Resources, Conservation and Recycling 55 (2011) 1185–1195

Contents lists available at ScienceDirect

Resources, Conservation and Recycling

journal homepage: www.elsevier.com/locate/resconrec

Assessing the potential of yield improvements, through process scrap reduction,

for energy and CO2 abatement in the steel and aluminium sectors
Rachel L. Milford, Julian M. Allwood ∗ , Jonathan M. Cullen
Department of Engineering, University of Cambridge, Trumpington Street, Cambridge CB2 1PZ, United Kingdom

a r t i c l e i n f o a b s t r a c t

Article history: Targets to cut 2050 CO2 emissions in the steel and aluminium sectors by 50%, whilst demand is expected
Received 24 March 2011 to double, cannot be met by energy efficiency measures alone, so options that reduce total demand for
Received in revised form 24 May 2011 liquid metal production must also be considered. Such reductions could occur through reduced demand
Accepted 26 May 2011
for final goods (for instance by life extension), reduced demand for material use in each product (for
instance by lightweight design) or reduced demand for material to make existing products. The last
Keywords:
option, improving the yield of manufacturing processes from liquid metal to final product, is attractive
Yield
in being invisible to the final customer, but has had little attention to date. Accordingly this paper aims
Energy
CO2
to provide an estimate of the potential to make existing products with less liquid metal production.
Lightweighting Yield ratios have been measured for five case study products, through a series of detailed factory visits,
Steel along each supply chain. The results of these studies, presented on graphs of cumulative energy against
Aluminium yield, demonstrate how the embodied energy in final products may be up to 15 times greater than the
energy required to make liquid metal, due to yield losses. A top-down evaluation of the global flows of
steel and aluminium showed that 26% of liquid steel and 41% of liquid aluminium produced does not make
it into final products, but is diverted as process scrap and recycled. Reducing scrap substitutes production
by recycling and could reduce total energy use by 17% and 6% and total CO2 emissions by 16% and 7% for
the steel and aluminium industries respectively, using forming and fabrication energy values from the
case studies. The abatement potential of process scrap elimination is similar in magnitude to worldwide
implementation of best available standards of energy efficiency and demonstrates how decreasing the
recycled content may sometimes result in emission reductions.
Evidence from the case studies suggests that whilst most companies are aware of their own yield ratios,
few, if any, are fully aware of cumulative losses along their whole supply chain. Addressing yield losses
requires this awareness to motivate collaborative approaches to improvement.
© 2011 Published by Elsevier B.V.

1. Introduction reducing emissions in industry (Bernstein et al., 2007; International

Energy Agency, 2008). However, the only material efficiency mea-
The production of steel and aluminium is energy and CO2 - sure described in detail by the IPCC is material substitution and no
intensive; in 2006, these industries were responsible for producing estimates were found of the associated emission mitigation poten-
10% of global energy and process CO2 emissions (Allwood et al., tial. A survey of all material efficiency options was conducted by
2010a). Demand for steel and aluminium is predicted to at least Allwood et al. (2010b), outlining all available strategies for “provid-
double by 2050, whilst over the same time period, global emission ing material services with less material production and processing”,
reduction targets of 50% or more are being negotiated. Allwood et al. in order to reduce environmental impacts; one strategy identified
(2010a) have shown that even the aggressive pursuit of current and was yield improvement.
developing energy efficiency technologies is not enough to meet Yield losses along the supply chain necessitate the production of
these targets. So what options are there for reducing emissions more liquid metal than ends up in the final metal product. As liquid
further? metal production is often the most energy and CO2 -intensive pro-
The 2007 IPCC report and the International Energy Agency cessing step along the supply chain, reducing liquid metal demand
(IEA) list material efficiency as an important strategy for through yield improvements could have a significant impact on
energy use and CO2 emissions. In the IPCC’s table of selected exam-
ples of industrial technology for reducing emissions (Table 7.5,
∗ Corresponding author. Tel.: +44(0)1223338181; fax: +44(0)1223332643. Bernstein et al., 2007), the reduction of process losses is listed as
E-mail address: jma42@cam.ac.uk (J.M. Allwood). a material efficiency measure in the iron and steel sector, but the

0921-3449/$ – see front matter © 2011 Published by Elsevier B.V.

doi:10.1016/j.resconrec.2011.05.021
1186 R.L. Milford et al. / Resources, Conservation and Recycling 55 (2011) 1185–1195

From upstream process To downstream process

Process Lost

From downstream To upstream

processes for processes for
recycling/ Internal reprocessing recycling/
reprocessing reprocessing

Fig. 1. Input and output process flows.

