Materials & Design. Vol. 17, No. 1. pp.

11-17, 1996
Copyright 8 1996 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
0261-3069/96 515.00 + 0.00

PII: SO261-3069(96)00024-6

Selection of materials to reduce environmental

impact: a case study on refrigerator insulation

P. M. Weaver, M. F. Ashby, S. Burgess and N. Shibaike

Engineering Design Centre, Cambridge University Engineering Department,
Trumpington Street, Cambridge CB2 IPZ, UK

Received 14 March 1996; accepted 24 June 1996

The creation and use of any engineering product carries with it environmental penalties. There is
a growing recognition that the minimization of these penalties must become a primary design
objective. Some of the considerations - and difficulties - in selection of materials to minimize
environmental impacts are discussed. A design strategy is developed which relates traditional
design objectives with environmental impact indicators. The selection of insulation for refrigera-
tors is used to illustrate the approach. Copyright 0 1996 Elsevier Science Ltd.

Keywords: environmental impact; materials selection; multi-objective design

Introduction: material use and the environment tion of environmental impact as an objective. It is un-
Powerful forces drive the development of new and realistic to think of it as the only objective, since mini-
improved materials, encourage substitution, and modify mizing cost is always a consideration also. But if a value
the way in which materials are produced and used. can be placed on each sort of impact, allowing the
Market forces - the demands for stiffer, stronger, formulation of a set of ‘exchange-rates’, converting
lighter, cheaper materials - remain one of the strongest. impact to dollars, the application of optimization
The ingenuity of research scientists, too, drives methods is possible.
change by revealing a remarkable spectrum of new To reach this goal, a need must be met: that for quan-
materials with exciting possibilities, though the time it titative accounting methods for environmental impact.
takes to develop and commercialize them is long - They do not need to be exact: useful conclusions can
typically 10 years or more - requiring sustained
be drawn from approximate data. The measures may
investment.
Until recently, these were the principal evolutionary require debate and will change with time, but tech-
forces of materials technology. But our damaging niques for obtaining them are only a little more difficult
impact on the environment can no longer be ignored’-3. than those established for energy accounting or for esti-
Any manufacturing process carries with it an environ- mating cost.
mental impact. For each type of impact, a natural This paper concerns one route towards achieving
recovery-rate exists (though it may vary enormously these goals. It describes how an established methodol-
from tropical jungle to high sierra), implying that a ogy for materials selection4 might be adapted to the task
steady state is possible, at least in principle. of minimizing the environmental impact of material
Increasingly, concern is expressed that present-day usage. In doing so, it is helpful to divide origins of envi-
processes cause impact at a rate which exceeds the ronmental impact in engineering production and use
capacity for recovery, leading to accumulation of into three broad classes.
damage.
Materials contribute to this damage at three points:
in their production, in the use of products made from Inefficient use of materials: the best defence against
them, and in the disposal of these products. Minimizing waste and the problems of its disposal is to decrease
the damage requires the selection of materials and the amount of waste that is produced in the first
processes which are less toxic, and can give products place by efficient design’-‘.
which are easier to recycle, lighter and less energy- Consumption of non-renewable resources: the life
intensive; and this must be achieved without compro- cycle of any engineering structure involves the
mising product quality. Different materials differ consumption of energy and materials, both very
greatly in their impact: by almost any measure, much derived from non-renewable resources.
cadmium is damaging, wood is benign. Specific damage to air, water and land, by chemical
There are two alternative views on how the negative contamination, particulates and solid wasteG8.
impacts of materials on the environment should be
countered. The first is to treat environmental damage as
a design constraint, listing permissible levels of impact To explore ways of choosing materials to minimize
and insisting that the actual impact level does not environmental impact it is helpful first to examine how
exceed this target. The second is to treat the minimiza- materials are selected to meet a design specification4,9.
Materials & Design Volume 17 Number 1 1996 11
Selection of materials to reduce environmental impact: P. M. Weaveret al.

