Heavy

Dominated Unit 8.
Metric 2: Mass m

solution Objectives in conflict:

trade off methods and
Non-dominated
penalty functions
solution

Trade-off
Mike Ashby
surface
Light

Department of Engineering
Cheap Metric 1: Cost C Expensive University of Cambridge

© Granta Design and M. F. Ashby, 2017

For reproduction guidance see back page

This lecture unit is part of a set created by Mike Ashby to help introduce students to materials, processes and rational selection.

The Teaching Resources website aims to support teaching of materials-related courses in Design, Engineering and Science.
Resources come in various formats and are aimed primarily at undergraduate education.
Some of the resources are open access and students can access them. Others are only available to educators using CES EduPack. www.teachingresources.grantadesign.com
 Learning objectives for this Lecture Unit

Intended Learning Outcomes

Knowledge and
Understanding Knowledge on graphical trade-off methods and penalty functions
Skills and
Ability to select systematically when design objectives conflict
Abilities
Values and
Attitudes Appreciation of the value of compromise in engineering design

Resources
 Text: “Materials Selection in Mechanical Design”, 5th Edition by M.F.
Ashby, Butterworth Heinemann, Oxford, 2016. Chapters 8-9
 Text: “Materials and the Environment”, 2nd Edition by M.F. Ashby,
Butterworth-Heinemann, Oxford 2012, UK. Chapters 9-10

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

 Heavy
Outline of Lecture Unit 8

Dominated
 Almost always 2+ objectives – they conflict
Metric 2: Mass m

solution
 Trade-off methods
 Penalty functions and exchange constants
Non-dominated
solution  Two-objective minimisation in EduPack

Trade-off
surface
Light

Cheap Metric 1: Cost C Expensive

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

 The selection strategy: materials

Design requirements: Data:

expressed as Material attributes
Constraints and Process attributes
Objectives Documentation

Able to be molded
Comparison engine
Density
Water and UV resistant  Screening Price
Stiff enough  Ranking Modulus
Strong enough Strength
 Documentation
As cheap as possible Durability
As light as possible Process compatibility
More…….
Final selection

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

 Multiple constraints and objectives

Design requirements set constraints – criteria for screening

objectives – criteria for optimising

Typical constraints Typical objectives

The material must be Minimize
 Electrically conducting  Mass m (satellite components)
 Optically transparent.....
 Volume (mobile phones)
And meet target values of
 Energy consumption (fridges)
 Stiffness
 Carbon footprint (cars)
 Strength…..
 Embodied energy (materials)
And be able to be
  Cost C (everything)
Die cast
 Welded ......

Dealing with multiple constraints is Dealing with multiple objectives

straightforward needs trade-off methods

Take, as example, simultaneously minimizing mass m and cost C

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com
 Multi-objective optimization: the words

Heavy
 “Solution”: a candidate that
meets the constraints, but not
necessarily optimum by either Dominated
solution
objective
 Plot solutions.

Mass m
(Convention: express objectives
to be minimized) Non-dominated
solution
 “Dominated solution”: one
that is definitely non-optimal Trade-off
surface
 “Non-dominated solution”:
Light

one that is optimal by one metric Cheap Cost C Expensive

(but not usually by both)

 “Trade-off surface”: the surface on which the non-

dominated solutions lie (Pareto Front). In our case a 2-
dimensional curve
Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com
 Finding a compromise: strategy 1

Heavy
 Make a trade-off plot
 Sketch a trade-off curve
 Use intuition to select a

Mass m
solution on the trade-off curve

 “Solutions” nearest the surface offer

Trade-off
the best compromise between curve
mass and cost Light

 8 out of 50 Cheap Cost C Expensive

 Choose from among these - depends on how highly you value light weight

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

 Finding a compromise: strategy 2

Heavy
 Reformulate all but one of the
objectives as constraints, Upper limit
setting an an upper limit for it on C

OK if budget limit

Mass m
 BUT….cheating Trade-off
curve
Cost is treated as constraint, not
objective. Best
choice
Light

Cheap Cost C
Expensive

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

 Finding a compromise: strategy 3

Heavy
Define locally-linear Z2 Z3 Z4
Z1
Penalty function Z Contours of
constant Z
Z  C  m
Seek solution with smallest Z

Mass m
 Make trade-off plot Trade-off
curve

Plot on it contours of Z Optimum

solution,
1 1 minimising Z
m  C Z
  Decreasing
values of Z
Lines of Z have slope -1/ 1/ 
(needs linear scales) Light
Cheap Cost C Expensive
 Read off solution with lowest Z

 Two issues: (Q1) What is the so called exchange constant,  ?

(Q2) What if we have Log, not Linear scales?

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com
 (Q1) Example of graphical solution for teaching

5000
Bicycles:
Trade-off price vs . mass
surface
4000
Titanium
CFRP
Alu-alloy

 determines a
Price (US $)

Alloy steel
3000 Plain steel
Slope
Slope location on the
$$
2000
200 // kg
kg
trade-off curve and
2000
reflects priorities
(price per kilo)
1000

Slope
$ 20 / kg
0
6 8 10 12 14 16
Bicycle mass (kg)

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

 (Q1) Example: materials for transport systems

Choice of material depends on system

 Mass, in transport systems, means fuel

$ kg
 Life cost = Initial cost, C + Fuel cost over life, scaling with mass m
$/kg
 Penalty function Z  C αm

 Must establish exchange constant, α

Steel Steel / Alu Alu / (composite) Alu / Ti / composites Composites

 3–6 6 – 20 100 – 600 600 – 2,000 (?) 5,000 – 10,000

($/
kg)

