Journal of Cleaner Production 40 (2013) 118e128

Contents lists available at SciVerse ScienceDirect Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro

Using sustainability reporting to assess the environmental footprint of copper mining

S. Northey a, b, N. Haque a, , G. Mudd b a CSIRO Minerals Down Under Flagship, Bag 312, Clayton South, Vic 3168, Australia

b Environmental Engineering, Department of Civil Engineering, Monash University, Clayton, Vic 3168, Australia

articleinfo abstract

Article history:

Received 3 April 2012 Received in revised form 17 September 2012 Accepted 20 September 2012 Available online 3 October 2012

Keywords:

Copper Embodied energy Greenhouse gases Water Life cycle assessment

The energy, greenhouse gas (GHG) emissions and water intensity, or environmental footprints, of global primary copper production have been estimated. The primary data have been collected from the sustainability reports published by copper producing mines, operations and companies. The mines analysed in this paper are from Australia, Chile, Peru, Argentina, Laos, Papua New Guniea, South Africa, Turkey, Finland, the USA and Canada. The typical range of energy intensity was found to be 10e70 GJ/ t Cu, with an average of 22.2 GJ/t Cu. The range of GHG emissions was 1e9 t CO2-e/t Cu, with an average of 2.6 t CO2-e/t Cu. The large variation exists largely due to the form of copper produced, ore grade, sources of fuel and electrical energy, and to a lesser extent the reporting methods and procedures used by the companies. The water footprint averages 70.4 kL/t Cu but can range from several kilolitres to up to 350 kL/t Cu. Variation in water intensity is generally due to inconsistencies in reporting method, the geographical location of the mining operations, limited economies of scale of production site, and the climate type (i.e. arid regions in Australia and Chile or temperate to sub- arctic climates in Canada or Finland). It is recommended that company sustainability reports should clearly specify fuel use by type for vehicles, heat or electrical energy, sources of electricity and their mixes (including GHG emissions factors), and the boundaries of the operation for meaningful use in life cycle assessment (LCA). Sustainability reports should be published at regular intervals so that improvements towards more sustainable performance can be measured, and an LCA of mining activities can be developed for primary copper production more readily. The paper provides a valuable insight into the strong value of sustainability reporting for an industrial sector such as copper mining and how such data can be linked to LCA studies.

1. Introduction

Copper is an important metal in modern society due to its wide variety of uses, including electrical wiring, heat exchangers, piping and roof construction and increasingly consumer electronics. The uses of these products have environmental impacts that often go unnoticed by the consumer. If the full environmental conse- quences of these products are to be considered then the associated impacts need to be described quantitatively. Life cycle assessment (LCA) methodology, or 'cradle-to-grave' analysis, can be used to quantify the environmental impacts of a product or process. LCA seeks to examine all stages of a product's life cycle such as material production, manufacturing, distribution, usage, and

* Corresponding author.

Crown Copyright ! 2012 Published by Elsevier Ltd. All rights reserved.

finally ultimate disposal or recycling. This is achieved by creating an inventory of all the energy and material flows used in the product's life cycle. The interactions of these flows with the environment can be used to evaluate the potential impacts of the product manufacture. The results of an LCA can be used to improve the design of products so that they have less impact on the environment. LCA of primary mineral and metal production is often carried out in the context of a 'cradle-to-gate' assessment, where only impacts of the mining and metal production stages of the life cycle are considered. This type of assessment can then be built upon to produce larger, whole of life assessments of products.

Demand for copper has increased dramatically over the past one hundred years, driving production rates almost exponentially. South America is largely responsible for meeting this demand, with Chile alone contributing one third of global production. Annual production of copper was estimated to be 16.1 million tonnes in

0959-6526/$ e see front matter Crown Copyright ! 2012 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jclepro.2012.09.027

2011 (USGS, 2012). Copper resources that are cheaper to extract are mined first because it is more profitable for mining companies to do so, typically meaning higher grade ores are mined first. A key factor that affects the energy and water requirements of copper extraction is the ore grade e that is, the ratio of contained copper per unit of ore (commonly expressed as a percentage, e.g. %Cu). It is expected that ore grades will steadily decline as high grade ores are prefer- entially mined (Mudd, 2010; Mudd and Weng, in press). The decline in ore grades has large ramifications regarding the potential envi- ronmental impacts of copper production. As ore grades decline the amount of ore mined and processed per tonne of copper produced increases and thus the specific energy of copper (i.e. the energy required to produce one tonne of copper) also increases (Norgate and Rankin, 2000).