Materials selection methodology identified without solving the complete design problem,
Function, objective und construints or even knowing all the details of F and G. This enables
Any engineering component has a junction: to carry considerable simplification: the performance for all F,
bending moments, to contain a pressure, to transmit G and S is maximized by maximizing f,(M), which is
heat, etc. In designing the component, the designer has called the ‘performance index’. Experience shows that
an objective; to make it as cheap as possible, or as light, the groups usually are separable. Numerous examples
or as safe, perhaps. This must be achieved subject to are given elsewhere3r5s6.
constraints: that the component can carry the given The steps in deriving an index are as follows:
loads without failure, that certain dimensions are fixed,
and that its cost is within certain limits. Function, 1. Identify the PRIMARY FUNCTION of the
objective and constraints (Tubfe I) define the boundary component for which a material is sought. A beum
conditions for selecting a material and - in the case of carries bending moments; a heat-exchanger tube
load-bearing components - a shape for its cross-section. transmits heat; a bus-bar transmits electric current.
The optimal selection of material for an engineering 2. Write down an equation for the OBJECTIVE; it is
component is usually overconstrained; and, additionally, called the ‘objective function’. The objective is the
it must commonly meet several compound objectives. first and most important quantity to be minimized
Methods for achieving this, based on the use of muter- or maximized. Commonly, it is weight or cost; but
iul indices and selection charts, are developed below. it could be energy dissiputed in i2R heating (a bus-
bar), or environment& impuct (refrigerator insula-
tion materials) - it depends on the application. In
The index-und-chart method general, the objective function contains one or more
Two concepts are used in the selection procedure. The free variables: dimensions (such as the cross-section
first is that of ‘performance indices’ which isolate the of the beam or the wall thickness of the heat-
combination of material properties and shape informa- exchanger tube) which are not specified, and which
tion that maximize performance; the second is that of we are free to choose.
materials-selection charts. These are discussed in detail Eliminate the free variable(s) in this equation by
elsewhere3. The main features are summarized briefly using the CONSTRAINTS. They are the design
here. requirements which must be met, and which there-
fore limit the optimization process of step (2).
Perjkwmunce indices. The idea of performance indices Commonly these are: a required value for the st@
in assessing the merits of a particular material in any ness S; a required value for the suf; loud F, or
design is a powerful one. The design of a mechanical moment M or torque T or pressure p that can be
component is specified by four groupings of variables: supported; a limit on operuting temperuture T,,,,,; or
the functional requirements (need to carry loads, trans- on resistance to sudden fracture, measured by the
mit heat etc.) F; the specifications on geometry, G; the jkcture toughness K,,.
properties of the material of which it is made, M; and Read-off the grouping of material properties (called
the section shape, S. The performance P of the compo- the PERFORMANCE INDEX) which maximize
nent can be described by an equation with the form the value of the objective.

P = f(F, G, M, S) (1) Applying the criteriu: selection churts

Figure 2 shows the steps involved in a single selection
where P may be its mass, or volume, or cost, or life, for stage. Typically, we have a performance index, M, made
example; and ‘f means ‘a function of. Optimum design up of two or more properties P, and P2 - in this
can be considered to be selection of the material and
geometry which maximize (or minimize) P. The opti-
mization is subject to constraints, some of them
imposed by the material properties. The four groups of
parameters in equation (1) are said to be ‘separable’
when the equation can be written

P = f,(F).f2(G).f,(M).f,(S) (2)

where f,, f2, f3 and f4 are functions. When the groups are
separable, the optimum choice of material becomes
independent of the details of the design: it is the same
for all geometries G, and all values of the functional
requirements F. Then the optimum material can be

Table 1 Function, objectives and constraints

Function What does component do?

Objective What is to be maximized or minimized? DENSITY p (Mg/m’ 1
Constraints What non-negotiable conditions are to be met?
What negotiable but desirable conditions .?
Figure I A material-selection chart

12 Materials & Design Volume 17 Number 1 1996

 Selection of materials to reduce environmental impact: P. M. Weaver et al.

example they are Young’s modulus, E and density, p.

We create a selection chart with axes of Pi and P,. The
performance index, M, plots as a diagonal line on the
chart. Its slope is important. A simple performance
index, typically, has the form

M+ (3)
2

The default axes of the charts are logarithmic. Taking

logs of the above equation gives:

log P, = n log P, + log M (34 I

0.01 L I I I
O.Ol 0.1 1 10 100
A line of slope n on the log-log plot describes the index; INDEX M,
its position is determined by the value of M. Moving
the line changes the value. The selection is optimized by Figure 2 The proper comparison of A with competing materials can
be made for various values of the weight-cost exchange constant Ca.
moving the line to the highest value of M which still When C, = IO. C is a better choice than A: when C, = 0. I, D is better
leaves a viable subset of materials exposed. than A

