You are on page 1of 10

A

SEMINAR REPORT

On

Approximate Probabilistic Optimization Using “Exact-Capacity


Approximate Response Distribution”

Submitted for the Partial Fulfilment of Requirements of the


Degree
of

MASTER OF ENGINEERING
in
MECHANICAL ENGINEERING
(DESIGN)

Guided By: Submitted by:


Prof. Kamlesh Purohit Vikram Patnecha
Professor & Supervisor ME-III Semester
Dr. Manish Bhandari 16MDE10506
Asst. Professor & Supervisor

Department of Mechanical Engineering


Faculty of Engineering & Architecture
(M.B.M. Engineering College)
J. N. V. University, Jodhpur (Raj.)
JAN-2018

i
CERTIFICATE
This is to certify that the Seminar Report entitled, “Approximate Probabilistic
Optimization Using “Exact-Capacity Approximate Response Distribution”, is being
submitted by Mr. Vikram Patnecha (Enroll. No.-12/20073, Roll No.-16MDE10506 ) for the
partial fulfilment of the requirements for the award degree of Master of Engineering in
Mechanical Engineering (Design) is a record of his own work carried out by him under my
supervision and guidance.

Countersigned By:

Prof. Kamlesh Purohit


Professor & Supervisor

Dr. Manish Bhandari


Asst. Professor & Supervisor

Dr. Rajat Bhagwat


Professor & Head of Department

Date:

Department of Mechanical Engineering


Faculty of Engineering & Architecture
(M.B.M. Engineering College)
J. N. V. University, Jodhpur (Raj.)
JAN-2018

ii
DECLARATION
This is to declare that the seminar report entitled, “Approximate Probabilistic
Optimization Using “Exact-Capacity Approximate Response Distribution”, is being
submitted for the partial fulfillment of the requirements of the degree of M.E. (Mechanical
Engineering) Design from Mechanical Engineering Department, M.B.M. Engineering
College, Jai Narayan Vyas University, Jodhpur, is a record of the dissertation work carried
out by me under the supervision of Dr. Kamlesh Purohit and Dr. Manish Bhandari. This
seminar report has not been submitted for the award of any degree or diploma by me or any
other student.

Vikram Patnecha
Enroll. No.12/20073
Roll No. 16MDE10506
M.E. III Semester

This is to certify that the declaration of the student is correct to the best of my knowledge.

Countersigned By:

Prof. Kamlesh Purohit


Professor & Supervisor

Dr. Manish Bhandari


Asst. Professor & Supervisor

Dr. Rajat Bhagwat


Professor & Head of Department
Date:

Department of Mechanical Engineering


Faculty of Engineering & Architecture
(M.B.M. Engineering College)
J. N. V. University, Jodhpur (Raj.)
JAN-2018

iii
ACKNOWLEDGEMENT
It is my pleasure and privilege to submit that with the blessings of the Almighty God, I’ve
been able to complete this dissertation/seminar work. I would like to acknowledge and
express my deep sense of gratitude towards the considerate people who have helped me in
this dissertation/seminar work.

I wish to specially thank Prof. Rajat Bhagawat, Head of Department of Mechanical


Engineering, for his precious support throughout the work.

I appreciatively acknowledge the keen supervision of my teacher and guide Prof. Kamlesh
Purohit and Dr. Manish Bhandari, Department of Mechanical Engineering, M.B.M.
Engineering College, Jodhpur, for inspiring and motivating me, despite her busy academic
schedule. She has always been kind enough to spare her valuable time and thought in giving
necessary guidance. Her rich and varied experience as an academician immensely helped
me in understanding this topic clearly.

I am grateful to Prof. Kamlesh Purohit, former Head of Department of Mechanical


Engineering who has always shown concern for my academic pursuits and for allowing me
to pursue M.E. under the blessings of such good teaching and non-teaching staff. Last and
most significantly, I would like to thank my family for the patience they have shown during
the course of my work and their constant blessings for my success.