Copper ores are mined using underground and open cut methods. The copper ores mined can broadly be categorised as being either sulfide or oxide ores. The type of ore mined has significant impacts on the downstream processing routes. There are two main processing methods used to produce copper. Sulfide ores that contain iron are commonly processed using the pyrometal- lurgical route because the iron enables them to be concentrated quite easily while the exothermic reaction heat from the oxidation of iron and sulphur can be utilised in the smelting stages. However, oxide ores and sulfide ores that contain low levels of iron are better suited to hydrometallurgical processing (Ayres et al., 2003). Bene- ficiation processes such as ore sorting, grinding and separation by flotation techniques may be used for both processes but are less common in hydrometallurgy. Pyrometallurgical processing has four main stages: mining, concentrating, smelting and refining. Sulfide ores undergo flotation to separate the copper minerals from gangue materials. The concentrate is then smelted to produce copper matte, iron oxide rich slag and sulphur dioxide. During smelting of sulfide concentrates both iron and sulphur are oxidised and their reactions are both exothermic. The sulphur dioxide produced is typically captured following this stage as sulphuric acid. Blister copper of approximately 98% purity is produced by further oxida- tion of the matte in the converting stage (Schlesinger et al., 2011). This is then fire refined and cast to produce copper anode. Finally, the copper anode is electro-refined in a refinery to produce cathode copper.

Hydrometallurgical processing has four main stages: mining, leaching, solvent extraction and electrowinning. Beneficiation also occurs to varying degrees after mining depending on the type of leaching method used. Leaching is the process of reacting the ore with acid to mobilise the contained metals. A variety of leaching methods are used to extract copper, with heap leaching being the most common. There are various forms of the heap leaching process. If the ore grade is low there is a possibility of using in-situ leaching (Independent Mining Consultants, 2011). After leaching, the copper is recovered from the leachate through the process of solvent extraction. Solvent extraction produces a copper sulphate solution which is then used to produce cathode copper by elec- trowinning. The solvent extractioneelectrowinning (SXeEW) process is continuous and produces copper cathode with similar purities to the pyrometallurgical process.

Companies within the mining industry are becoming more transparent in releasing information in relation to their interactions with the environment and following various standards and proto- cols (Perez and Sanchez, 2009). This is due in part to the Global Reporting Initiative (GRI) which has seen increasing uptake by mining companies over the past decade (Mudd, 2009a). The GRI is a framework for the reporting of environmental, social and economic performance in the context of sustainability. Conse- quently, there is now a large body of publicly available data relating to the energy, water consumption and greenhouse gas (GHG)

emissions of copper production. However, there are some incon- sistencies and variations in the methods or boundary used by the companies. Thus the sense, purpose and the use of the reported numbers often make it difficult for the stakeholders to judge the performance of different companies and operations. The drivers for reporting performance of the sustainability by the companies are varied (Pellegrino and Lodhia, 2012) and often LCA is not the prime purpose. The main intention of the companies for placing an emphasis on sustainability is to maintain a social license to operate and to be compliant with the voluntary or mandatory requirements where imposed.

Although there are a number of companies now reporting sustainability indicators in greater detail, there is a lack of analysis of this information in the literature to provide a basis for improving performance. The purpose of this paper is to report energy, GHG and water footprint of copper production based on the review of sustainability information reported by the companies relating to primary copper production. Operational data relating to the energy and water consumption, and GHG emissions of copper mines have been compiled from publicly available sustainability and financial reports. These reports have been critically analysed for variations, trends and judged for their usability for LCA. The effects of ore grades and processing routes on water, energy and GHG intensities have been analysed to determine the key issues and relationships of copper production globally. Generally companies only report scope 1 and 2 GHG emissions data, and so direct comparison with previous LCA work, which includes scope 3, is difficult. Importantly, the widespread variability between copper operations is identified and highlights the limitations of applying conventional LCA models to the industry as a whole. In the context of this study, scope 1 covers the emissions from the fuels burnt on-site within the boundary of production. Scope 2 generally considers purchased electricity from an off-site grid. Scope 3 is all the remaining energy associated with the emission generated by third party material suppliers, travel of the employees, etc. that are generally outside the boundary of the production site.

2. LCA of copper

LCA work on copper production has been undertaken previously by Australia's CSIRO and other research organisations (Norgate et al., 2007; Norgate and Jahanshahi, 2010; Alvarado et al., 2002; Kuckshinridis et al., 2007). CSIRO carried out a series of LCAs on primary metal production processes in the early 2000's. Norgate et al. (2007) amalgamated the findings of these studies and several others to provide an overview of the potential environ- mental impacts of metal production and the factors that affect these impacts. This analysis found that for a 3% copper sulfide ore processed via pyrometallurgical processes, the specific energy consumption was approximately 33 MJ/kg Cu, resulting in a global warming potential (GWP) of 3.3 kg CO2-e/kg Cu. For a 2% copper sulfide ore processed via a hydrometallurgical process, the required energy input was 64 MJ/kg Cu, with an associated GWP of 6.2 kg CO2-e/kg.