Environmental impact versus functional energy a total lifetime cost can be established. Then the
performance value, V, in equation (4) becomes the total lifetime cost
The index-and-chart method outlined above can be and C, is the cost of electrical energy.
readily extended for the case of compound objectives: Furthermore, the value function method can also be
the task of meeting two objectives in one designi2. For used to relate seemingly unrelated objective functions.
this scenario the design is best analysed by forming the Again using a refrigerator as an example we consider
performance index for each objective in turn, and the increase in amount of money a consumer is
combining the two in an appropriate way to form a prepared to pay for an increase in fire resistance. The
vulue function. This procedure can be applied to envi- performance index that minimizes the cost of material is
ronmental design. Here, an environmental objective is once again given by M, in equation (4). The perform-
formed and its performance index, M,, is calculated; the ance index which gives a measure of fire resistance is
result giving a measure of environmental impact. M,. The exchange constant is found as follows. Note
Environmental impact can then be compared with the that for the general case and a fixed value of V
traditional function (design) objective (characterized by
the index M2), by making a chart with these quantities -=Ac -c (44
E
as the axes. Selection is made by plotting coupling t AM, 1
contours onto the selection chart, as illustrated in
Figure 2. A coupling contour is a plot of the value func- that is, C, measures the change in cost, AC for a given
tion by change in M,, at constant V. Then for our example, C,
is the increase in price a consumer is prepared to pay
V= C,M, +M, (4) for an increase in fire resistance. These examples are
discussed in more detail in the following section.
Here, C, is the appropriate exchange (coupling)
constant; it can be thought of as a weighting factor, or
- more precisely - as ‘the relative value of environmen- Simple measures of impact: a case study on
tal impact and performance’. The power of this method refrigerator insulation
is that potential materials are compared with a reference Environmental impact is difficult to quantify. One
material (A in Figure 2) in the light of both performance component of this impact relates to the energy content
indices. All materials that lie on a contour have the of the material, and that cun be quantified. We shall use
same overall value (e.g. life-cost) as the reference ma- this, together with the energy dissipated during the life
terial. All materials that lie above the line have a lower of a component and fire resistance of a material, as
life-cost and are a better choice while those that lie examples in the design of insulation for refrigerators,
below are worse. The slope of the contour is dependent returning to the more general measures of environmen-
on C,: its position depends on V. tal damage at the end of the paper.
The method is best explained by means of an
example. Consider the insulation of refrigerators. The Muterids for rejiigerutor insulation
manufacturer may wish to minimize the cost of the Many refrigerators use polyurethane foam as the insu-
material, while, on the other hand, a consumer may lation material.* We choose a polyurethane foam with
want to minimize the energy lost over the lifetime by
conduction through the refrigerator walls. These objec-
tives can be transformed into performance indices that * Chlorofluorocarbon foams (CFCs) are not considered in this case
study, as they have been excluded for use in refrigerators in Europe
are given by M2 and M,, respectively. By relating the and other countries. However, it should be noted that their thermal
energy losses to a cost through the price of electrical conductivities are among the lowest available for foamed materials.

Materials & Design Volume 17 Number 1 1996 13

 Selection of materials to reduce environmental impact: P. M. Weaveret al.

- 0.6m -__) Table 2 Design requirement for insulation

4 ~:.~..;.:.:.~..::..:.:~~:~..:~.:;:.::’~
1 INSULATION Function Refrigerator insulation
Objectives
Functional Minimize cost of insulation while satisfving
. L

constraints
Environmental Minimize energy consumption over lifetime
Safety Material needs to have good fire resistance

Constraints Thickness I 2 cm
-x- Lifetime > 10 years
Outside/inside temperature difference > 16°C
=2cm
Depth = 0.6m
AAT
uT O T = xpq+- [I