abatement impact of this strategy is not explored any further in metal to the downstream process (Fig. 1), which must include the
the report. The IEA suggests that yield losses result in the loss of “wasted” embodied energy in the metal that does not continue
invested energy (International Energy Agency, 2007), for exam- straight down the supply chain. Yield data are sometimes reported
ple, direct casting of steel is credited with indirect energy savings without accounting for internal reprocessing of material as this can
compared with conventional casting due to reduced yield losses be difficult to measure. However, for the majority of processes, it
(International Energy Agency, 2008). So how can material efficiency is the total amount of metal entering a process that drives energy
through yield improvements be achieved? consumption, regardless of its origin, so in this analysis, internal
A great deal of literature exists on improving material utilisation reprocessing is included. Therefore, in this paper, yield and yield
in two stages of the sheet metal supply chain: blanking and stamp- loss are defined as follows:
ing operations. Different nesting algorithms have been proposed Metal to downstream process
in order to optimise yield for different shaped parts. However, the Yield =  , Yield loss = 1 − Yield
All process inputs
motivation for these yield improvements is the material cost bene-
(1)
fits, as in Nye (2001), rather than the energy or emissions benefits.
Additionally, this approach only addresses yield losses in part of
the supply chain, rather than the yield along the entire supply 2.2. Gathering case study data
chain from liquid metal to final product, and describes incremental
yield improvements rather than the maximum potential material To date, no complete data set of energy and yield figures
savings. has been reported for the fabrication of different metal products.
Whilst blanking and stamping losses are planned yield losses Accordingly, in the work leading up to this paper, a primary data
that are inherent to the process, unplanned yield losses also arise set has been collected for five case study products, through a com-
from issues of quality. This total view of yield loss is explored in the prehensive set of site visits. The case studies were chosen to span
work of Kamalapurkar and Date (2006), who model the failure rate a range of different types of metal products, for example, bulk and
of components by considering the statistical probability of quality sheet products, and a range of sectors, including transport, con-
failures from analysis of the areas of greatest stress during the man- struction and packaging. The case study products investigated are:
ufacturing process, as well as looking at planned losses due to the a 500 kg steel I-beam; a steel and an aluminium outer panel for a
blank layout. front car door; an aluminium beverage can body; an aluminium
The objective of this paper is to determine whether yield wing skin panel for an aeroplane.
improvement is a potentially significant energy and CO2 abate- The aim of the case studies is to evaluate the energy and CO2
ment strategy, firstly, by evaluating the impact yield has on the impact of yield losses from liquid metal to the final product. To
embodied energy and CO2 of five case study products, and sec- do this, the yield, energy consumption and CO2 emissions for each
ondly, by evaluating global steel and aluminium flows to estimate process from liquid metal production to the final fabrication step
the global emissions abatement potential of eliminating process were collected and combined along the supply chain to give values
scrap. for the cumulative yield, energy consumption and CO2 emissions.
Cumulative values were chosen as this is equivalent to the embod-
2. Methodology ied energy or CO2 of the output of each process. The energy and
CO2 impact of yield losses is then evaluated by comparing the
This section describes the methodology used in this analysis. embodied energy or CO2 of liquid metal with that of the final
Section 2.1 provides the definition of material yield that determines product.
energy consumption; Section 2.2 summarises what information The embodied energy and CO2 of liquid metal is calculated using
was collected for the case studies; Section 2.3 describes how the yield, energy and CO2 emission information on upstream processes
results of the case study are presented, and Section 2.4 describes collected predominantly from the literature, where global, aver-
how energy data from the case study results are used with global age data exist. Equivalent information for the fabrication processes
flows of metal to provide an estimate of the global impact of downstream of liquid metal production is not currently available
yield. in the literature and so was collected from communications with
six companies and by physically walking down the supply chain
2.1. Defining yield in seven production sites. Due to the sensitive nature of some of
the data collected, these companies and sites will not be identified
The term “yield” can be defined in several different ways, but, the data used are collected from current operations of major
depending on the process characteristic of interest and the mate- companies active in producing these components.
rial flows included in the definition of the yield ratio. For example, Yield data for the fabrication stages were also collected during
yield may be defined as the proportion of material that remains the site visits, where possible, to ensure it was consistent with the
in the metal cycle, i.e. that is not lost or dissipated to the envi- definition of yield in Section 1. The cumulative yield describes the
ronment. In this analysis, the outflow of interest is the flow of proportion of the original liquid metal that continues down the
 R.L. Milford et al. / Resources, Conservation and Recycling 55 (2011) 1185–1195 1187

Fig. 2. Example cumulative energy versus cumulative yield graph.

supply chain after each process, assuming the supply chain started
with 1 t of liquid metal, and is defined as follows:

For 1 < i ≥ n, where n is the number of process steps Cumulative yieldi = Cumulative yieldi−1 × Process yieldi
(2)
For i = 1, Cumulative yield0 = 1

Energy data for fabrication stages were collected from in-line

metering, where possible, but for some sites, process energy data 2.3. Data representation
had to be allocated from line or plant-wide metering. Final energy
values collected from site visits were converted into primary energy Having collected yield, energy and CO2 data for the case study
values using the process fuel mix and the same conversion fac- products, how can the impact of yield losses on embodied energy
tor as Worrell et al. (2008): the efficiency of electricity production and CO2 be communicated? Although the data collected are specific
is assumed to be 33%, to account for generation and transmission to a single process, embodied characteristics are determined by
losses. Where necessary, process energy values were converted into considering cumulative effects along the supply chain. However,
the form GJ/t process output. The cumulative energy describes the process energy or CO2 values cannot be added together directly to
absolute energy invested in the mass of metal passing down the give the cumulative value; upstream process values must be scaled
supply chain and is defined as follows: to include the effects of yield, as shown in Eq. (3).
Cumulative energyi−1 Process values can be scaled to assume either a constant output,
Cumulative energyi = Process energyi + i.e. cumulative embodied energy or CO2 per tonne of output from
Process yieldi
the process, or assuming a constant input, i.e. cumulative embodied
(3)
energy or CO2 per tonne of liquid metal input at the start of the
supply chain. The novel presentation of cumulative energy, CO2
As the majority of the fabrication sites visited were in the UK, and yield used in this paper illustrates both of these perspectives.
an emissions factor reflecting the UK’s electricity mix is used along An example graph of cumulative energy against cumulative
with the process fuel mix to infer the CO2 emissions for each yield is shown in Fig. 2. The amount of metal processed at each
process. The emission factors for fuel and electricity are taken stage is determined by the cumulative process yield, depicted on
from the Department for Environment Food and Rural Affairs and the x-axis. The contours of the graph (where y/x = constant) show
Department of Energy and Climate Change (2010). For some pro- the cumulative embodied energy per tonne of process output. In
cesses, data on the fuel mix were not available and so had to be effect, these show the “embodied energy” of the product, whereas
estimated. Cumulative CO2 emissions for each process are calcu- the y-axis shows the total energy expended up to and including
lated as follows: each process, assuming the supply chain started with 1 t of liquid
metal. In the example shown in Fig. 2, 1 t of liquid metal could be
Cumulative CO2 emissionsi = Process CO2 emissionsi used to make about 520 kg of component C and over 8.4 GJ of energy
would have been used. The embodied energy of component C is
Cumulative CO2 emissionsi−1 shown by the contours, and is about 16 GJ per tonne of component
+ (4)
Process yieldi C, i.e. 8.4 GJ/0.52 t.
Fig. 2 has been normalised per tonne of liquid metal; an alter-
Eqs. (3) and (4) assume a linear supply chain, i.e. all process scrap native presentation would be to show the non-normalised, or
is sent for recycling rather than for further fabrication into products. absolute, graph of cumulative energy against cumulative yield. This
This was representative for the case studies observed, but may not would show the actual mass of metal and energy required at each
be the case for all steel and aluminium supply chains. process to make a specific product. The shape of the graph would
1188 R.L. Milford et al. / Resources, Conservation and Recycling 55 (2011) 1185–1195

