Skip to main content
Log in

Optimization formulations for the maximum nonlinear buckling load of composite structures

  • Research Paper
  • Published:
Structural and Multidisciplinary Optimization Aims and scope Submit manuscript

Abstract

This paper focuses on criterion functions for gradient based optimization of the buckling load of laminated composite structures considering different types of buckling behaviour. A local criterion is developed, and is, together with a range of local and global criterion functions from literature, benchmarked on a number of numerical examples of laminated composite structures for the maximization of the buckling load considering fiber angle design variables. The optimization formulations are based on either linear or geometrically nonlinear analysis and formulated as mathematical programming problems solved using gradient based techniques. The developed local criterion is formulated such it captures nonlinear effects upon loading and proves useful for both analysis purposes and as a criterion for use in nonlinear buckling optimization.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13
Fig. 14
Fig. 15

Similar content being viewed by others

References

  • Abrate S (1994) Optimal design of laminated plates and shells. Compos Struct 29:269–286

    Article  Google Scholar 

  • Almroth BO, Brogan FA (1972) Bifurcation buckling as an approximation of the collapse load for general shells. AIAA J 10(4):463–467

    Article  Google Scholar 

  • Bendsøe MP, Sigmund O (2003) Topology optimization—theory, methods and applications, 2nd edn. Springer, Berlin. ISBN: 3-540-42992-1

    Google Scholar 

  • Bendsøe MP, Olhoff N, Taylor J (1983) A variational formulation for multicriteria structural optimization. J Struct Mech 11:523–544

    Google Scholar 

  • Borri C, Hufendiek HW (1985) Geometrically nonlinear behaviour of space beam structures. J Struct Mech 13:1–26

    Google Scholar 

  • Brendel B, Ramm E (1980) Linear and nonlinear stability analysis of cylindrical shells. Comput Struct 12:549–558

    Article  MATH  Google Scholar 

  • Brush DO, Almroth BO (1975) Buckling of bars, plates, and shells. McGraw-Hill, Kogakusha. ISBN: 07-085028-3

    MATH  Google Scholar 

  • Bushnell D (1985) Computerized buckling analysis of shells. Martinus Nijhoff, Dordrecht. ISBN: 90-247-3099-6

    Google Scholar 

  • Courant R, Hilbert D (1953) Methods of mathematical physics, vol 1. Interscience, New York

    Google Scholar 

  • Crisfield MA (1981) A fast incremental/iterative solution procedure that handles “snap-through”. Comput Struct 13:55–62

    Article  MATH  Google Scholar 

  • Dvorkin EN, Bathe KJ (1984) A continuum mechanics based four-node shell element for general non-linear analysis. Eng Comput 1:77–88

    Article  Google Scholar 

  • Foldager JP, Hansen JS, Olhoff N (2001) Optimization of the buckling load for composite structure taking thermal effects into account. Struct Multidisc Optim 21:14–31

    Article  Google Scholar 

  • Harnau M, Schweizerhof K (2002) About linear and quadratic “solid-shell” elements at large deformations. Comput Struct 80(9–10):805–817

    Article  Google Scholar 

  • Hinton E (ed) (1992) NAFEMS introduction to nonlinear finite element analysis. Bell and Bain, Glasgow. ISBN 1-874376-00-X

    Google Scholar 

  • Hu HT, Wang SS (1992) Optimization for buckling resistance of fiber-composite laminate shells with and without cutouts. Compos Struct 22(2):3–13

    Article  Google Scholar 

  • Hu HT, Yang JS (2007) Buckling optimization of laminated cylindrical panels subjected to axial compressive load. Compos Struct 81:374–385

    Article  Google Scholar 

  • Hyer MW, Lee HH (1991) The use of curvilinear fiber format to improve buckling resistance of composite plates with central circular holes. Compos Struct 18(5/6):239–261

    Article  Google Scholar 

  • Johansen L, Lund E, Kleist J (2009) Failure optimization of geometrically linear/nonlinear laminated composite structures using a two-step hierarchical model adaptivity. Comput Methods Appl Mech Eng 198(30–32):2421–2438

    Article  Google Scholar 

  • Jones RM (2006) Buckling of bars, plates, and shells. Bull Ridge. ISBN: 0-9787223-0-2

  • Klinkel S, Gruttmann F, Wagner W (1999) A continuum based three-dimensional shell element for laminated structures. Comput Struct 71(1):43–62