=-- 1 ATf;+x (9
Ml x M3

with M, = l/A and M, = llqp, and where A. is the

l Weight of PU(O.oS) thermal conductivity of the material, x is its wall thick-
foam - 1Okg ness (2 cm), t, is the lifetime of the insulation (10 years),
l Typical cost of fridge q is its energy content and p is its density. AT is the
= f250 of which - f40 temperature difference between the inside and the
is cost of insulation outside of the insulation layer. Each M is a performance
index, to be maximized if each contribution was treated
Figure 3 Insulation reduces the loss of energy from a refrigerator,
but contains energy itself. The best choice minimizes the total life- separately. Figure 4 shows the contribution from each
energy. An existing foam is polyurethane (density 80 kg/m’) energy to the total by plotting M, against MJt. The
better the material, the greater the value on each axis.
density of 80 kg/m3 as our reference material. Using the Superimposed on the chart is a contour of total energy
typical dimensions of a refrigerator shown in Figure 3, using polyurethane from (density 80 kg/m)) as a refer-
a total insulation mass of approximately 10 kg can be ence. This is obtained directly from equation (5) but
calculated. This translates into a cost of around &40 for plots as a curve because of logarithmic scales. All
this foam. Refrigerators of this type sell for a price near materials which cut through such a line have the same
f250, which shows that insulation material costs energy associated with them over their lifetime. The
account for around 15% of the selling price. With this contour, for this example, is near-vertical, indicating
knowledge we shall examine whether other foams that the energy content of the material is small
perform the same function more economically and compared with the energy losses over the lifetime.
explore also certain environmental issues associated Therefore, to a good approximation, the total energy
with the choice of material. associated with a material can be measured solely in
The traditional objective, in insulating a refrigerator terms of energy losses over the lifetime of the refrigera-
(Figure 3) is to minimize the material cost while simul- tor.
taneously satisfying constraints such as wall thickness The calculation can be adapted to minimize cost,
and maintaining the interior at the required temperature rather than energy. Equation (5) the value function,
(-4°C). In recent years, however, environmental as well becomes, instead
as thermal management issues influence the design. Two
types of environmental impact are considered in this AAT
example. The first concerns energy losses due to conduc- cTOT = xcmP+- WE

tion through the refrigerator’s walls over its lifetime. But X AT$a 1 (6)
=-++-
the insulation has a finite energy content itself, and it has
a finite life. The most environmentally sound choice of M, x M,
material for insulation is that which minimizes the total.
Furthermore, by introducing the cost of energy the with M, = l/&p and MZ = l/it. The cost of energy, C,,
energy cost can be summed with material cost to give a has been introduced to convert energy losses to an
lifetime cost. A second environmental issue - fire resis- energy cost. In practice, this is the cost of electricity to
tance - influences safety aspects of design. the consumer. Here, a figure of f lO/GJ is assumed for
There are at least three constraints. There is an upper illustrative purposes.
limit on the thickness of the insulation. The design life Figure 5 shows a chart of M, against Mz.
is 10 years and the internal temperature must be held Traditionally the best materials would have been chosen
below 4°C (Table 2). on cost alone as indicated by large values of M,. On
The total energy dissipated per unit area during the environmental grounds the best materials have large
lifetime of the insulation is the sum of energy content of values of M,. However, the two competing objectives
the material and the energy loss over the lifetime giving
the value function:
’ All charts are constructed using the Cambridge Materials Selector”.

14 Materials & Design Volume 17 Number 1 1996

 Selection of materials to reduce environmental impakt: P. M. Weaver et al.

MELAMINE-DENSIM (0 01)

POLYURETHANE-PU (0 019)

......-LIFE-ENI
..._.....
PHEN (Oh)
-

SE?,
PS (0.05)
c

Po&hacrylimide Foam. RigidbhWR (0.110)

\ I
Y
-%--- --z
l/The:smal Conducti vity(mKM? M,
Figure 4 Insulation reduces the loss of energy from a refrigerator, but contains energy itseff. The contour shows all materials with the same iife-
energy as polyurethane (density 0.08 Mg/m)). The total life-energy contour is near-vertical, indicating that energy content of the material is small
compared with energy losses over the lifetime