The emission of GHGs as a result of copper production is an important issue that must be considered. The mining and benefi- ciation stages of pyrometallurgical processing produces a larger relative proportion of GHG emissions than in hydrometallurgical processing (Norgate et al., 2007). Declining ore grade mostly impact on the energy required for the mining and beneficiation stages of copper production. Once a concentrate or mineral product of a specified grade is produced, the downstream processing is not significantly affected by the original ore grade. Therefore the effects of declining ore grade will have a larger impact on pyrometallur- gical copper production.

S. Northey et al. / Journal of Cleaner Production 40 (2013) 118e128 119

120 S. Northey et al. / Journal of Cleaner Production 40 (2013) 118e128

A considerable amount of operational data has been made available recently via companies adopting the GRI (Mudd, 2009b). A 'bottom-up' statistical analysis of these data provides an approach to quantifying the impacts of copper production, albeit with a reduced scope of analysis when compared with typical LCA. Importantly, this enables the variability of performance between sites to be quantified and provides an insight into the current state of sustainability reporting.

3. Methodology

Data for this paper were sourced from numerous companies' sustainability and financial reports over several years, as shown in Table 1. The material, energy and water consumption input data from these reports have been used to construct input data equiv- alent to life cycle inventory data tables. While financial reports typically report production data consistently from year to year, the quality of data from sustainability reports, however, varied signif- icantly between companies and individual operations. Generally, companies that had adopted some form of GRI were more consis- tent in their reporting from year to year. Companies that chose to implement an in-house sustainability reporting framework were sometimes inconsistent with what indicators they report in any given year. Another key issue was companies revising reported data of previous years in a table or graph without explaining the reasons behind the change. In these cases, it can be difficult to ascertain whether they have changed their reporting methodology or if the company is simply correcting a calculation error. Despite these inconsistencies, reported data were used as is to gain an insight in the production of copper.

The essential steps were collection of data from the sustain- ability and financial reports, tabulation for each mine and production site depending upon the process configuration, analysis of energy data by source, addition of sources of electricity collected from open sources, analysis of water data, and finally determina- tion of the trend between the specific footprint with ore grade. The results have been assessed for the causes of variation and identi- fication of any potential improvement measures for reducing

environmental impact and to recommend what data in sustain- ability reports are potentially suitable or required for LCA.

A range of copper operations from around the world have been included in this analysis. The inclusion of particular operations was dependent on the availability of public reports and site data. Re- ported energy usage, GHG emissions, water consumption and production data were compiled for each operation where available. For most operations, production data were only compiled for years in which there were sustainability data reported. Where incom- plete or partial information was provided in reports, other data points were calculated where possible (i.e., the calculation of indirect energy consumption as total energy consumption minus direct energy consumption).

All energy, water, and GHG data presented in this paper have been weighted to reflect the proportion of copper produced from that operation relative to other products, in accordance with the ISO 14044 LCA standard (ISO, 2006). The allocation was done using annual average prices, primarily sourced from the USGS's Minerals Commodities Summary for the period 1991e2008 (USGS, 2009) or the London Metals Exchange from 2009 to 2010 (LME, 2011). Non- metallic by-products of copper production such as sulphuric acid were not included in the economic weighting of the data.

The boundaries of reported data vary from company to company depending on their reporting methodology. Therefore it was diffi- cult to apply strict boundaries in the analysis of the data. For the purpose of analysis, each mine site has been categorised based on the major mining and extraction methods used. These are under- ground and/or open pit mining, concentration, smelting, refining, and leaching, solvent extraction and electrowinning (LSE). Trend lines have been plotted where appropriate to aid visual analysis and each data point has an equal weighting in producing the trend line. However each data point actually represents a different amount of copper production.

For LCA, the general methodology is to collect and estimate life cycle inventory data for each stages of operation at the individual production site. In contrast, this study's methodology is a more useful approach to analyse a large number of production site based reports to find the environmental footprints in real terms. Using this methodology, the environmental footprints of copper mining can be determined from the gross operational data reported by the companies. Although there are limitations in the data quality, this methodology is relatively quick compared with conventional LCA. There are also problems with data quality and assessment in conventional LCA databases (e.g. variable local GHG emissions intensity factors), although the data uncertainty can be analysed to some extent using models.