Table 1
Global data from Allwood et al. (2011a).

Metal Liquid metal (Mt) Semi-finished products (Mt) Final products (Mt)

(a) Global metal flows

Steel 1400 1280 1040
Aluminium 76 56 45

Metal Ore Forming scrap Fabrication scrap End-of-life scrap

(b) Liquid metal sources

Steel 64% 6% 15% 15%
Aluminium 50% 14% 26% 10%

be the same, however, the values on the axes would be differ- abatement strategies on cumulative yield and energy for the steel
ent. Similar normalised or absolute calculations and graphs can be car door panel. Top-down analysis based on global metal flows is
constructed for CO2 emissions. presented in Section 3.4 in order to determine whether process
scrap reduction is a significant energy and CO2 abatement strategy.
2.4. The global impact of yield
3.1. Process data for the case studies
The global implications of yield losses along the supply chain
are evaluated using data on the global flow of steel and aluminium The yield and energy data collected for processes all along the
taken from Allwood et al. (2011a) and are shown in Table 1a and b. supply chain for each case study product and their sources are sum-
Liquid metal is cast and formed into semi-finished products, such as marised in Tables 3 and 4, and discussed in detail in Supporting
coil, plate, rod and bar. Semi-finished products are then fabricated Information (SI). Unlike the aluminium industry, the steel indus-
into final metal products. try does not collect yield and energy data for mining, therefore the
Table 2 shows the values of energy and CO2 intensity of liquid steel supply chain starts with coking and sintering. For some pro-
metal production from ore and from scrap used in this analysis, cesses, yield or energy data were not available, so estimates were
including their sources and derivations.
The total energy required to produce liquid metal is defined in
Table 2
Eq. (5), where εprimary is the energy intensity of production from Energy and CO2 intensity values.
ore and εsecondary is the energy intensity of production from scrap.
Metal Steel
Liquid metal energy = Global mass of liquid metal Primary energy 18
intensity (GJ/t) 1 t of primary liquid steel requires production of
× [(%scrap × εsecondary ) + (%ore × εprimary )]
1600 kg sinter, 390 kg coke, 1140 kg pig iron,
1140 kg crude steel and 1010 kg steel for refining
(5)
The energy intensity of production is 1.6 GJ/t
sinter, 3.2 GJ/t coke and 12.3 GJ/t iron in the blast
The total forming and fabrication energy is defined in Eqs. (6) furnace. Steelmaking in the basic oxygen furnace is
and (7). Representative forming and fabrication energy intensity exothermic and this energy, −0.26 GJ/t crude steel,
values, εforming and εfabrication are inferred from the results of the can be recovered in integrated steelworks. Refining
has an energy intensity of 0.4 GJ/t refined steel
case studies. (derived using energy and yield data from a and b )
Secondary energy 5.5a
Forming energy = Global mass of liquid metal
intensity (GJ/t) Primary energy value
Primary CO2 intensity 1.6c
× εforming (6)
(t CO2 /t) Average blast furnace – BOF route
Secondary CO2 0.4c
intensity (t CO2 /t) Average scrap – EAF route
Fabrication energy = Global mass of semi-fabricated products