Vikram Patnecha
M.E. (Design)
Date: Mechanical Engineering Department
Place: Jodhpur J.N.V. University, Jodhpur (Raj.)

iv
ABSTRACT

There are two major barriers in front of probabilistic structural design. First,
uncertainties associated with errors in structural and aerodynamic modeling and quality of
construction are not well characterized as statistical distributions and insufficient
information may lead to large errors in probability calculations. Second, probabilistic
design is computationally expensive because repeated stress calculations (typically finite
element analysis) are required for updating probability calculation as the structure is being
re-designed. Targeting these two barriers, we propose a probabilistic design optimization
method, where the probability of failure calculation is confined to failure stresses, to take
advantage of the fact that statistical characterization of failure stresses is required by Federal
Aviation Administration (FAA) regulations [25]. The stress distribution is
condensed into a representative single value thereby eliminating the need for expensive
stress distribution calculation, so a probabilistic optimization problem is transformed into a
semi-deterministic optimization problem. Since the procedure starts from the deterministic
optimum, a small number of iterations is expected, and a reliability analysis is required only
once in each iteration. The proposed method provides approximate sensitivity of failure
probability with respect to design variables, which is essential in risk allocation.

The method is demonstrated with (i) a beam problem with two failure modes, and (ii) ten-
bar truss problem. In the ten-bar truss problem, risk is allocated between the truss elements,
while risk is allocated between different failure modes in the beam example.

Probabilistic structural design deals with uncertainties in response (e.g. stresses) and
capacity (e.g. failure stresses). The calculation of the structural response is typically
expensive (e.g., finite element simulations), while the capacity is usually available from
tests. Furthermore, the random variables that influence response and capacity are often
disjoint. We have shown that this disjoint property can be used to reduce the cost of
obtaining the probability of failure via Monte Carlo simulations. We propose to use this
property for an approximate probabilistic optimization based on exact capacity and
approximate response distributions (ECARD). In Approximate Probabilistic Optimization
Using ECARD, the change in response distribution is approximated as the structure is re-
designed while the capacity distribution is kept exact, thus significantly reducing the
number of expensive response simulations. ECARD may be viewed as an extension of
SORA (Sequential Optimization and Reliability Assessment), which proceeds with

v
deterministic optimization iterations. In contrast, ECARD has probabilistic optimization
iterations, but in each iteration, the response distribution is approximated so as not to require
additional response calculations. The use of inexpensive probabilistic optimization allows
easy incorporation of system reliability constraints and optimal allocation of risk between
failure modes. The method is demonstrated using a beam problem and a ten-bar truss
problem. The former allocates risk between two different failure modes, while the latter
allocates risk between members. It is shown that ECARD provides most of the improvement
from risk re-allocation that can be obtained from full probabilistic optimization.

The use of probabilistic optimization in structural design applications is hindered by the


huge computational cost associated with evaluating probabilistic characteristics, where the
computationally expensive finite element method (FEM) is often used for simulating design
performance. A Sequential Optimization and Reliability Assessment (SORA) method with
analytical derivatives is applied to improve the efficiency of probabilistic structural
optimization. With the SORA method, a single loop strategy that decouples the
optimization and the reliability assessment is used to significantly reduce the computational
demand of probabilistic optimization. Analytical sensitivities of displacement and stress
functional derived from finite element formulations are incorporated into the probability
analysis without recurring excessive cost. The benefits of our proposed methods are
demonstrated through two truss design problems by comparing the results with using
conventional approaches. Results show that the SORA method with analytical derivatives
is the most efficient with satisfactory accuracy [14].

vi
NOMENCLATURE

β = reliability index

c.o.v. = coefficient of variation

𝑐𝑓, 𝑐𝜎, = coefficients of variation of the failure stress and the stress,
respectively.

