Reliability Based Design Optimization

Reliability Based Design
Optimization
Outline
 RBDO problem definition

 Reliability Calculation
 Transformation from X-space to u-space
 RBDO Formulations
 Methods for solving inner loop (RIA , PMA)
 Methods of MPP estimation
Terminologies
 X : vector of uncertain variables

 η : vector of certain variables
 Θ : vector of distribution parameters of
uncertain variable X( means , s.d.)
 d : consists of θ and η whose values can be
changed
 p : consists of θ and η whose values can not
be changed
Terminologies(contd..)
 Soft constraint: depends upon η only.

 Hard constraint: depends upon both X(θ)
and η
 [θ,η] = [d,p]
 Reliability = 1 – probability of failure
RBDO problem
Optimization problem
 min F (X,η) objective
 fi (η) > 0
 gj (X, η ) > 0
RBDO formulation
 min F (d,p) objective
 fi (d,p) > 0 soft constraints
 P (gj (d,p ) > 0) > Pt hard constraints
Comparison b/w RBDO and Deterministic
Optimization
Feasible Region
Reliability Based Optimum
Deterministic Optimum
Basic reliability problem
Reliabilty Calculation
Probability of failure
Reliability Index
Reliability index
Formulation of structural reliability problem
Vector of basic random variables
represents basic uncertain quantities that define the state of the

structure, e.g., loads, material property constants, member sizes.
x2 Limit state function
Safe domain
f X ( x ) = const. f
Failure domain
g(x) = 0
s
Limit state surface
0 x1
Geometrical interpretation
uS
failure domain Transformation to the

 f standard normal space
limit state surface
safe domain
 S
0 uR
Distance from the origin [uR, uS] to the Cornell reliability index
linear limit state surface
Hasofer-Lind reliability index
• Lack of invariance, characteristic for the Cornell reliability
index, can be resolved by expanding the Taylor series
around a point on the limit state surface. Since alternative
formulation of the limit state function correspond to the same
surface, the linearization remains invariant of the formulation.
• The point chosen for the linearization is one which has the
minimum distance from the origin in the space of transformed
standard random variables . The point is known as the
design point or most probable point since it has the highest
likelihood among all points in the failure domain.
Geometrical interpretation
For the linear limit state function, the absolute value of the
reliability index, defined as , is equal to the distance
from the origin of the space (standard normal space) to
the limit state surface.
Hasofer-Lind reliability index
u2
failure domain
f
safe domain u*
S 
u1
0 G3 ( u ) = 0
G2 ( u ) = 0
G1( u ) = 0
RBDO formulations
RBDO Methods
Double Loop Decoupled Single Loop

Double loop Method
Objective
function
Reliability Reliability
Evaluation Evaluation
For 1st For mth
constraint constraint
Decoupled method(SORA)
Deterministic optimization loop
Objective function : min F(d,µx)
Subject to : f(d,µx) < 0
g(d,p,µx-si,) < 0
k = k+1
si = µxk – xkmpp dk,µxk
xkmpp ,pmpp
Inverse reliability analysis for

Each limit state
Single Loop Method
 Lower level loop does not exist.
 min { F(µx) }
 fi (µx) ≤ 0 deterministic constraints
gi (x) ≥ 0
 where
 x- µx = -βt*α*σ
 α=grad(gu(d,x))/||grad(gu(d,x))||
 µxl ≤ µx ≤ µxu
Inner Level Optimization
(Checking Reliability Constraints)
 Reliability Index  Performance Measure

Approach(RIA) Approach(PMA)
 min ||u||  min gi ( u,µx )

 subject to gi(u,µx)=0  subject to ||u|| = βt
 if min ||u|| >βt(feasible)  If g(u*, µx )>0(feasible)

Most Probable Point(MPP)
 The probability of failure is maximum corresponding
to the mpp.
 For the PMA approach , -grad(g) at mpp is parallel
to the vector from the origin to that point.
 MPP lies on the β-circle for PMA approach and on
the curve boundary in RIA approach.
 Exact MPP calculation is an optimization problem.
 MPP esimation methods have been developed.
MPP estimation
active
constraint
RIA
MPP
 inactive
constraint
PMA MPP
PMA MPP
RIA MPP
U Space
Methods for reliability computation
Numerical computation of the integral in definition for
large number of random variables (n > 5) is extremely difficult
or even impossible. In practice, for the probability of failure
assessment the following methods are employed:
 First Order Reliability Method (FORM)

 Second Order Reliability Method (SORM)
 Simulation methods: Monte Carlo, Importance Sampling
FORM – First Order Reliability Method
u2
f
n ( u,0,I ) = const

u* region of most
contribution to
probability integral
*
G (u) = 0
0 u1
l(u) = 0
s
SORM – Second Order Reliability Method
u2 v n  v n
f
~
v
v*
Gv(v) = f v ( ~
v ) – vn = 0
v n = sv ( ~
v)
0 u1
s
Gradient Based Method for finding MPP
 find α = -grad(uk)/||grad(uk)||
 uk+1=βt* α
 If |uk+1-uk|<ε, stop
 uk+1 is the mpp point
 else goto start
 If g(uk+1)>g(uk), then perform an arc search
which is a uni-directional optimization
Abdo-Rackwitz-Fiessler algorithm
find
subject to
Rackwitz-Fiessler iteration formula
Gradient vector in the standard space:

Abdo-Rackwitz-Fiessler algorithm
Convergence criterion
for every and
Very often to improve the effectiveness of the RF algorithm

the line search procedure is employed
where is a constant < 1, is the other indication

of the point in the RF formula.
Merit function proposed by Abdo

Alternate Problem Model
 solution to :
 min ‘f’ s.t
 atleast 1 of the reliability constraint is exactly
tangent to the beta circle and all others are satisfied.
 Assumptions:
 minimum of f occurs at the aforesaid point
Alternate Problem Model
x1 β1
Reliability based
β2
optimum
x2
Scope for Future Research
 Developing computationally inexpensive

models to solve RBDO problem
 The methods developed thus far are not
sufficiently accurate
 Including robustness along with reliability
 Developing exact methods to calculate
probability of failure

Reliability Based Design Optimization

Uploaded by

Copyright:

Available Formats

Reliability Based Design Optimization

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Reliability Based Design Optimization

Uploaded by

Copyright:

Available Formats

Reliability Based Design

 RBDO problem definition

 X : vector of uncertain variables

 Soft constraint: depends upon η only.

Reliability Based Optimum

represents basic uncertain quantities that define the state of the

failure domain Transformation to the

Double Loop Decoupled Single Loop

Inverse reliability analysis for

 Reliability Index  Performance Measure

 min ||u||  min gi ( u,µx )

 if min ||u|| >βt(feasible)  If g(u*, µx )>0(feasible)

 First Order Reliability Method (FORM)

Rackwitz-Fiessler iteration formula

Gradient vector in the standard space:

for every and

Very often to improve the effectiveness of the RF algorithm

where is a constant < 1, is the other indication

Merit function proposed by Abdo

 Developing computationally inexpensive

You might also like