× εfabrication (7) Metal Aluminium

Primary energy 168

intensity (GJ/t) 1 t of primary cast aluminium requires production
The total energy required to produce metal products globally is of 5272 kg bauxite, 1923 kg alumina, 1020 kg liquid
the sum of the liquid metal energy, forming energy and fabrication aluminium and 435 kg anode
The energy intensity of production is 0.1 GJ/t
energy values. An equivalent calculation for CO2 can be performed
bauxite, 15 GJ/t alumina, 132 GJ/t aluminium and
using CO2 intensity values instead of energy intensity values. 4 GJ/t anode (data from d , e , f )
Secondary energy 7.6a
3. Results intensity (GJ/t) Primary energy value
Primary CO2 intensity 8.6d
(t CO2 /t) Excluding PFCs
The previous section has defined yield, described the set of case Secondary CO2 0.43
studies, proposed a novel presentation for the data and explored intensity (t CO2 /t) Final energy value from a multiplied by electricity
how to scale the case study results to global estimates. This section emission factorg
now reports and discusses the results of the data gathering, and a
Worrell et al. (2008).
the implications of these results will then be discussed. Process b
International Iron and Steel Institute (1998).
c
yield, energy and CO2 data for the case studies are presented in International Energy Agency (2007).
d
International Aluminium Institute (2007).
Section 3.1. The cumulative yield, energy and CO2 results for the e
International Aluminium Institute (2010).
case study products are presented as general results in Section 3.2 f
BCS Incorporated (2007).
and demonstrate the impact of yield losses on embodied energy g
Department for Environment Food and Rural Affairs and Department of Energy
and CO2 in final products. Section 3.3 presents the effect of three and Climate Change (2010).
 R.L. Milford et al. / Resources, Conservation and Recycling 55 (2011) 1185–1195 1189

Table 3
Steel process summary table.

Process Yield (%) Primary energy Emissions t Process Yield (%) Primary energy Emissions t
GJ/t output CO2 /t output GJ/t output CO2 /t output

Sintering 82 1.6 0.27 Fabrication 99 0.30 0.02

Coking 100 3.2 0.30 Pickling 95 0.75 0.04
Ironmaking 99 12 0.50 Cold rolling 99 0.9 0.05
Steelmaking 88 −0.3 0.10 Galvanising 91 1.5 0.09
Refining 99 0.4 0.02 Slitting 98 0.1 0.01
Casting 97 0.1 0.01 Blanking 90 0.3 0.02
Hot rolling 95–97 2.2–2.4 0.17–0.19 Stamping 68 1.2 0.07

Table 4
Aluminium process summary table.

Process Yield (%) Primary energy Emissions t Process Yield (%) Primary energy Emissions t
GJ/t output CO2 /t output GJ/t output CO2 /t output

Bauxite extraction 86 0.1 0.01 De-reeling and lubrication 100 1.1 0.06
Alumina production 37 15 0.77 Cupping 85 1.6 0.09
Anode manufacture 98 4 0.62 Body making and trimming 98 11 0.61
Smelting 53 132 6.78 Casting 88 7.5 0.40
Ingot casting 98 1.4 0.13 Homogenising 99 1.4 0.07
Remelting 92 4.7 0.27 Edge cropping 92 0.1 0.01
Sawing 84–96 0.0–0.5 0.00–0.03 Solution heat treatment 99 5.7 0.28
Scalping 87–95 0.1–0.3 0.01 Levelling 99 0.3 0.02
Preheating 99–100 0.2–4.6 0.01–0.25 Stretching 99 0.2 0.01
Hot rolling 89–92 1.4–2.3 0.08–0.12 Precipitation heat treatment 99 1.0 0.06
Cold rolling 95 1.3 0.08 Testing 100 0.0–1.0 0.00–0.05
Annealing 100 1.2 0.07 Machining 29–55 1.1–3.2 0.05–0.16
Slitting 96 0.1 0.00 Inspection 100 0.0 0.00
Blanking 90 0.3 0.02 Packing 100 0.0 0.00
Stamping 72 1.2 0.07 Releasing 100 0.5 0.03

made. For example, the energy of the blanking process for the car allocated between processes. The error bounds are estimated to
door panel was assumed to be half of the energy for the beverage be ±20%, which is the approximate contribution of the auxiliary
can cupping process, which involves blanking and some drawing energy to the overall process energy.
of the cup. As far as possible these were verified as representative
by the relevant companies and are identified in SI. 3.2. General case study results
Collecting the yield data for some processes presented chal-
lenges. Yield data were not always available for both technical The normalised cumulative energy against cumulative yield
and commercial reasons. Data on planned blanking and stamping graphs for the steel and aluminium case studies are shown in
losses were more readily available than yield data including data Figs. 3 and 4. As the energy required to produce the liquid metal
on quality losses. Collecting process-specific energy data was also is much greater than the energy used in forming and fabrication
non-trivial as few sites had in-line energy metering, relying instead processes, the y-axis and data sets for both graphs do not start at
on company and author estimates from line or plant-wide meter- 0 GJ in order that the forming and fabrication steps can be illus-
ing and, where available, auxiliary energy consumption had to be trated more clearly. The steel graph starts with 1 t of liquid steel,

Fig. 3. Cumulative energy against cumulative yield – normalised steel.

1190 R.L. Milford et al. / Resources, Conservation and Recycling 55 (2011) 1185–1195

Fig. 4. Cumulative energy against cumulative yield – normalised aluminium.

assumed to have a scrap content of 20%, which is the minimum finishing, more than in forming and fabricating the product from
scrap content used in the basic oxygen furnace (Corus Construction liquid metal. This is an interesting finding and further work could
Centre, 2001), and therefore an embodied energy of 14 GJ/t liquid establish whether this is the case for other steel and aluminium
steel. Whilst steel products may also be produced via the electric products.
arc furnace route, the supply chains observed in the case studies Figs. 5 and 6 show graphs of normalised cumulative CO2 against
went via the primary BOF route, which is considered representa- cumulative yield for the steel and aluminium case studies respec-
tive of I-beam and sheet production in the UK. The aluminium graph tively. For these graphs, the y-axis is shown starting from the origin
has two starting points, reflecting the two different scrap contents (unlike Figs. 3 and 4) to demonstrate the difference in emissions
observed during the site visits. The liquid aluminium for the wing arising from liquid metal production and component fabrication.
skin panel has a scrap content of about 40% and an embodied energy Intermediate product labels are not included for clarity, but are the
of just over 100 GJ/t liquid aluminium. The liquid aluminium for the same as in Figs. 3 and 4.
beverage can and car door panel has a scrap content of about 50% The general form of all of the energy and CO2 graphs is a rotated-
and an embodied energy of 86 GJ/t liquid aluminium. L shape. The size of the vertical stem is determined by the energy
The curve for the aluminium beverage can is shown as a solid line required or CO2 emitted when producing liquid metal. The size of
up until the production of the trimmed can body. The downstream the vertical stem is also influenced by the scrap content of the liq-
coating and decorating stages are shown as a dotted line as equiv- uid metal, the greater the scrap content, the less primary metal
alent data were not collected for the car door and wing skin panels required and the shorter the stem. The size of the vertical stem is
and suggests that significant energy is used during painting and important for the embodied characteristics of the component: the

