Académique Documents
Professionnel Documents
Culture Documents
Abstract: Nonlinear solution of reinforced concrete structures, particularly complete load-deflection response, requires
tracing of the equilibrium path and proper treatment of the limit and bifurcation points. In this regard, ordinary solution
techniques lead to instability near the limit points and also have problems in case of snap-through and snap-back. Thus they
fail to predict the complete load-displacement response. The arc-length method serves the purpose well in principle, received
wide acceptance in finite element analysis, and has been used extensively. However modifications to the basic idea are vital
to meet the particular needs of the analysis. This paper reviews some of the recent developments of the method in the last two
decades, with particular emphasis on nonlinear finite element analysis of reinforced concrete structures.
Key words: Arc-length method, Nonlinear analysis, Finite element method, Reinforced concrete, Load-deflection path
Document code: A CLC number: TU31
idea of the method can be explained as follows. arc-length, and can be written in differential form
Since the method treats the load-factor as a as
variable, it becomes an additional unknown in
equilibrium equations resulting from finite element s = ∫ dpT dp + dλ 2ψ 2 q T q (2)
procedure, and yield (N+1) unknowns, where N is
the number of elements in the displacement vector.
or in increment form, written as
The solution of (N+1) unknowns requires an addi-
tional constraint equation expressed in terms of
a = ∆p T ∆p + ∆λ 2ψ 2 q T q − ∆l 2 = 0 (3)
current displacement, load-factor and arc-length.
Two approaches, fixed arc-length and varying
arc-length are generally used. In the former the where ∆p is vector of incremental displacement, ∆λ
arc-length is kept fixed for current increment, is incremental load-factor, ∆l is fixed radius of
whereas in the latter case, new arc-length is evalu- desired intersection, and ψ is the scaling parameter
ated at the beginning of each load step to ensure the for loading terms. With some simplification Eqs.(1)
achievement of the solution procedure. Simplifica- and (3) can be directly used to compute the iterative
tion of the constraint equation leads to a quadratic change in displacement vector and load-factor, and
equation, whose roots are used for determining the are written as
load-factor. Proper selection of root is one of the
−1
key issues of the method, whose details will be δ p KT −q gold
discussed in subsequent sections. Generally, for the = − 2 T (4)
δλ 2∆p 2∆λψ q q aold
T
gi (λi ) = fi − λi q (1) and the incremental displacement for the next in-
crement can be written as
where fi is vector of internal equivalent nodal forces,
q is the external applied load vector, λ is the ∆pi +1 = ∆pi + δ p (6)
load-level parameter, and gi is out-of-balance force
vector. The arc-length method is aimed to find the Substituting values from Eqs.(5) and (6) into the
intersection of Eq.(1) with constant s termed as the constraint equation yields the expression
620 Memon et al. / J Zhejiang Univ SCI 2004 5(5):618-628
(p)
where
c1 = δ pTTδ pT + ψ 2 q T q
c2 = 2δ pT (∆p + δ g ) + 2∆λψ 2 q T q (8)
c3 = (∆p + δ g ) (∆p + δ g ) − ∆l + ∆λ ψ q q
T 2 2 2 T
1971)
(p)
∆λoψ q ) (Fig.3); the technique of Ramm is closely After the introduction of arc-length method by
related to the work of Riks and Wempner. To avoid Riks (1972; 1979) and Wempner (1971), the me-
the dependency of solution process on either pre- thod received wide acceptance in the field of finite
dictor or secant change, Fried (1984) suggested use element analysis. However as Crisfield (1981)
of (δ pT ,(1/ψ 2 q T q)) instead of (∆po ,∆λo ). All these stated, the Riks method was not suitable for stan-
techniques are termed as linearised versions of dard finite element analysis even with modified
arc-length method, in which the constraint equation Newton-Raphson (mN-R) procedure, because equ-
leads to only one solution, hence no issue of selec- ations proposed by Riks destroy the banded nature
tion of roots. However it is possible that the method of the stiffness matrix. For one-dimensional prob-
may sometime miss the equilibrium path, and lead lem with N displacement variables, Crisfield (1981)
to numerical difficulties. gave the modification of the method and suggested
Memon et al. / J Zhejiang Univ SCI 2004 5(5):618-628 621
the fixation of incremental length ∆l during load where gi is the out-of-balance force vector at the
increment. Therefore instead of applying the con- end of the previous iteration. The near optimum
straint Eq.(3), the following equation was used value of µ is suggested as 0.8. The procedure works
well; but if Eq.(12) is violated, then a new
∆piT ∆pi = ∆l 2 (9) arc-length ηi,2 must be tried as a second attempt to
satisfy the equation. The simplest method to com-
The proposed technique is termed as cylin- pute second attempt is to use linear interpolation or
drical arc-length method. Solution of Eq.(9) yields extrapolation using s1 and s0 values.