are linked through C,. Superimposed on the chart is a chart are three contours of constant value using
contour of constant total cost using polyurethane polyurethane (0.08) as the reference material. Each
(0.080) as the reference material.* All materials with contour is associated with a different perceived value of
lower lifetime cost lie towards the top right hand quad- fire resistance. The lower contour is given by the lowest
rant as indicated by large values of M, against M,. The premium for a 10% increase in fire resistance, that is,
results are analysed in the next section. only &lo. The middle contour is for g30 and the upper
Safety aspects are often considered alongside envi- one for &loo. An indication of the appropriate accurate
ronmental concerns. Fire resistance, in particular, could figure could be obtained from consumer surveys. The
be identified as an issue for which consumers are trend of each contour is easily explained. At high values
prepared to pay a premium. If the amount of money of fire resistance (large M4) the material cost is higher
could be identified for which customers are willing to (low M,), as expected from the increase in price a
pay extra for an increase in fire resistance the analysis consumer is prepared to pay. On the other hand,
above could be adapted to two separate objectives. The materials with worse fire resistance than polyurethane
lifetime cost calculations assumed that the cost of elec- (0.08) are cheaper. The contour represents a curve of
trical energy, C,, was known. To relate fire resistance to constant value.
cost of material it is necessary to determine the per- The two limits of perceived value are also examined
ceived increase in value to the consumer associated with on this chart. If customers do not value fire resistance at
an increase in fire resistance. Then all, i.e. they are willing to pay &O for increase in fire
resistance, the contour of constant value is a horizontal
Increase in value = value added due to increase in fire line passing through the reference material (poly-
resistance urethane 0.08). Conversely, the greater the premium a
- increase in material costs consumer is prepared to pay for an increase in fire resis-
tance, the more vertical the contour becomes, again
Here, it is proposed that customers are prepared to pay passing through the reference material.
for fire resistance.We examine three levels of this will-
ingness: f 10, f30 and f 100 per 10% increase in fire resis-
Results
tance. A useful measure of fire resistance in polymeric
The lifetime cost of a refrigerator with polyurethane
foams is the oxygen index. This figure gives the percent-
(0.08) as an insulation material is equal to approxi-
age of oxygen in the atmosphere needed to sustain
mately ~5600. It is made up of f40 on PU (0.08), f200 on
burning, the larger the figure the more fire-resistant the
the rest of refrigerator and f360 on energy losses over
material. Figure 6 shows a chart of M, versus M, where
the latter is the oxygen index. Superimposed on the 10 years, as calculated by inserting the constants in
equation (6).
The results of the selection strategy are summarized
in Tuble 4. Each row corresponds to a different envi-
* The numbers in brackets following the material name in Figures 4-6 ronmental indicator. The first row shows the relative
are the densities in units of M&m’. lifetime cost of alternative foams compared to a refrig-
Materials & Design Volume 17 Number 1 1996 15
Selection of materials to reduce environmental impact: P. M. Weaver et al.

i p” (O.lQ T~LOW ENERGY I

~-a
O-l-.~~~..~~i-~-~--i~~.~~. .*.--...i,,‘..,,..,i ~r..-.r,lr,,...r-rI-r--“‘r’
30 32 34 36 38 40 42 44 46 48 50
l/Thermal Conductivity (mK/W) I&
Figure S Insulation reduces the loss of energy (which has a value in f) from the refrigerator, but has an intrinsic cost itself. The choice which
minimizes the life-cost of the system can be made with this chart, again allowing for a constraint on total thickness

erator with PU (0.08). The phenolic foam of density 35 Again the phenolic foam rates highly. For the same cost
kg/m3 shows a lifetime cost saving of 20% (f120) over refrigerator, phenolic foams are 29% better in value for
PU (0.08). money in terms of fire resistance.
The second row shows the results of added value to
the consumer, for the same cost, associated with an Discussion
increase in fire resistance. The results are normalized to Before phenolic (0.035) foam is confirmed as a better
the cost of a refrigerator with PU (0.08), i.e. E250. environmental foam for insulation in refrigerators other

Insulation Materials for

Refrigerators: Chart 3
: mw”DDlm

tl-~--‘-.--t-------q
bl c
‘. \

Oxy& Index (Fi&Re~istmce) (%) M.

Figure 6 Polymeric foam materials are often highly flammable, The premium a customer is willing to pay for increase fire resistance can be used
to plot contours of constant value

16 Materials & Design Volume 17 Number 1 1996

 Selection of materials to reduce environmental impact: P. M. Weaver et al.

Table 4 Results of environmental insulation material selection

PU PU PU PVC PVC PVC Phenolic foam

0.08 0.032 0.028 0.06 0.046 0.03 I 0.035

Lifetime cost of refrigerator/lifetime I 0.99 0.98 1.04 0.96 0.91 0.80

cost of reference
Added value cost/fire resistance I 1.12 1.13 0.88 0.93 0.99 1.29