Fig. 5. Cumulative CO2 against cumulative yield – normalised steel.

 R.L. Milford et al. / Resources, Conservation and Recycling 55 (2011) 1185–1195 1191

Table 5
Embodied energy of case study products.

Product Embodied energy (GJ/t)

(a) Steel
Liquid steel 14
I-beam 19
Car door panel 42
(b) Aluminium
Liquid aluminium (50% scrap) 86
Liquid aluminium (40% scrap) 100
Car door panel 220
Beverage can 210
Wing skin panel 1500

taller the stem, the greater the gradient of the contours of embod-
ied energy or CO2 , therefore, the greater the absolute increase in
embodied energy or CO2 for the same yield loss. Fig. 6. Cumulative CO2 against cumulative yield – normalised aluminium.
The size of the horizontal branch is determined by the size
of yield losses in the forming and fabrication processes. Greater
yield losses were observed for products made from sheet (car door
panel and beverage can) than from bulk products (I-beam). Whilst the use phase energy over the life of the component is included after
databases of the embodied characteristics of semi-finished prod- fabrication. Also shown are the predicted effects of three abate-
ucts (plate, sheet, etc.) exist, for example, the Bath Inventory of ment strategies: a 20% improvement in energy efficiency; a 20%
Carbon and Energy (Hammond and Jones, 2008) and the World improvement in process yield; and lightweighting the product by
Steel Association’s Life Cycle Inventory for metal products (World 20%.
Steel Association, 2010b), equivalent databases do not exist for The use phase energy was calculated assuming a compact car of
finished metal products (car door panel, beverage can, etc.). The mass 1260 kg has a fuel consumption of 200 MJ/100 km over a life of
significant impact of fabrication yield on the embodied energy or 200,000 km, based on data from the University of California Santa
CO2 of a finished product makes it difficult to estimate accurately Barbara’s greenhouse gas model (University of California Santa
embodied characteristics solely from the finished product weight Barbara, 2010). For the 1260 kg car, the energy usage was attributed
and embodied characteristics of semi-finished products. equally across the mass of the car, i.e. 317 MJ/kg. Mass reduction
Table 5 shows the embodied energy of the original liquid metal is not directly proportional to a reduction in fuel consumption;
and the finished component (excluding coating/decorating pro- assuming a 20% reduction in mass results in a 10% reduction in
cesses); the embodied energy of the finished product is up to fuel consumption, the total fuel consumption in the lightweighting
15 times greater than the embodied energy of the original liquid scenario will be 90% of the original fuel consumption. The data used
metal. to construct Fig. 7 are reported in SI.
The effect of a 20% reduction in the energy intensity of each
3.3. Absolute case study result – steel car door panel forming and fabrication process step is to shift the graph down the
y-axis. The total energy used is reduced by 1%. It is assumed that no
Due to the large difference in scale of the case study products, energy efficiency improvement can be realised for liquid metal pro-
it is not possible to plot the absolute cumulative energy and cumu- duction as it is already very efficient. The effects of a 20% reduction
lative yield for all products on the same graph. Fig. 7 shows the in process scrap are shown assuming the energy intensity of each
absolute energy and yield for the steel car door panel. An estimate of process is independent of throughput mass, and therefore process

Fig. 7. Cumulative energy against mass – absolute values.

1192 R.L. Milford et al. / Resources, Conservation and Recycling 55 (2011) 1185–1195

30

20
Total energy (EJ)

/t
GJ
14
Liquid metal
Forming
/t
GJ
Fabrication 15

Scenario with perfect yield

Actual production 2008

10

0
0 200 400 600 800 1000 1200 1400 1600
Mass (Mt)
(a) Steel

J/t J/t
Total energy (EJ)

Liquid metal
Forming 0G 88G
14
Fabrication

4
Scenario with perfect yield

Actual production 2007

2

0
0 10 20 30 40 50 60 70 80
Mass (Mt)

(b) Aluminium

Fig. 8. Global energy use in production of steel and aluminium products.