∆* = relative change in characteristic stress σ* corresponding to a relative


change of ∆ in stress σ

∆ = relative change in stress

F( ) = cumulative distribution function of the failure stress

f( ) = probability density function of the failure stress

k = correction factor for characteristic stress

µ𝑓 , µ𝜎 = mean values of the failure stress and the stress, respectively.

s( ) = probability density function of the stress

𝜎𝑓 = failure stress

𝜎∗ = characteristic stress

*𝜎𝑝 = characteristic stress for previous design

σ = stress

𝑃𝑓 = approximate probability of failure of probabilistic design

P𝑓𝑝 = probability of failure at previous design

P𝑓 = actual probability of failure

𝑃𝑓𝑑 = probability of failure at given deterministic design

𝑊𝑑 = weight of deterministic design

W = weight of probabilistic design

vii
CONTENTS

1. Abstract v
2. Nomenclature vii
3. List of Figures ix
4. List of Tables x
5. CHAPTER 1: INTRODUCTION 1-3
6. CHAPTER 2: LITERATURE REVIEW 4-22
2.1. Optimization 4
2.1.1. Design Deterministic optimization 4
2.1.2. Probabilistic optimization 6
2.1.3. Reliability based design optimization 7
2.1.4. Double loop method 10
2.1.5. SORA Method 10
2.1.6. Probability density failure 14
2.1.7. ECARD 17
7. CHAPTER 3: Exact capacity approximate response distribution
probabilistic structural design 23-36
3.1. Application of ECARD using Beam Problem 24
3.1.1. Deterministic optimization 24
3.1.2. Probabilistic optimization 25
3.1.3. Probabilistic optimization using ECARD 27
3.2. Application of ECARD using ten bar truss Problem 29
3.2.1. Deterministic optimization 29
3.2.2. Probabilistic optimization 33
3.2.3. Probabilistic optimization using ECARD 34
8. CONCLUSION 37
9. REFERENCES 38-41

viii
LIST OF FIGURES

S No. Figure Page

1 Figure 2.1 Double-loop RBDO process 8

2 Figure 2.2 Flowchart of double loop strategy 10

3 Figure 2.3 The flowchart for SORA 11

4 Figure 2.4 R. Percentile performance of A constraint function 12

5 Figure 2.5 Flow chart for SORA method 13

6 Figure 2.6 shifting distance of probabilistic boundaries 14

7 Figure 2.7 PDF of Failure stress or capacity curve 15

8 Figure 2.8 Log-normal probability density functions 15

9 Figure 2.9 Normal Probability density function 16

Figure 2.10: Calculation of characteristic stress σ*From the given


10 18
probability of failure and CDF of the failure stress

Figure 2.11 Stress distribution s(σ) before and after redesign in


11 19
relation to failure-stress distribution f (σ f)

Figure2.12
(a) Comparison of approximate and exact ∆ and ∆* and
12 (b) the resulting probabilities of failure for lognormal failure stress 20
(with µf=100 and cf=8%) and normal stress (with µσ=42.49 and
cσ=20% )

13 Figure 3.1 Cantilever beam: geometry and loading 23

Figure 3.2 Deterministic, probabilistic and approximate optimum


14 26
design using FORM

15 Figure 3.3 Ten-bar truss example 29

ix
LIST OF TABLES

S No. Table Page

Table-3.1: The mean and coefficient of variation of the random


1 23
variables. All variables follow normal distribution.

2 Table-3.2: Deterministic optimum of the beam problem 25

Table-3.3: Probabilistic optimum of the beam problem using


3 26
FORM

Table 3.4. Probabilistic optimum of the beam problem using


4 26
MCS

Table 3.5 lists the designs attained in the iterations of the


5 27
approximate optimization using FORM.

Table 3.6. Iterations of approximate ECARD probabilistic


6 28
optimization for the cantilever beam problem using. MCS.

7 Table 3.7: Input data for truss problem 30

8 Table 3.8: Deterministic optimum of ten-bar truss problem 30

9 Table 3.9: Error and variability in ten-bar truss problem 32

Table 3.10: Probabilistic optimum of ten-bar truss problem using


10 33
Separable MCS

Table 3.11: Results of approximate probabilistic optimization


11 34
and progress towards the true optimum.

You might also like