    Article  Google Scholar 

  • Lee SJ, Hinton E (2000) Dangers inherited in shells optimized with linear assumptions. Comput Struct 78:473–486

    Article  Google Scholar 

  • Lin CC, Yu AJ (1991) Optimum weight design of composite laminated plates. Comput Struct 38(5/6):581–587

    Article  MATH  Google Scholar 

  • Lindgaard E, Lund E (2010a) Nonlinear buckling optimization of composite structures. Comput Methods Appl Mech Eng 199(37–40):2319–2330. doi:10.1016/j.cma.2010.02.005

    Article  MathSciNet  Google Scholar 

  • Lindgaard E, Lund E (2010b) A unified approach to nonlinear buckling optimization of composite structures. Comput Struct. doi:10.1016/j.compstruc.2010.11.008

    MATH  Google Scholar 

  • Lindgaard E, Lund E, Rasmussen K (2010) Nonlinear buckling optimization of composite structures considering “worst” shape imperfections. Int J Solids Struct 47:3186–3202. doi:10.1016/j.ijsolstr.2010.07.020

    Article  MATH  Google Scholar 

  • Lund E (2009) Buckling topology optimization of laminated multi-material composite shell structures. Compos Struct 91:158–167

    Article  Google Scholar 

  • Lund E, Stegmann J (2005) On structural optimization of composite shell structures using a discrete constitutive parametrization. Wind Energy 8:109–124. doi:10.1002/we.132

    Article  Google Scholar 

  • Mateus HC, Soares CMM, Soares CAM (1997) Buckling sensitivity analysis and optimal design of thin laminated structures. Comput Struct 64(1–4):461–472

    Article  MATH  Google Scholar 

  • Moita JS, Barbosa JI, Soares CMM, Soares CAM (2000) Sensitivity analysis and optimal design of geometrically non-linear laminated plates and shells. Comput Struct 76:407–420

    Article  Google Scholar 

  • Noguchi H, Hisada T (1993) Sensitivity analysis in post-buckling problems of shell structures. Comput Struct 47(4/5):699–710

    Article  MATH  Google Scholar 

  • Olhoff N (1989) Multicriterion structural optimization via bound formulation and mathematical programming. Struct Optim 1:11–17

    Article  Google Scholar 

  • Overgaard LCT, Lund E (2005) Structural design sensitivity analysis and optimization of vestas V52 wind turbine blade. In: Proc. 6th world congress on structural and multidisciplinary optimization

  • Overgaard LCT, Lund E, Thomsen OT (2010) Structural collapse of a wind turbine blade. Part A: Static test and equivalent single layered models. Compos Part A Appl Sci Manuf 41(2):257–270. doi:10.1016/j.compositesa.2009.10.011

    Article  Google Scholar 

  • Singer J, Arbocz J, Weller T (1998) Buckling experiments: experimental methods in buckling of thin-walled structures, vol 1. Wiley, New York. ISBN: 0-471-95661-9

    MATH  Google Scholar 

  • Svanberg K (1987) Method of moving asymptotes—a new method for structural optimization. Int J Numer Methods Eng 24:359–373

    Article  MathSciNet  MATH  Google Scholar 

  • Taylor JE, Bendsøe MP (1984) An interpretation of min-max structural design problems including a method for relaxing constraints. Int J Solids Struct 20(4):301–314

    Article  MATH  Google Scholar 

  • Thompson JMT, Hunt GW (1973) A general theory of elastic stability. Wiley, New York. ISBN: 0-471-85991-5

    MATH  Google Scholar 

  • Topal U (2009) Multiobjective optimization of laminated composite cylindrical shells for maximum frequency and buckling load. Mater Des 30:2584–2594

    Article  Google Scholar 

  • Topal U, Uzman Ü (2008) Maximization of buckling load of laminated composite plates with central circular holes using mfd method. Struct Multidisc Optim 35:131–139

    Article  Google Scholar 

  • Walker M (2001) Multiobjective design of laminated plates for maximum stability using the finite element method. Compos Struct 54:389–393

    Article  Google Scholar 

  • Walker M, Adali S, Verijenko V (1996) Optimization of symmetric laminates for maximum buckling load including the effects of bending-twisting coupling. Comput Struct 58(2):313–319

    Article  MATH  Google Scholar 

  • Wilson EL, Itoh T (1983) An eigensolution strategy for large systems. Comput Struct 16:259–265

    Article  MATH  Google Scholar 

  • Wittrick WH (1962) Rates of change of eigenvalues, with reference to buckling and vibration problems. J R Aeronaut Soc 66:590–591

    Google Scholar 

Download references

Acknowledgement

The authors gratefully acknowledge the support from the Danish Center for Scientific Computing (DCSC) for the hybrid Linux Cluster “Fyrkat” at Aalborg University, Denmark.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Esben Lindgaard.