yield. This abatement strategy reduces the total energy used 3.4. Global results
by 2%.
Neither improvements in energy efficiency or process yield have Global energy use and the average embodied energy of liquid
any effect on the use-phase emissions. However, the lightweight- metal and finished metal products are shown in Fig. 8 for steel
ing strategy does. The effect of a 20% reduction in product weight and aluminium, comparing current production with a scenario
(assuming process yield and energy values remain constant) is to with perfect yield along the supply chain. Based on the findings
move the original data points towards the origin, until the mass on from the case studies, the average forming energy is estimated
the x-axis is at 80% of the original value. This abatement strategy to be 5 GJ/t and the average fabrication energy is estimated to be
reduces the total energy used by 11%. 1 GJ/t. The curve on the right presents the energy use and yield
Currently, use phase emissions are much greater than manufac- losses associated with 2008 production for steel and 2007 produc-
turing emissions, which reduce the impact of abatement strategies tion for aluminium. The curve on the left presents the theoretical
in the manufacturing phase. However, legislation on tail-pipe emis- energy use associated with 2008 or 2007 production assuming
sions and new technologies, such as hybrid and electric cars, all processes downstream of liquid metal production can elim-
may significantly reduce future use phase emissions. Therefore, inate process scrap. This scenario assumes all end-of-life scrap
embodied emissions abatement strategies will become increas- is still recycled and primary production makes up the remaining
ingly important. demand.
 R.L. Milford et al. / Resources, Conservation and Recycling 55 (2011) 1185–1195 1193

Table 6
Key results from global analysis.

Actual production Scenario with perfect yield

Steel Aluminium Steel Aluminium

Liquid metal demand (Mt) 1400 76 1040 45

Average liquid metal embodied energy (GJ/t) 14 88 15 140
Average liquid metal embodied carbon (t CO2 /t) 1.2 4.5 1.4 7.1
Formed product demand (Mt) 1280 56 1040 45
Final product embodied energy (GJ/t) 26 160 21 150
Final product embodied carbon (t CO2 /t) 2.0 8.0 1.7 7.5
Total energy use (EJ) 27 7.0 22 6.6
Total CO2 emissions (Mt CO2 ) 2100 360 1800 340

In the scenario with perfect yield, the embodied energy of liquid content will only reduce overall energy use if the additional
metal is greater than the current liquid metal embodied energy as scrap collected is diverted from being lost, for example, diverted
it has a lower average scrap content. However, as less liquid metal from landfill, as this will displace primary metal production. As
is produced and processed, the total energy use, and therefore the the recovery rate of process scrap is already very high (close to
embodied energy of the final products, is lower in the scenario with 100%), energy reduction through increased recycled content is only
perfect yield. possible through increasing the recovery rate of post-consumer
Table 6 shows the key results from the global analysis. The aver- scrap.
age forming CO2 emissions are 0.29 t CO2 /t, assuming the average Using the average embodied energy of liquid metal, or even
final forming energy is 2.5 GJ/t, of which 1.25 GJ/t is from gas and of semi-finished products, as a proxy for the embodied energy of
1.25 GJ/t is from electricity consumption. The average fabrication final products may give misleading results when comparing the
CO2 emissions are 0.06 t CO2 /t, assuming fabrication processes use effects of two scenarios on the embodied energy of final products.
100% electricity. In Fig. 8, the average embodied energy of both the liquid metal
On a global scale, the overall process losses for aluminium and semi-finished products is greater in the scenario with perfect
(41%) are much greater than that for steel (26%). Allwood et al. yield, however the final total energy use is lower than in the current
(2011b) suggest this is due to mechanical operations, like scalping, scenario.
to remove surface oxide, and the requirement for a higher quality The analysis shows that strategies that reduce scrap content may
surface finish. If all process scrap could be eliminated, 26% less liq- still result in overall energy savings and that normalised values,
uid steel and 41% less liquid aluminium would be required to meet like embodied characteristics which are defined per tonne, may
final product demand in 2008. not show the full picture; an understanding of absolute values may
Surprisingly, the global percentage reduction in energy use from be required to correctly evaluate the relative abatement potential
eliminating process scrap is much smaller than the percentage of different scenarios.
reduction in liquid metal; reducing process scrap only displaces
secondary production, assuming 100% of the process scrap arising 4. Discussion
is collected and recycled within a year. As a result, despite the dra-
matic savings in liquid metal production shown in Table 6, total As liquid metal production is the most energy intensive process
energy use would be reduced by just 17% and 6% and total CO2 step along the supply chain, the steel and aluminium indus-
emissions would be reduced by 16% and 7%, for steel and aluminium tries have been highly motivated to reduce energy consumption,
respectively. As the energy for producing liquid metal from ore is in order to reduce operating costs and environmental impacts.
so much greater than the energy for producing liquid metal from As a result, it is a highly efficient process and there is lim-
scrap, reducing energy use from secondary production has a less ited scope for further reduction in energy intensity. To achieve
significant effect. The difference between primary and secondary significant energy and CO2 reductions in the steel and alu-
energy is more marked for aluminium than for steel, therefore minium sectors, strategies that reduce demand for liquid metal are
eliminating process scrap has only a small impact on total energy required.
use. However, if the future production mix were to have an Liquid metal production is driven by demand for metal prod-
increased proportion of scrap due to stabilising demand and greater ucts. The volume of liquid metal required to meet this demand
end-of-life scrap arisings, the abatement effect of yield improve- depends both on the amount of metal in the final product and
ment would be more significant. the yield of the forming and fabrication processes. The results in
Calculated values for total energy use compare well with val- Section 3 demonstrate that the yield losses during the fabrication
ues from the literature: in 2004, the IEA reported energy use in process significantly increase the embodied energy of finished com-
the steel and aluminium sectors to be 21 EJ and 5.3 EJ respectively ponents compared to the liquid metal. Improving process yields
(2007). Scaling 2004 to 2008 (steel) and 2007 (aluminium) levels of could significantly reduce embodied energy and overall energy con-
production from IISI and IAI data (World Steel Association, 2010a; sumption by reducing liquid metal demand. Lightweighting and
International Aluminium Institute, 2006, 2008) gives a projected life extension are two other potentially significant strategies for
total energy use of 27 EJ and 6.5 EJ for steel and aluminium, which reducing liquid metal demand that should be investigated along-
is within 10% of the reported values. side process yield improvements: lightweighting reduces the liquid
The results in Fig. 8 illustrate how increasing the recycled con- metal demand per product and life extension reduces the demand
tent of a product may not always reduce its embodied energy. for products.
If all process scrap was eliminated, the average recycled con- At a global scale, eliminating process scrap in the steel indus-
tent of liquid metal would decrease and the embodied energy try would save up to 340 Mt CO2 , the impact of this strategy is
would increase to a level higher than today. However, as the equal in magnitude to the CO2 emission reduction thought possi-
absolute volume of liquid metal required would be reduced, the ble if best available technology could be implemented worldwide,
total energy used would also be lower. An increase in recycled based on 2005 levels of production (International Energy Agency,
1194 R.L. Milford et al. / Resources, Conservation and Recycling 55 (2011) 1185–1195