4. Results

4.1. Production

Production data are the most common data reported by mining companies. Commonly reported production statistics include mass of ore milled, ore mined, saleable product, and ore grade. The summary of production by company is shown in Table 2. Fig. 1 shows the relationship between the average annual copper production and ore grade over the years considered for each operation. This figure indicates that the majority of recent production occurred from ore grades in the range 0.5e1.5% Cu. This agrees with prior analysis by Mudd (2009b) that found current global average ore grades are less than 1% and in gradual decline. These results also confirm that the cut-off for profitable ore grades is around 0.5% Cu (Schlesinger et al., 2011). Crowson (2012) provides a more detailed analysis of the historical trends in copper ore grades which shows clear decline from 1970 to 2009.

Table 1

List of financial Report type

Financial Sustainability Financial Sustainability Financial Sustainability Financial Sustainability Financial Sustainability Financial Sustainability Financial Sustainability Financial Sustainability Financial Sustainability Financial Sustainability Financial Sustainability Sustainability Financial Sustainability

reports and sustainability reports reviewed and analysed.

Year

2002e2010 2002e2009 2000e2010 2001e2010 1999e2010 2000e2010 2005e2010 2005e2009 2009e2010 2009e2010 2003e2010 2004e2010 2000e2010 2002e2007 2004e2008 2004e2007 2008e2010 2008e2010 2000e2010 2000e2010 2000e2010 2001e2009 1996e2004 2002e2010 2002e2010

Companies

Anglo American Anglo American BHP Billiton BHP Billiton Codelco Codelco

Inmet Mining Inmet Mining Minerals and Metals Group Minerals and Metals Group Newcrest Mining Newcrest Mining Ok Tedi Mining Ok Tedi Mining Oxiana Oxiana Oz Minerals Oz Minerals Rio Tinto Rio Tinto Teck Resources Teck Resources WMC Resources Xstrata Xstrata

S. Northey et al. / Journal of Cleaner Production 40 (2013) 118e128 121

Table 2

Average operational data and weighted energy, GHG and water intensities.

Operation

Operating Deposit company type

Metals extracted

Mine Process type

kt ore/yr

%Cu t Cu/year

Electricity source

Direct GJ/t Cu

Indirect Total GJ/t Cu GJ/t Cu

Total t CO2e/t Cu

kL/t Cu

Weighting Period Cu%

Australia

Cadia-Ridgeway Ernest Henry Golden Grove Mount Isa Northparkes Olympic Dam Prominent Hill Rosebery

Newcrest Porphyry Xstrata IOCG MMG VMS Xstrata SedEx Rio Tinto80% Porphyry BHP Billiton IOCG

AueCu CueAu AgePbeZneCu Cu CueAu CueAueAgeU CueAueAg AgePbeZneCueAu AueCu

O/U MC OP MC UG MC UG MCS O/U MC UG MCSRL OP MC UG MC O/U MC

22,317 10,232 1530 5787 5290 7841 8038 760 17,903

12.4(7) 16.5(4) 4.0(4) 10.1(4) 7.3(3) 11.1(6) 9.0(2)

12.7(4) 21.2(6) 19.5(3) 19.4(5) 14.8(7) 27.2(19)

100 2002e2010 85.7 2003e2010 95.3 1991e2010 80.9 2009e2010 10.5 2009e2010 29.4 2004/05e2009/10

Telfer

OZ Minerals IOCG MMG VMS Newcrest OGC

8.7(2) 17.7(2) 12.8(2) 18.8(2) 19.8(6) 40.6(6)

1046.9(1) 161.1(6)

Argentina

Alumbrera

Xstrata50% Porphyry

CueAu CueMo

OP MCL

36,127

0.55 174,078

GAS, Hydro 23.9(4)

13.4(5) 37.3(4)

91.3(6)

68.0 2002e2010 82.3 2004e2007

Canada

Highland Valley Kidd Creek Chile Andina

Teck Porphyry Xstrata VMS

CueZn

UG MC OP MCSR

48,882 2451

0.40 164,827 2.07 87,009

Hydro 6.6(4) Hydro/Nuclear

17.1(4) 23.7(4) 20.0(2)

0.9(4) 2.3(2)

135.4(2) 76.7(2)

56.2 2006e2010 91.3 2001e2010

Codelco Norte Collahuasi El Soldado El Teniente

CueMo CueMoeAueAg Cu CueMo

OP MCSL OP MCL O/U MCL UG MCS

126,120 44,422 7995 44,053

0.83 888,618 SING 10.8(9) 9.4(10) 1.12 479,263 SING 5.9(4) 9.1(4) 1.00 62,626 SIC 1.00 409,692 SIC 5.2(9) 14.1(10)