to a quadratic equation similar to Eq.(7) but with ψ Forde and Stiemer (1987) introduced a general
= 0, and thus two roots. To avoid doubling back on arc-length procedure based on orthogonality prin-
original load/deflection path, Crisfield suggested ciples. In their work the authors suggested selection
that the angle between the incremental displace- of an arbitrary direction n(i) with reference to tan-
ment vector before the current iteration and in- gent t(i) of the current incremental load-displacement
cremental load vector after the current iteration configuration. The scalar product of these vectors
must be minimum. To achieve this, the appropriate yields a residual g(i). The tangential and normal
root is the one that gives positive angle, and in case vectors (t and n) consist of m dimensions from the
that both roots are positive, the appropriate root is displacement vector and one dimension from the
the one that is closest to the linear solution. Cris- load parameter. These components are combined
field applied this method for large-deflection using a scaling factor β to form vectors with m+1
elasto-plastic analysis of imperfectly stiffened dimension, which can be written as
plates and shells and got satisfactory results.
Crisfield (1983) pointed out that although the t (i ) = u( i ) + βλ (i ) (13)
arc-length method works well and has been suc-
n = ∆u + β ∆ λ
(i )
(14)
cessfully incorporated into finite element pro-
gramming, yet only partial success was achieved in
solving the material nonlinearity of beams and The scalar product of t(i) and n(i) results in the gen-
slabs with significant strain softening. However the eral expression for ∆λ, as follows
same problem when solved by the displacement T
Eq.(10) is too stringent a condition to meet in ARC-LENGTH METHOD DURING THE 90’S
practice and instead it is desirable to satisfy
Al-Rasby (1991) gave modified arc-length
s j (ηi , j ) < µ s0 (ηi ,0 = 0) (11) method almost similar to that of Forde and Stiemer
(1987) but introduced scaling matrices for calcula-
where tion of the arc-length for the purpose of non-dimen-
s0 (ηi ,0 = 0) = δ iT gi ( pi ) (12) sionalizing the vectors that define the arc-length
622 Memon et al. / J Zhejiang Univ SCI 2004 5(5):618-628
constraint equations. These matrices can also be length method given by Crisfield (1981) and Ramm
used to gauge the relative contribution of load and (1981). Fafard and Massicotte in their work used
displacement components to the arc- length con- the constraint equation given by Ramm (1981), and
straint equation. Based on the idea, the equation for used updated hyper-plane technique to evaluate
∆λ takes the form arc-length as follows
{∆pi } { K −1 g}
T
g (i ) − U (i ) Vsδ ( i )
T
∆λ = (16) ∆λ = (17)
{∆pi } { K −1q}
T
U ( i ) Vsδ ( i ) + λ ( i ) Rref
T
Ws Rref T
where Vs is the displacement scaling diagonal ma- and the current displacement was computed as
trix, and Ws is the load scaling diagonal matrix. By given in Eq.(6). This displacement does not fall on
varying the residual g, all of the known arc-length the defined hypersphere; so to bring it on the hy-
methods can be used. With proper variation of Vs persphere, a reduction in the computed displace-
and Ws, load control and displacement control ment is suggested, as
methods are the special cases of this technique.
Since variation of the scaling diagonal matrices
∆pi +1 = ∆l = α ∆pi +1 (18)
required to follow the solution process, the varia-
tion can be done in infinite variety. However the from which
author suggests the following four types of varia- ∆l
α= (19)
tions: ∆pi +1
i) Vs ≡ Ws ≡ I. Setting both diagonal scaling
matrices equal to identity means that all parameters Hence the desired incremental displacement can be
are considered equally. This may not be the ideal expressed as
case, since on the load-displacement path, at par-
ticular points some parameters are more significant {∆pi +1} = α {∆pi } + α {δ i } (20)
than others. Also this procedure does not introduce
any scaling on either load or displacement com- Since the technique combines the advantages
ponent. of two methods, it can be effectively used for
ii) Vs ≡ I and [Ws]i,i = ( Rref )t−2 . No sum on i. load-deflection tracing in nonlinear analysis and it
T
This is equivalent to specifying Rref Ws Rref ≡ I. can also overcome both the individual drawbacks of
the other methods and the convergence problem;
iii) Vs ≡ I and Ws = cI,
however near failure point convergence can cause
δ (1)Tδ (1) problem, which can be avoided by reducing the ∆l
Where c = 2 T CSP
∆λ1 Rref Rref value. In addition Fafard and Massicotte (1993)
λ (1) δ (1) Rref gave the geometrical representation of arc-length
and CSP = (i ) where δ(i) is the initial methods introduced by Crisfield and Ramm, which
∆λ1 U Rref
helps to visualize the method geometrically.