2008). Eliminating process scrap in the aluminium industry also and commercial barriers to the comprehensive collection of yield
has a similar impact to worldwide implementation of best avail- data need to be overcome. Senior management should recognise
able technology: eliminating scrap reduces emissions by 20 Mt CO2 , that yield losses are not necessarily a measure of inefficiency,
whilst best available technology reduces emissions by 12 Mt CO2 but that an accurate knowledge of yield is required in order to
(International Energy Agency, 2008). evaluate and improve the total yield loss along the whole supply
For the case study products, three processes emerge as par- chain.
ticularly significant for yield losses. Machining, as used in the The reduction of process scrap in the steel and aluminium
production of the wing skin, has the highest yield losses. The stamp- industries would have long-term structural implications: capac-
ing process used to produce the car door panel and the blanking ity requirements for secondary production may be reduced and
process used in both the car door panel and the beverage can also the post-consumer scrap stream would become more important.
have yield losses of between 10 and 32%. The practical limits to pro- Although challenging, structural changes like these may be neces-
cess scrap reduction have not been explored in this analysis, but sary in order to meet long-term emissions targets.
three strategies exist for reducing yield loss: develop new manu-
facturing processes with higher yields; operate existing processes 5. Conclusions
more effectively; design components with geometries more similar
to those of semi-finished products. This analysis has shown that reducing process scrap through
New manufacturing processes. Near-net shape casting processes yield improvements may offer a significant carbon abatement
are currently being developed, which produce an intermediate opportunity as yield is one of the factors that drives demand
product much closer in geometry to the final component, therefore for energy-intensive liquid metal production. Eliminating process
reducing downstream yield losses. However, large-scale imple- scrap downstream from liquid metal production would reduce liq-
mentation of these processes is currently limited by a lack of uid metal production by 26% and 41% for steel and aluminium.
economies of scale and the challenge of producing the same This could save a maximum of 17% and 6% of total energy use
microstructure as the conventional processing route. Innovation and 16% and 7% of total CO2 emissions in the steel and aluminium
in the blanking process could also reduce yield loss by develop- sectors respectively. This reduction is approximately the same as
ing a cutting process where a clean cut around a blank could be that which could be achieved by worldwide implementation of all
achieved without the need for a sacrificial skeleton, allowing bet- known best available technologies for energy efficiency in the steel
ter component tessellation. Novel processes could be developed for and aluminium sectors.
stamping to eliminate the need to produce a blank larger than the The graphs of cumulative energy and CO2 against cumulative
final component in order that the tooling can grip the blank whilst yield both for the case study products and the global case illus-
forming. trate the significant impact yield has on determining the embodied
Efficient process operation. Companies continuously work to energy and CO2 of metal products. Graphs of this form can also be
improve yield in existing processes as yield improvements increase used to illustrate the impact of energy efficiency and lightweight-
utilisation and have a financial benefit, for example, the transition ing strategies, though yield improvements may be more easily
from ingot to continuous casting for steel and casting variable- achieved than energy efficiency gains in what are already extremely
width ingots in order to produce coils of customer-specific width. efficient processes.
However, these initiatives are often limited to within the bound- Whilst yield improvements are already being pursued by com-
aries of the site; material utilisation for blanking could be improved panies, motivated by cost savings, an integrated approach to
by tessellating a larger range of parts for different customers from measuring and improving yield along the whole supply chain is
the same coil. required to make significant gains in yield improvement and there-
Modified component design. Modifying the design of components fore energy and CO2 reductions.
to improve tessellation could reduce yield losses, however, this
strategy would only reduce liquid metal demand if the displaced Acknowledgements
scrap took the place of some of the original product mass. If the
total mass of the product and the scrap remains constant, liquid The authors would like to acknowledge the input of all of the
metal demand will be unchanged. companies that provided data and facilitated site visits. All of the
Structural and business factors in the current supply chain result authors are supported by a Leadership Fellowship provided by the
in yield loss, for example, liquid metal is used to make semi-finished UK Engineering and Physical Sciences Research Council (EPSRC)
products, from which final metal goods can be manufactured. By reference EP/G007217/1.
limiting the range of geometries for these intermediate stock prod-
ucts, manufacturing processes can utilise economies of scale. Stock Appendix A. Supplementary data
product geometries are also kept simple to allow for maximum
versatility. However, final metal goods are diverse and often com- Supplementary data associated with this article can be found, in
plex in their geometries and the conversion from stock products to the online version, at doi:10.1016/j.resconrec.2011.05.021.
final products necessitates the production of significant amounts
of scrap. References
The dispersed nature of the current supply chain means that the
energy and emissions implications of yield loss along the supply Allwood JM, Cullen JM, Milford RL. Options for achieving a 50% cut in
industrial carbon emissions by 2050. Environmental Science & Technol-
chain are not readily identifiable. Yield losses will also influence ogy 2010a;44(March (6)):1888–94, doi:10.1021/es902909k, ISSN 0013-936X.
the “metal footprint” of a country, for example, the metal foot- http://dx.doi.org/10.1021/es902909k.
print of the UK will be greater than its net metal imports, due Allwood JM, Cullen JM, Carruth MA, Milford RL, Patel ACH, Moynihan M, et al. Going
on a metal diet-using less liquid metal to deliver the same service in order to save
to external yield losses. In order to motivate change, a whole
energy and carbon. WellMet2050. [online] Cambridge: University of Cambridge;
supply chain approach to process scrap is required, where the 2011a, Available at http://www.lcmp.eng.cam.ac.uk/wellmet2/introduction,
production and recycling of scrap can be treated as an internal http://www.lcmp.eng.cam.ac.uk/wellmet2/%20introduction [accessed 27 July
flow. Before collaboration along the supply chain can take place, 2011].
Allwood JM, Cullen JM, Carruth MA, Milford RL, Patel ACH, Moynihan
each company must have a complete knowledge of their pro- M, et al. Going on a metal diet; 2011b, ISBN 978-0-903428-32-3,
cess yields, both planned and unplanned. Therefore, the technical http://www.lcmp.eng.cam.ac.uk/wellmet2/publications.
 R.L. Milford et al. / Resources, Conservation and Recycling 55 (2011) 1185–1195 1195