Appendix A Design sensitivity analysis

Appendix A Design sensitivity analysis

1.1 A.1 Linear displacement sensitivity

The displacement sensitivities \(\frac{d \mathbf{D}}{d a_i}\) are computed by direct differentiation of the static equilibrium equation, see (1), w.r.t. a design variable a i , i = 1,..., I.

$$\label{eqn:displacement_sensitivity_pseudoload_elo} \mathbf{K_0} \frac{d {\bf D}}{d a_i}=-\frac{d \mathbf{K_0}}{d a_i}{\bf D} + \frac{d {\bf R}}{d a_i}, \quad i=1,\ldots,I $$
(15)

The displacement sensitivity \(\frac{d {\bf D}}{d a_i}\) can be evaluated by backsubstitution of the factored global initial stiffness matrix in (15). The initial stiffness matrix has already been factored when solving the static problem in (1) and can here be reused, whereby only the new terms on the right hand side of (15), called the pseudo load vector, need to be calculated. Note that the force vector derivative, \(\frac{{d \bf R}}{d a_i}\), is zero for design independent loads as in the case for CFAO. The global initial stiffness matrix derivatives \(\frac{d \mathbf{K_0}}{d a_i}\) are determined semi-analytically at the element level by central difference approximations and assembled to global matrix derivatives.

$$\frac{d {\bf k}_{\bf 0}}{d a_i} \approx \frac{\mathbf{k_{0}}(a_i+\Delta a_i) - \mathbf{k_{0}}(a_i-\Delta a_i)}{2 \Delta a_i} \label{eqn:k0_sensitivity_elementlevel_elo}\\ $$
(16)
$$\frac{d \mathbf{K_0}}{d a_i}=\sum_{n=1}^{N_e^{as}}\frac{d \mathbf{k_0}}{d a_i}, \quad i=1,\ldots,I \label{eqn:assembly_global_stiffness_matrix_elo} $$
(17)

k 0 is the element initial stiffness matrix, Δa i is the design perturbation, and \(N_e^{as}\) is the number of elements in the finite element model associated to the design variable a i .

1.2 A.2 Nonlinear displacement sensitivity

The nonlinear displacement sensitivities are computed by considering the residual or force unbalance equation at a converged load step n,

$${\bf Q}^n({\bf D}^n({\bf a}),{\bf a})={\bf F}^n-{\bf R}^n=\boldsymbol 0 $$
(18)

where Q n(D n(a),a) is the so-called residual or force unbalance, F n is the global internal force vector, and R n is the global applied load vector. Taking the total derivative of this equilibrium equation with respect to any of the design variables a i ,i = 1,...,I, we obtain

$$\frac{d {\bf Q}^n}{d a_i}=\frac{\partial {\bf Q}^n}{\partial a_i} +\frac{\partial {\bf Q}^n}{\partial {\bf D}^n}\frac{d {\bf D}^n}{d a_i}=\boldsymbol 0 \label{eqn:total_deriv_unbalance_desvar} \\ $$
(19)
$$\text{where \quad} \frac{\partial {\bf Q}^n}{\partial {\bf D}^n}=\frac{\partial {\bf F}^n}{\partial {\bf D}^n}-\frac{\partial {\bf R}^n}{\partial {\bf D}^n} \label{eqn:partial_unbalance_D_elo}\\ $$
(20)
$$\text{and \quad} \frac{\partial {\bf Q}^n}{\partial a_i}=\frac{\partial {\bf F}^n}{\partial a_i}-\frac{\partial {\bf R}^n}{\partial a_i} \label{eqn:partial_forceunbalance_desvar} $$
(21)

We note that (20) reduces to the tangent stiffness matrix. Since it is assumed that the current load is independent of deformation, \(\frac{{\partial\bf R}^n}{{\partial\bf D}^n}=\boldsymbol 0\), we obtain

$$ \frac{\partial {\bf F}^n}{\partial {\bf D}^n}= {\bf K}_{{\bf T}}^n $$
(22)