displacement vector in the first step, ∆λ1 is the
Carrera (1994) gave a modification for selec-
initial load parameter in the first step and CSP is the
tion of proper root of the nonlinear constraint
current stiffness parameter.
equation. The author argues that although for some
iv) Vs = [K]i,i and Ws= [ K ]i−,1i , no sum on i. In problems arc-length method works well; yet it fails
this scaling procedure the diagonal terms of the in some cases, and mostly it is due to the selection
stiffness matrix are used as scaling parameters. of the proper root of the governing nonlinear con-
Fafard and Massicotte (1993) modified the straint equation. Thus Carrera proposed selection of
method by making use of the advantages of arc- root that is closest to constraint linear solution. The
Memon et al. / J Zhejiang Univ SCI 2004 5(5):618-628 623
dominant element in which some or all of the ma- minimum residual norm. However to preserve
terial is either in damaged zone or in failure zone. computer resources the idea is applied only when
Thus the relative displacement vector for a plane sharp snap-back is observed, otherwise the usual
element with n degrees of freedom is given as method is continued.
Regarding the issue of selection of root when
∆δ = [δ1 − δ n , δ 2 − δ1 , δ 3 − δ 2 ,..., δ n − δ n −1 ]T (27) divergence occurs, Kweon and Hong (1994) dis-
carded every existing possibility of selection of
where δ1 to δn are the displacements at nodes within roots and suggested restarting of the corresponding
the element. In adjacent dominant elements relative load step with arc-length reduced to half. To pre-
displacement of common nodes is computed twice, vent the number of iterations being too large the
since each element has its own relative displace- authors also suggested that the maximum number
ment vector. The equations for arc-length and load- of iterations be fixed. If the number of iterations
factor are then computed as exceeds the maximum number, then the load step is
restarted with arc-length reduced to half. Generally
m the direction of load step is evaluated according to
∑ ( ∆δ ) ( ∆δ )
T
1 e i e = ∆l 2 (28) the sign of the determinant of stiffness matrix, but
e =1
Kweon and Hong (1994) suggested that the load
m
direction must be independent of the sign of the
∑ ( ∆U ) ( ∆δ + ∆U F )e
T
P e i −1
∆λi = ∆λ1 − e =1
(29) determinant of the stiffness matrix; and imposed
m
the condition that the first incremental displace-
∑ ( ∆U ) ( ∆U )
T
P e P e
e =1
ment vector in the present load step should make an
in which acute angle with the total incremental displacement
∆l vector at the last step. They argued that the criterion
∆λ1 = (30) leads to better results, and that numerical problems
m
∑ ( ∆U ) ( ∆U )
T
P e P e for post buckling analysis do not occur. Although
e =1
the reported results of post buckling analysis of
isotropic shell under compression showed the va-
where ∆UP is the relative reference deformation in lidity of the reported procedure, the method will
the element and ∆UF is the relative incremental probably lead to slower convergence as the arc-
deformation due to unbalanced force. The tech- length is reduced to half of previous or no con-
nique mentioned successfully predicts the load- vergence at all for problems with critical points.