Allwood JM, Ashby MF, Gutowski TG, Worrell E. Material efficiency: a white International Aluminium Institute. Personal communication; 2010.
paper. Resources, Conservation and Recycling 2010b;55(December):362–81, International Energy Agency. Tracking industrial energy efficiency and CO2
doi:10.1016/j.resconrec.2010.11.002, ISSN 09213449. http://linkinghub. emissions: in support of the G8 plan of action, vol. 1. OECD Pub-
elsevier.com/retrieve/pii/S0921344910002405. lishing; June 2007. ISBN 9789264030169. doi:10.1787/9789264030404-en.
BCS Incorporated. U.S. energy requirements for aluminum production, his- http://www.oecd-ilibrary.org/energy/.
torical perspective, theoretical limits and current practices; 2007, International Energy Agency. Energy technology perspectives 2008: scenarios &
http://www1.eere.energy.gov/industry/aluminum/pdfs/al theoretical.pdf. strategies to 2050, in support of the G8 plan of action, vol. 3. OECD/IEA; 2008,
Bernstein L, Roy J, Delhotal KC, Harnisch J, Matsuhashi R, Price L, et al. Industry, http://wwwsoc.nii.ac.jp/gtsj/jgpp/v03n01tp01.pdf.
in climate change 2007: mitigation. Contribution of working group III to the International Iron and Steel Institute. Energy use in the steel industry. Brussels:
fourth assessment report of the Intergovernmental Panel on Climate Change. International Iron and Steel Institute; 1998, ISBN 2-930069-36-8.
Cambridge, United Kingdom/New York, NY, USA: Cambridge University Press; Kamalapurkar R, Date P. Minimizing wastage of sheet metal for economical
2007 [chapter 7]. manufacturing. Journal of Materials Processing Technology 2006;177(July
Corus Construction Centre. Recycling and re-use in steel, environmental design in (1–3)):81–3, doi:10.1016/j.jmatprotec.2006.03.179, ISSN 09240136.
steel; 2001, www.steelconstruction.org. http://linkinghub.elsevier.com/retrieve/pii/S0924013606003517.
Department for Environment Food and Rural Affairs and Department of Nye TJ. Stamping blank optimal layout and coil slitting widths for single
Energy and Climate Change. 2010 guidelines to DEFRA/DECC’s GHG con- and multiple blanks. Journal of Engineering Materials and Tech-
version factors for company reporting; 2010, http://www.defra.gov.uk/ nology 2001;123(4):482, doi:10.1115/1.1395020, ISSN 00944289.
environment/business/reporting/conversion-factors.htm. http://link.aip.org/link/JEMTA8/v123/i4/p482/s1&Agg=doi.
Hammond G, Jones C. Inventory of carbon & energy (ICE) version 1.6a; 2008, University of California Santa Barbara. Greenhouse gas material comparison
http://www.bath.ac.uk/mech-eng/sert/embodied/. model; 2010, http://www.worldautosteel.org/Projects/LCA-Study/2010-UCSB-
International Aluminium Institute. Aluminium for future generations sustainability model.aspx.
update 2006; 2006, http://www.world-aluminium.org/?pg=140. World Steel Association. Steel statistical yearbook 2009; 2010a,
International Aluminium Institute. Life cycle assessment of aluminium: inven- http://www.worldsteel.org/?action=stats search.
tory data for the primary aluminium industry year 2005 update; 2007, World Steel Association. LCI data for steel products; November 2010,
http://www.world-aluminium.org/?pg=140. http://www.worldsteel.org/index.php?action=lcaform.
International Aluminium Institute. Aluminium for future generations sus- Worrell E, Price L, Neelis M, Galitsky C, Zhou N. World best practice energy
tainability update 2008; 2008, http://www.world-aluminium.org/cache/ intensity values for selected industrial sectors, LBNL-62806, rev. 2; 2008,
fl0000286.pdf. http://escholarship.org/uc/item/77n9d4sp.pdf.