factors should be examined. These include material here too, if appropriate charts could be constructed.
properties such as ease of recycling, biodegradability And therein lies the challenge: devising quantitative
and toxicity.The manufacturing process should be measures of environmental impact (generalized replace-
examined to identify the amount of gaseous emissions ments for qp) against which design-limiting properties
(NO,, SO,, CO,) and other contaminants. In addition, (strength or thermal conductivity) can be plotted.
the toxicity of component chemicals and the finished One candidate is cost: the cost, per m3 of material, of
material could be analysed. exactly reversing the damage that creating, using and
The selection chart method outlined here, whereby a disposing of the materials has caused. The difficulties
traditional functional objective is plotted against an involved can be seen by attempting to apply it to our
environmental impact objective, could be applied to all restrictive case. The cost of energy is, of course, known;
these environmental measures. The objectives are it is currently about f10 per GJ. But the energy content
plotted in terms of performance indices. The two or of a material measures more than that: it also includes
more objectives are combined to give a value function a measure of the COZ, SO, and H,S produced, the
by expressing one in terms of the other via an exchange depletion of a natural energy resource, and the dust and
constant, C,. Examples considered here include the slag produced. Energy content was chosen for the
energy loss for the insulation material which is charts shown here because it has this greater dimension.
converted into a cost using the cost of energy, and fire It is clear that a need exists for quantitative account-
safety, which is converted to cost with a perceived ing methods for environmental impact. As mentioned
benefit. But before this can be done, data of sufficient earlier, they need not be exact. As with energy content,
accuracy are required. If stored in ranges (lower and different materials differ greatly. Useful conclusions can
upper bounds) the data do not need to be exact. This, be drawn from approximate data. The measures will
at the present time, is the most urgent need. require debate and will change with time, but tech-
Having identified the most energy-efficient material niques for obtaining them are only a little more difficult
for refrigerator insulation there is still a question of the than those established for energy or cost accounting.
mechanism or market forces that will cause this design These, with methods like those described in this paper,
solution to be chosen. One possible scenario is that are a logical step forward in environmentally conscious
manufacturers, themselves, will make a marketing deci- engineering.
sion to promote the design that gives the best lifetime
energy performance, even though there may be a References
penalty on the initial purchase cost. This approach 1 Forrest, D. and Szekely, J. Journul of’Mefub 1991, December, 23
becomes more viable if the government makes it 2 Navichandra, D. Design for environmentability. ASME Design
mandatory for every refrigerator manufacturer to state Theory and Methodology Conference, American Society of
the lifetime energy-efficiency of their refrigeratory so Mechanical Engineers, Miami, Florida, USA, 1991
Geiser, K. The greening of industry. Technology Review 1991,
that a potential customer is more confident about the August/September
decisions on lifetime costs. Ashby, M. F. Muterids Selection in Mechunicul Design,
Pergamon Press, Oxford, 1992
Beitz, W. and Grote, K.-H. Design for material saving. Muteriuls
Conclusions and challenges & Design, 1994, 15 (4)
If engineers are to minimize the environmental impact Thurston, D. L., Lloyd S. M. and Wallace, J. Considering
customer preferences for environmental protection in materials
of the products they design, they require tools and selection. Muteriuls & Desk. 1994. 15 (41
Yl .--\I

methods to guide their choice of material and process. I Bor, J. M. The influence of waste strategies on product design.
Ideally, this methodology should point, in any given Muteriuls & Design 1994. 15 (4)
application, to the subgroup of materials which perform 8 Ishii, K., Eubanks, C. F. and bi Marco, P. Design for product
retirement and material life-cycle. Muterid & Design 1994,
best by environment-sensitive criteria. 15 (4)
This paper outlines one approach to the problem and Asib;, M. F. Materials and the environment. Phys. Stut. Sol.
illustrates it by taking two restricted measures of 1992. 131. 625-638
impact, the energy content and the fire resistance of the Ashdy, c. F. and Cebon, D. Cuse Studies in Muteriuls Selection.
Granta Design, 20 Trumpington St, Cambridge, UK, 1996
material. Energy content, as said earlier, is only one Weaver, P. M. and Ashby, M. F. The optimal selection of
measure of the environmental impact of material usage. material and shape. Journul oJ’ Engineering Design 1996. 1 (2)
In many circumstances it is not the important one; the Ashby, M. F. Muteridr Selection: Multiple Construints untl
emission of a toxic by-product, the difficulty of recy- Objectives, Cambridge University Engineering Dept Report
CUED/C-MATSflR229, September 1995
cling, or the resistance to biodegrading can be the real The Cumbridge Muteriuls Selector, (CMS) Software, Granta
environmental threat. The chart method could be used Design Ltd, 20 Trumpington St, Cambridge, UK, 1995

Materials & Design Volume 17 Number 1 1996 17