Genetic Algorithms Based Optimisation of Composite Ship Structures

1
Genetic Algorithms Based Optimisation of
Composite Ship Structures

• Komsan Maneepan
• Supervisor Prof. R.A. Shenoi
• Co-supervisor Dr. J.I.R Blake
• Sponsor Royal Thai Navy
• Acknowledgements Dr. H.K. Jeong, EU FP6
  Marstruct Program

2
Content

• Introduction
• Literature reviews
• Optimisation methodology
• Structural analysis
• Genetic algorithm
• Applications and problem definition
• Unidirectional stiffened plate
• Cross stiffened plate
• Unstiffened plate
• Conclusion
• Further works

3
Introduction

• FRP composite ? fibre, resin, fibre volume
  fraction ? composite form, woven roving, CSM,
  unidirectional laminate ? production process
• Marine applications ?small craft, Naval ship
  ?Top-hat stiffened plates, sandwich plate

4
Compare composite with metal
5
(No Transcript)
6
Physical significance of stiffness terms in Force
and Moment Resultants
7
Production cost
Production Costlabour rate x (working time of
based plate and stiffeners)
8
Reviews on optimisation methods

• Many papers using the classical optimisation
  method, show the optimum fibre angle, the main
  design variable for composite structure, is not
  suitable for manufacturing processes. Moreover,
  the global optimum could not be obtained if the
  starting point is near a local optimum.
• The GA is the most popular in the stochastic
  types of optimisation methods as a results of
  their simplicity.

9
Reviews on the structural analysis

• Stiffened plate can be solved by Grillage
  assumption, although force method provide the
  exact solution, the computational time increase
  when the number of intersection is high. The
  energy method required only one equation to be
  solved.
• Other method is orthotropic plate method. It is
  justified only when stiffeners are closely
  spaced.
• For unstiffened plate, 3-D, CLPT, FSDT, HSDT are
  available. The 3-D is exact solution. CLPT is
  unique, but the less accuracy comparing to
  others. HSDT is the best in the equivalent single
  layer theories.
• FEM is the universal method for the solutions of
  mechanics problems and there are plenty of
  commercial FEM software packages available.
  

10
Structural Analysis
11
Stress component and orthotropic lamina
12
Optimisation methodology
13
Structural analysis

• For stiffened plate, the energy method based on
  Naviers solution is adapted by introducing the
  flexural rigidity of composite material cross
  section into energy terms.
• For unstiffened plate, HSDT is selected because
  the displacement field is close to the
  deformation of plate in real situation.
  

14
Assumption of stiffened plate
15
Grillage Analysis
Equivalent elastic properties of stiffener
component
Flexural rigidity of stiffener is obtained and
then the total strain energy for all stiffeners
is
Total strain energy Work done
Stresses on crown or base plate and shear stress
on web
16
Equivalent elastic properties
17
Output

• Deflection shape throughout the plate
• Stresses results for both longitudinal and
  transverse stiffeners
• Stress variation through stiffeners cross
  section
• Shear stress on Web element

P
18
Genetic Algorithm (GA)

• A stochastic search method mimic the laws of
  natural evolution
• Initial population it is randomly generated.
• Evaluation the fitness is directly related to
  the value of objective function of the phenotype
  of chromosome
• Selection operator it is for exploiting the
  design space by a randomised procedure to create
  a new population. For example, Roulette wheel
  selection select chromosomes from pool by
  determining their survival probability.
• Crossover it creates offspring by combining
  parents features
• Mutation it cause GA search to increase
  population diversity by introducing new genetic
  material.

19
Flow chart of GA and design variables
representation
20
Validation of structural analysis
21
Validation of GA program
22
Comparison of the proposed methodology with Ansys
23
Application to composite ship structures
Sp/4

• Cross stiffened plates
• Unidirectional stiffened plates
• Unstiffened plate

Sp/4
Problem definitions
Girder spacing (Sp)

• Plate dimensions are specified for each problem
  (maximise stiffness, strength or weight)
• Design variables are, fibre angle, number of
  plies, fibre type, Areal weight

24
Design variables
25
Considering positions
26
Material properties of resin and fibre
27
Unidirectional stiffened plates
Def.max lt 10.0 FI.max lt1.0

• Maximise stiffness
• UHM(8)? high modulus material is selected
• Girder spacing is upper-bound value
• Maximise strength
• E-glass(8)?low modulus material with high areal
  weight is selected.
• Weight minimisation
• UHM(8) ?reduce the maximum deflection
• HS(2)?reduce the magnitude of failure index

28
(No Transcript)
29
Weight minimisation of unidirectional stiffened
plate
30
Cross-stiffened plates
Def.max lt 25.40 FI.max lt1.0

• Maximise stiffness
• High modulus fibre (UHM) is building material
• Beam spacing is wider than girder spacing
• Maximise strength
• E-glass is used to increase the strength at the
  web
• Weight minimisation
• Crown element is highest thickness
• The weight of plate mainly rely on stiffness
  constraints

45/-45/45s, UHM/HS/HSs, 0.2/0.3/0.5s
90s, HMs, 0.2s
90s, HMs 0.2s
0s, UHMs, 0.2s
609.60 mm
158.75 mm
90s, UHMs, 0.2s
31
(No Transcript)
32
Weight minimisation of cross stiffened plate
33
Un-stiffened plate

• Maximise stiffness
• Maximise strength
• Maximise critical buckling load

Each design variable has a large influence on
optimal results Thickness of the plate mainly
influence on the optimal results. (E-glass with
its highest areal weight is used)
34
(No Transcript)
35
Maximisation of buckling load for unstiffened
laminated plate
36
Conclusion

• The unique optimisation framework is introduced
  for composite ship framework by adapting grillage
  analysis.
• The validation of the subroutine program is
  shown.
• Applications of the optimisation framework to
  three plate types is presented.

37
Further works

• The extension this work for midship section.
• Multi-objective could be added.
• Plates with other supported conditions and
  various load types.
• Material data base will be replace the evaluation
  formulation of material properties
• Parallel computing should be used to help GA for
  fast convergence to optimum results.

38
List of publications

• Maneepan K., Jeong H.K. and Shnoi R.A. (2005)
  Optimisation of FRP tophat stiffened single skin
  and monocoque sandwich plates using genetic
  algorithm, ISOPE, June 19-24, Seoul, Korea, pp.
  513-518
• Maneepan K., Shenoi R.A., Jeong H.K. and Blake
  J.I.R. (2006) Multi-objective optimisation of
  orthogonally tophat-stiffened composite laminates
  plates, OMAE, June 4-9,Hamburge, Germany
• Maneepan K., Shenoi R.A., Blake J.I.R. and Jeong
  H.K. (2006) Genetic Algorithms (GAs) based
  optimisation of FRP composite plated grillages in
  ship structures, International Journal of
  Maritime Engineering (RINA).

39
(No Transcript)