displacement behavior of materials with strain While performing geometrical nonlinear post-
softening as mentioned in reported results. How- buckling analysis of truss members Kuo and Yang
ever, the reported method was applied to plain (1995) found that most existing arc-length control
concrete element and there was no mention of the methods failed to predict load-displacement re-
use of the method for compression softening, which sponse, and argued that it is because iterations were
is frequent and dominant in concrete structures. not performed in proper direction for problems with
Hellweg and Crisfield (1998) also attempted to multi-winding loops. To overcome this problem,
solve a double cantilever beam with softening type Kuo and Yang introduced two control parameters
damage model and reported divergence of the so- for detecting change in direction and for guiding the
lution even if proper root for cylindrical arc-length direction of iteration. It was accomplished by
method was selected. The authors argue about the obtaining the dot product of two adjacent tangent
situation when divergence is due to the sharp vectors; and once negative value is obtained, the
snap-back, in which neither root can lead the solu- direction of loading should be reversed. The tech-
tion on the proper path. Therefore they suggested nique has advantage in that negative value will only
that the selection of proper root be based on the be obtained when the control passes a limit point;
626 Memon et al. / J Zhejiang Univ SCI 2004 5(5):618-628
however for large variation in curvature it is pos- ral and reinforced concrete in particular, is that the
sible that the above parameter may not yield nega- initial solution which makes the basis of further
tive value. In such cases a vector directed along the iterations, may be well away from the final equi-
secant line of N-dimensional curve is defined. Then librium state, and thus can lead to numerical dif-
the dot product of two adjacent tangent vectors ficulties and/or divergence of solution process. In
along this newly defined secant vector will give the addition strain-softening is one of the major effects
correct sign. The validation of the proposed tech- in reinforced concrete analysis, which must be
nique is given by solving numerical examples. included to produce a realistic response. The effect
can lead to localized failure of material and often
exhibit snap-back behavior in load-deformation
APPLICATION IN NONLINEAR ANALYSIS OF space. In such cases, traditional methods fail to
REINFORCED CONCRETE STRUCTURES trace the complete load-deflection response of re-
inforced concrete structures. Therefore the arc-
Most nonlinear analyses of reinforced con- length method is used in combination with New-
crete structures use Newton-Raphson (NR) or ton-Raphson method to trace the complete response
modified Newton-Raphson (m-NR) method to trace in load-deformation space. The method has been
the equilibrium path. Modified Newton-Raphson successfully applied by several researchers for the
method leads to some simplification, but conver- analysis of reinforced concrete structures, who
gence of the solution process is slow. Both of these reported good agreement between experimental and
methods work well with linear or bilinear material analyzed results. In the following table, a few
relationship, but become inefficient with higher examples are summarized to show the validity of
degree of nonlinearity, i.e., cracking, bond-slip, the method. Details of problems included are given
material nonlinear behavior, time dependent effects, in relevant references.
etc., are introduced. Furthermore a disadvantage of Table 1 summarizes five examples, among
these methods is that without some special tech- which examples 1 to 3 were analyzed by Foster
nique the descending branch of the load-displace- (1992) and examples 4 and 5 were analyzed by
ment path cannot be traced. An early method of Crisfield (1983). All the examples included were
dealing with limit points was to use the displace- compared with the modified Newton-Raphson
ment-control of Baltoz and Dhatt (1979). The method to demonstrate the validity and effective-
method has been applied successfully in many ness of the arc-length method. For all five examples
situations; however the method is not suitable for mentioned, the modified Newton-Raphson method
problems with many degrees of freedom, which is almost failed to converge at the early stages. In
quite common in case of nonlinear analysis of re- addition Table 1 also shows the faster convergence
inforced concrete structures. A further problem rate of the solution process when the arc-length
when trying to apply the method to highly material method is combined with line searches.
nonlinear finite elements, such as concrete in gene- Lam and Morley (1992) used their modified arc-
length method to analyze doubly reinforced con- Mechanics, Tapir press, p.647-669.
crete beam with fixed and sliding support and a Bergan, P.G., Horrigmoe, G., Krakeland, B., Soreide, T.H.,
1978. Solution techniques for nonlinear finite element
cantilever beam with nearly uniform bending mo-
problems. International Journal for Numerical
ment. For both of the examples, strain-softening Methods in Engineering, 12:1677-1696.
constitutive relation was incorporated in the ana- Carrera, E., 1994. A study on arc-length type methods and
lysis which produced singular points on the load- their operation failures illustrated by a simple model.
deflection path and strain localization. Due to the Computer and Structures, 50(2):217-229.
use of the arc-length method, it was possible to Crisfield, M.A., 1981. A fast incremental/iterative solution
procedure that handles snap-through. Computer and
trace the complete load-deflection path of the ex-
Structures, 13:55-62.
amples which otherwise was not possible with tra- Crisfield, M.A., 1983. An arc-length method including line
ditional methods. searches and accelerations. International Journal for
Numerical Methods in Engineering, 19:1269-1289.
Fan, Z.L., 1994. A study of variable step-length incre-
CONCLUSION mental/iterative methods for nonlinear finite element
equations. Computer and Structures, 52(6):1269-
1275.