By inserting the tangent stiffness and (21) into (19), we obtain the displacement sensitivities \(\frac{{d \bf D}^n}{d a_i}\) as

$$\label{eqn:nonlinear_displacement_sensitivity} {\bf K}_{{\bf T}}^n\frac{d {\bf D}^n}{d a_i}=\frac{\partial {\bf R}^n}{\partial a_i}-\frac{\partial {\bf F}^n}{\partial a_i} $$
(23)

The partial derivative of the load vector, \(\frac{\partial {\bf R}^n}{\partial a_i}\), can explicitly be expressed by two terms by taking the partial derivative to (6)

$$ \frac{\partial {\bf R}^n}{\partial a_i}=\gamma^n \frac{\partial {\bf R}}{\partial a_i} + \frac{\partial \gamma^n}{\partial a_i} \mathbf{R} $$
(24)

For design independent loads \(\frac{\partial {\bf R}}{\partial a_i}=\boldsymbol 0\) and for a fixed load level \(\frac{\partial \gamma^n}{\partial a_i}=0\). The pseudo load vector, i.e. the right hand side to (23), is determined at the element level by central difference approximations and assembled to global vector derivatives.

1.3 A.3 Linear compliance

The design sensitivity of linear compliance is obtained by applying the adjoint approach, see e.g. Bendsøe and Sigmund (2003) and Lund and Stegmann (2005), and obtaining the sensitivity with respect to any design variable a i , i = 1,...,I as

$$\label{eqn:DSA_linear_compliance} \frac{d C_L}{d a_i} = - \mathbf{D}^\text{T} \, \frac{d \mathbf{K_0}}{d a_i} \, \mathbf{D} $$
(25)

The global initial stiffness matrix derivatives \(\frac{d \mathbf{K_0}}{d a_i}\) are determined semi-analytically at the element level by central difference approximations and assembled to global matrix derivatives as in (16) and (17).

1.4 A.4 Nonlinear end compliance

The design sensitivity of nonlinear end compliance at a converged load step n with respect to any design variable, a i , i = 1,...,I, is obtained by the adjoint approach, see e.g. Bendsøe and Sigmund (2003)

$$\label{eqn:DSA_nonlinear_compliance} \frac{d C_{\rm GNL}}{d a_i}= \boldsymbol{\lambda}^\text{T} \frac{\partial {\bf Q}^n}{\partial a_i} = \boldsymbol{\lambda}^\text{T} \left( \frac{\partial {\bf F}^n}{\partial a_i}-\frac{\partial {\bf R}^n}{\partial a_i} \right) $$
(26)

Assuming the end load fixed and independent of design changes we have that \(\frac{\partial {\bf R}^n}{\partial a_i}=\boldsymbol 0\). The adjoint vector \(\boldsymbol{\lambda}\), which is not to be confused with the eigenvector, is obtained as the solution to the adjoint equation

$$\label{eqn:adjointvector_nonlinear} {\bf K}_{{\bf T}}^n \, \boldsymbol{\lambda}=-{\bf R}^n $$
(27)

The partial derivatives in the right hand side of (26) are determined at the element level by central difference approximations and assembled to global vector derivatives.

1.5 A.5 Nonlinear first principal element strain

The design sensitivities of the first principal element strain, \(\frac{d \varepsilon_1}{d a_i}\), are determined semi-analytically by forward differences at the element level.

$$\label{eqn:DSA_Principal_Strain} \frac{d \varepsilon_1}{d a_i} \approx \frac{\varepsilon_1(\mathbf{D}^n + \Delta \mathbf{D}^n) - \varepsilon_1(\mathbf{D}^n)}{\Delta a_i} $$
(28)

The displacement field is perturbed via the calculated displacement sensitivities in (23) such that \(\Delta \mathbf{D}^n \approx \frac{d \mathbf{D}^n}{d a_i} \Delta a_i\).

