1 INTRODUCTION
The main objective of design optimization is to ob-
tain a set of design variables that minim-
ize/maximize objective function(s) of interest while
satisfying given constraints. If design optimization is
performed in a deterministic manner, i.e. uncertain-
ties are not taken into account during the optimiza-
tion, the resultant optimal design may have unquan-
tified risk of violating the given constraints.
Recently, various reliability based design optimiza-
tion (RBDO) methods have been developed to
achieve optimal designs with acceptable failure
probabilities (see Frangopol & Maute 2004 for a
state-of-the-art review of RBDO and recent applica-
tions to civil and aerospace structural systems). Dur-
ing RBDO, the probability of violating given con-
straint(s), i.e. the failure probability, is often
computed by reliability analysis employing methods
such as first order reliability method (FORM),
second order reliability method (SORM) or response
surface method.
Traditionally, RBDO has been performed by use
of a nested or “double loop” approach, that is, each
step of the iteration for design optimization involves
another loop of iteration for reliability analysis. For
example, reliability index approach (RIA; Enevold-
sen & Sorensen 1994) and performance measure ap-
proach (PMA; Tu et al. 1999) employ FORM as the
inner loop to perform the reliability analysis effi-
ciently. If the constraints are active, the two ap-
proaches yield the same results. However, it is
known that PMA is generally more efficient and sta-
ble than RIA (Tu et al. 1999).
In general, the double loop approach is computa-
tionally expensive. Recently, a single-loop approach
(Liang et al. 2004) was proposed to improve effi-
ciency of RBDO. The Karush-Kuhn-Tucker (KKT)
optimality condition is used to approximate the de-
sign point (or most probable point, MPP) in the in-
ner loop for each constraint. As a result, the inner
loop is replaced by a deterministic constraint, which
transforms the double loop RBDO problem into an
equivalent deterministic optimization problem.
When multiple failure modes need to be consi-
dered as the constraints of a design optimization,
RBDO is often formulated such that the optimal
structure satisfies each failure mode with pre-
determined probabilities. This approach is termed as
component reliability-based design optimization
(CRBDO) in this paper. In some cases, however, the
failure event is better described by a system event,
i.e. a logical (or Boolean) function of multiple fail-
ure modes. In this case, the probabilistic constraint
should be given for the system event. This approach
is called system reliability-based design optimiza-
tion (SRBDO). The SRBDO requires system relia-
bility analysis, which is not trivial especially for sta-
tistically dependent component events, or for a
system event that is not series or parallel system.
Single-Loop System Reliability Based Design Optimization (SRBDO)
Using Matrix-based System Reliability (MSR) Method
T.H. Nguyen, J. Song, and G.H. Paulino
University of Illinois, Urbana, Illinois, USA
ABSTRACT: This paper proposes a single-loop system reliability based design optimization (SRBDO) ap-
proach using the recently developed matrix-based system reliability (MSR) method. A single-loop method
was employed to eliminate the inner loop of SRBDO that evaluates probabilistic constraints. The MSR me-
thod computes the system failure probability and its parameter sensitivities efficiently and accurately through
efficient matrix calculations. The SRBDO/MSR approach proposed in this paper is uniformly applicable to
general systems including series, parallel, cut-set and link-set system events. Two numerical examples dem-
onstrate the proposed approach. In the first example, the cross-sectional areas of the members of a statistically
indeterminate truss structure are determined for minimum total weight with a constraint on the system failure
probability satisfied. The second example demonstrates the application of the proposed approach to topology
optimization. The influences of the statistical correlation and the types of constraints, i.e. deterministic, prob-
abilistic (component) and probabilistic (system) on the optimal topology are investigated.
Safety, Reliability and Risk of Structures, Infrastructures and
Engineering Systems – Furuta, Frangopol & Shinozuka (eds)
© 2010Taylor & Francis Group, London, ISBN 978-0-415-47557-0
1534