The arc-length method has become a powerful Fafard, M., Massicotte, B., 1993. Geometrical interpreta-
tool to use with finite element formulation for tion of the arc-length method. Computer and Struc-
complete analysis of the load-deflection path. After tures, 46(4):603-615.
introduction, the method got high praise of re- Feng, Y.T., Peric, D., Owen, D.R.J., 1996. A new criterion
searchers and many research papers on it have been for determination of initial loading parameter in
arc-length methods. Computer and Structures, 58(3):
published. This paper reviews some of the modi-
479-485.
fications and provides in-depth insight of the Forde, B.W.R., Stiemer, S.F., 1987. Improved arc-length
method. In addition to papers on the modification orthogonality methods for nonlinear finite element
of the existing method, a good number of papers analysis. Computer and Structures, 27(5):625-630.
have also been published for use of the method in Foster, S., 1992. An application of the arc-length method
finite element analysis. It can be seen from the involving concrete cracking. International Journal for
Numerical Methods in Engineering, 33: 269-285.
review that the method is performing quite well, but
Fried, I., 1984. Orthogonal trajectory accession to the non-
to the best knowledge of the authors, the validity of linear equilibrium curve. Computer Methods in Ap-
the method has not reported for three-dimensional plied Mechanics and Engineering, 47:283-298.
solid modeling of reinforced concrete structures in Hellweg, H.B., Crisfield, M.A., 1998. A new arc-length
general or for strain softening in compression in method for handling sharp snap-backs. Computer and
particular. Structures, 66(5):705-709.
Kuo, S.R., Yang, Y.B., 1995. Tracing post buckling paths of
structures containing multi-loops. International Jour-
References nal for Numerical Methods in Engineering, 38:4053-
Al-Rasby, S.N., 1991. Solution techniques in nonlinear 4075.
structural analysis. Computer and Structures, 40(4): Kweon, J.H., Hong, C.S., 1994. An improved arc-length
985-993. method for post-buckling analysis of composite cy-
Baltoz, J.L., Dhatt, G., 1979. Incremental displacement lindrical panels. Computer and Structures, 53(3):
algorithms for nonlinear problems. International 541-549.
Journal for Numerical Methods in Engineering, 14: Lam, W.F., Morley, C.T., 1992. Arc-length method for
1262-1266. passing limit points in structural calculation. Journal
Bellini, P.X., Chulya, A., 1987. An improved automatic of Structural Engineering, 118(1):169-185.
incremental algorithm for the efficient solution of May, I.M., Duan, Y., 1997. A local arc-length procedure for
nonlinear finite element equations. Computer and strain softening. Computer and Structures, 64(1-4):
Structures, 26(1-2):99-110. 297-303.
Bergan, P.G., Soreide, T.H., 1978. Solution of Large Dis- Ramm, E., 1981. Strategies for Tracing the Nonlinear Re-
placement and Instability Problems Using the Current sponse near Limit Points. In: Nonlinear Finite Element
stiffness Parameter. In: Finite Element in Nonlinear Analysis in Structural Mechanics. Springer, New York,
628 Memon et al. / J Zhejiang Univ SCI 2004 5(5):618-628
http://www.zju.edu.cn/jzus
Journal of Zhejiang University SCIENCE (ISSN 1009-3095, Monthly)
♦ The Journal has been accepted by Ei Compendex, Index Medicus/MEDLINE, CA, BIOSIS,
AJ, CBA, ZB1, INSPEC, and CSA for abstracting and indexing respectively, since founded in
2000.
♦ The Journal aims to present the latest development and achievement in scientific research in
China and overseas to the world’s scientific community.
♦ The Journal is edited by an international board of distinguished foreign and Chinese scientists.
♦ The Journal mainly covers the subjects of Science & Engineering, Life Sciences & Biotech-
nology.
♦ A thoroughly internationalized standard peer review system is an essential tool for this Jour-
nal’s development.
The editors welcome your opinions & comments on, your contributions to, and subscription
of the journal.
Please write to: Helen Zhang, Managing Editor of JZUS
E-mail: jzus@zju.edu.cn Tel/Fax: 86-571-87952276
English Editorial Office, Journal of Zhejiang University SCIENCE
20 Yugu Road, Hangzhou 310027, China
● Individual US $200/ ¥200 (12 issues/year);
● Institutional US $240/¥240(12 issues/year)