1.6 A.6 Element nonlinearity factor

The design sensitivities of the element nonlinearity factor, \(\varepsilon_{\text{GNL}}\), are determined semi-analytically by forward differences at the element level.

$$\label{eqn:DSA_GNL_Factor} \frac{d \varepsilon_{\text{GNL}}}{d a_i} \approx \frac{\varepsilon_{\text{GNL}}(\mathbf{D}^1 + \Delta \mathbf{D}^1,\mathbf{D}^n + \Delta \mathbf{D}^n) - \varepsilon_{\text{GNL}}(\mathbf{D}^1,\mathbf{D}^n)}{\Delta a_i} $$
(29)

It is assumed that the initial load level and the final load level are fixed whereby the perturbed element nonlinearity factor is determined by

$$\begin{array}{lll} &\varepsilon_{\text{GNL}}(\mathbf{D}^1 + \Delta \mathbf{D}^1,\mathbf{D}^n + \Delta \mathbf{D}^n) \\ &{\kern4pt} = 1 + \left| \frac{\varepsilon_1^n(\mathbf{D}^n + \Delta \mathbf{D}^n)/\gamma^n-\varepsilon_1^1(\mathbf{D}^1 + \Delta \mathbf{D}^1)/\gamma^1}{\varepsilon_1^1(\mathbf{D}^1 + \Delta \mathbf{D}^1)/\gamma^1} \right| \end{array} $$
(30)

Since the element nonlinearity factor is determined by information at two equilibrium points, i.e. the initial load step and the final step n, the displacement sensitivities have to be calculated at both load steps by (23). The perturbation of the displacement fields at both equilibrium points may then be evaluated by \(\Delta \mathbf{D}^n \approx \frac{d \mathbf{D}^n}{d a_i} \Delta a_i\) and \(\Delta \mathbf{D}^1 \approx \frac{d \mathbf{D}^1}{d a_i} \Delta a_i\), respectively.

1.7 A.7 Linear buckling

The linear buckling load factor sensitivities may be determined by

$$\label{eqn:linear_buckling_DSA_elo} \frac{d \lambda_j}{d a_i}=\boldsymbol{\phi}_j^T \left( \frac{d \mathbf{K_0}}{d a_i} + \lambda_j \frac{d {\bf K}_{\boldsymbol \sigma}}{d a_i} \right) \boldsymbol{\phi}_j $$
(31)

where the eigenvalue problem in (2) has been differentiated with respect to any design variable, a i , i = 1,...,I, assuming that λ j is simple, see e.g. Courant and Hilbert (1953) and Wittrick (1962). The global matrix derivatives of K 0 and \({\bf K}_{\boldsymbol \sigma}\) are determined semi-analytically at the element level by central difference approximations and assembled to global matrix derivatives, see (16) and (17). The stress stiffness matrix is an implicit function of the displacement field, i.e. \({\bf K}_{\boldsymbol \sigma}(\mathbf{D}(\mathbf{a}),\mathbf{a})\), thus both displacement field and design variables need to be perturbed in the element central difference approximation. The displacement field is perturbed via the calculated displacement sensitivities in (23) such that \(\Delta \mathbf{D}^n \approx \frac{d \mathbf{D}^n}{d a_i} \Delta a_i\).

1.8 A.8 Nonlinear buckling

The nonlinear buckling load factor sensitivities at load step n are determined by

$$ \label{eqn:DSA_nonlin_eigprb_elo} \frac{d \lambda_j}{d a_i}=\boldsymbol{\phi}_j^T \left( \frac{d \mathbf{K_0}}{d a_i} + \frac{d {\bf K}_{{\bf L}}^n}{d a_i} + \lambda_j \frac{d {\bf K}_{\boldsymbol \sigma}^n}{d a_i} \right) \boldsymbol{\phi}_j $$
(32)

and

$$ \frac{d \gamma_j^c}{d a_i}=\frac{d \lambda_j}{d a_i} \, \gamma^n $$
(33)

where the eigenvalue problem in (11) has been differentiated with respect to any design variable, a i , i = 1,...,I, assuming that λ j is simple, see Lindgaard and Lund (2010a). It is assumed that the final load level is fixed and that the nonlinear buckling load has been determined at load step n by evaluation of (10) and (11). The global matrix derivatives of K 0 , \({\bf K}_{L}^n\), and \({\bf K}_{\boldsymbol \sigma}^n\) are determined in the same manner as for the linear buckling load sensitivities, i.e. semi-analytical central difference approximations at the element level and assembly to global matrix derivatives.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Lindgaard, E., Lund, E. Optimization formulations for the maximum nonlinear buckling load of composite structures. Struct Multidisc Optim 43, 631–646 (2011). https://doi.org/10.1007/s00158-010-0593-8

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00158-010-0593-8

Keywords

Navigation