, 5
[23] D. G. Fertis. Nonlinear structural engineering. Springer-Verlag, New York, USA, 2006. 3, 5
[24] COSMOS/M. v.2.95. Los Angeles: Structural Research and Analysis Corp., 2006. 6
[25] H. Chen and G. E. Blandford. Work-increment-control method for non-linear analysis. Int. Jour. of
Num. Meth. in Eng., 36:909–930, 1993. 7

10

View publication stats