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

 (Q2) Linear penalty functions go with linear axes

Heavy

Z3 Z4 Z5 Z5

Z4
Z2

Z3

Log (Mass, m)
Z
1
Z2
Mass, m

Trade-off Z
1
surface

Decreasing
Decreasing values of Z
values of Z
Best Best
choice choice Trade-off
-1/a

Light
surface
Light

Cheap Cost, C Expensive Cheap Log (Cost, C) Expensive

 Set your axes to linear before plotting property charts for linear penalty functions
 Logarithmic scales give the same best choice but Z no longer appears as straight

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

 (Q2) Example of two-objective Log chart in EduPack

Minimum mass and cost for member in tensile or compressive load

and stiffness-limited design: Log scale axes

Minimize
M2 =  / E

Minimize M1 = Cm *  / E
Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com
 (Q2) How to use a penalty function in bubble charts

1e7 Trade-off in typical Log scale chart

1e6 (1) Make trade-off plot

M2=Density / Young's modulus

Trade-off curve
100000
(2) Convert to linear scales
10000
(double-click on axes)

1000

100

10

0.1 1 10 100 1000 10000 100000 1e6 1e7 1e8 1e9

M1=Price * Density / Young's modulus

(3) Convert back to log scales

Selection line now curved
Move down until few materials left

(3) Apply selection line, slope -1/.

Select “minimize index”

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

 Two ways to find materials for auto bumpers

Beam

Function Absorb impact, transmit load to energy-absorbing units or supports

Objectives Minimize mass and material cost

Criteria Mass m per unit Cost C per unit

bending strength bending strength
 Cm  Cm = Material cost / kg
Beam in bending m C
Index to minimize:  y2/3  y2/3 ρ = Density, kg/m3
 y = Yield strength, MPa
 = exchange constant , $/kg

Penalty function Z  C m 


 y2/3
C m  
Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com
 Bar chart selection using the penalty function

 
Z   C m   
 2/3 
y


Use the The value of the

“Advanced”
(Density / (Yield strength^0.66)) exchange constant
facility to make
the penalty *(Price + 10)
function

+ - / * ^ ( )

List of properties
 Density
 Price
 Tensile strength
 etc

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

Bar chart selection using the penalty function

α = $1/kg

Best choice: steels

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

Bar chart selection using the penalty function

α = $10/kg

Best choice: Mg alloys

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

Bar chart selection using the penalty function

α = $100/kg

Best choice: CFRP

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

 Bubble chart selection using penalty function
Strong bumper, minimum
weight and cost


Minimize weight M2 
 y2 / 3

Cm  10
 100
1
Minimize cost M1 
 y2 / 3

Penalty function Z  M1   M2

 = 1 $/kg Low alloy steels, Carbon steels,

 = 10 $/kg Aluminum alloys, Magnesium alloys

 = 100 $/kg Carbon-fiber reinforced composites

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com
 So what?

 Real design involves conflicting objectives –

often technical performance vs. economic performance (cost).

 Trade-off plots reveal options

 If the exchange constant is known –

penalty function allows unambiguous choice

 The penalty function technique can be

applied to bar charts or bubble charts in CES
EduPack for interactive and visual selection

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

 Lecture Unit series

These PowerPoint lecture units, as well as many other types of resources, are on the Teaching Resource Website.

Finding and Displaying Information Sustainability

Unit 1 The materials of engineering What is a sustainable development? a materials perspective
Unit 2 Material property charts: mapping materials Materials for low carbon power
Unit 3 The Elements database: properties, relationships and resources Special Topics
Material Properties Architecture and the built environment: materials for construction
Unit 4 Manipulating properties: composition, microstructure, architecture Structural sections: shape in action
Unit 5 Designing new materials: filling the materials-property space Materials in industrial design: Why do consumers buy products?
Unit 6 Materials Science and Engineering The Products, Materials and Processes database
Selection Materials for Bioengineering
Unit 7 Material selection: translation, screening, ranking, documentation Advanced Teaching and Research
Unit 8 Objectives in conflict: trade off methods and penalty functions Advanced databases: a lightning tour
Unit 9 Material and shape: materials for efficient structures The Aerospace edition
Unit 10 Manufacturing processes and cost modeling The Polymer edition
Unit 11 Eco-informed material selection The Synthesizer tool: hybrids and other models
Unit 12 Eco design and the Eco Audit tool Editing and creating new databases: CES Constructor
CES Selector and Constructor in research

Granta Design and Mike Ashby, 2017 www.teachingresources.grantadesign.com

 Author
Mike Ashby
University of Cambridge, Granta Design Ltd.
www.grantadesign.com
www.eng.cam.ac.uk

Reproduction
There are 300+ resources
These resources can be reproduced for teaching purposes provided
you have purchased access to Granta’s Teaching Resources (usually as (Available to Educators)
part of a CES EduPack license). Please make sure that Granta Design is
credited on any reproductions. You cannot use these resources for
commercial purposes. Including:
 PowerPoint lecture units
 Exercises with worked solutions
Accuracy
 Recorded webinars
We try hard to make sure these resources are of a high  Posters
quality. If you have any suggestions for improvements,
please contact us by email at  White Papers
 Solution Manuals
teachingresources@grantadesign.com  Interactive Case Studies

© Granta Design and M. F. Ashby, 2017

This lecture unit is part of a set created by Mike Ashby to help introduce students to materials, processes and rational selection.

The Teaching Resources website aims to support teaching of materials-related courses in Design, Engineering and Science.
Resources come in various formats and are aimed primarily at undergraduate education.
Some of the resources are open access and students can access them. Others are only available to educators using CES EduPack. www.teachingresources.grantadesign.com