17.1(9) 15.0(4) 27.9(7) 19.3(9)

3.1(9) 2.5(4) 2.0(7) 1.4(9)

53.3(8) 31.9(2) 48.2(7)

87.4 1999e2010

Escondida Lomas Bayas Los Bronces

Porphyry BHP57.5% Porphyry Xstrata Porphyry

CueAueAg Cu CueMo

OP MCL OP MH OP MCL

116,425 41,834 20,469

1.25 1,242,644 SING 4.4(6) 7.5(6)

11.9(6) 26.9(5) 16.9(7)

1.7(6) 1.1(7)

52.5(6) 75.5(3) 80.7(7)

97.6 2003e2008 100 2006e2010 90.7 2003e2009

Mantos Blancos Mantoverde Salvador Quebrada

Anglo American IOCG Anglo American IOCG Codelco Porphyry Teck76% Porphyry

Cu Cu CueMo CueZn

OP MCL OP ML O/U MCSL OP ML

14,924 14,418 16,704 17,626

0.69 89,957 Diesel, SING 0.56 61,093 SIC 0.59 74,379 SIC 0.86 85,000 Diesel, SING

20.2(9) 23.4(10) 51.1(1) 0.6(1)

47.2(7) 22.8(7) 41.5(9) 51.6(1)

3.1(7) 2.3(7) 3.5(9) 3.9(1)

226.5(7) 46.6(7) 321.3(8) 21.9(1)

100 2003e2009 100 2003e2009 89.2 1999e2010

Laos

Blanca

94.3 2008 51.8 2005e2010

Finland

Pyhasalmi

Inmet Mining VMS

ZneCu CueAu

UG MC OP ML

1374 3646

1.08 16,650 1.99 53,370 Hydro

4.5(7) 10.4(7) 12.6(4) 13.7(4)

14.9(6) 26.4(4)

0.4(5) 2.8(4)

211.0(7) 34.0(5)

Sepon

MMG Hybrid VMS-SedEx

79.0 2005e2010 70.8 2000e2010

South Africa

Palabora

Palabora Carbonatitic Mining Company

Cu

UG MCSR

16,932

0.61 77,656 Coal

34.5(6) 20.4(6)

55.5(10)

8.5(10)

94.4(11)

Turkey

Cayeli

Inmet Mining VMS Xstrata Skarn OTML Porphyry Rio Tinto Porphyry

CueZn CueAu CueAueAg CueAueAgeMo

UG MC OP MCL OP MC OP MCSR

967 6944 24,945 48,676

1.5(7) 5.3(7) 21.6(1) 9.1(5) 0.83 176,460 Hydro 6.8(6) 8.1(4)

6.7(7) 31.4(1) 15.8(5) 47.0(8)

1.0(7)

87.3(7) 42.5(5) 38.8(7)

65.7 2003e2009 95.7 2006e2010 69.6 2003e2010 58.0 2003e2010

Peru

3.78 30,229 1.34 103,179 Hydro

Tintaya

PNG

Ok Tedi

1.2(6) 4.3(8) 9.8(7)

USA

Kennecott Utah

0.59 242,122 Coal

Codelco Breccia- Porphyry Codelco Porphyry Xstrata-Anglo Porphyry

CueMo

O/U MC

23,446

1.05 226,932

SIC 3.1(9)

8.7(10) 11.8 (9)

1.0(9)

99.4(7)

Anglo American IOCG Codelco Breccia-

139.7(7)

95.0 2006e2010 100 2003e2009 92.4 1999e2010

Anglo American Breccia- Porphyry

0.33 64,228 0.98 95,130 1.85 19,446 3.21 222,261 0.78 40,678 2.46 136,827 1.54 104,241 0.36 1985 0.24 30,519

Coal Gas Hydro Coal, gas Coal Coal, wind Coal, wind Hydro 6.1(2) Gas 20.9(6)

16.0(7) 28.4(7) 8.0(4) 23.2(6) 5.9(3) 10.1(3)

5.2(7) 2.3(6) 1.7(4) 2.1(6) 4.1(5) 5.0(16) 2.2(2) 1.2(2) 4.7(6)

49.0(7) 42.3(7) 33.0(3) 19.5(7) 74.0(8) 46.6(14) 39.8(2)

43.0 2003/04e2009/10 84.8 2002e2010 35.9 2005e2010

0.31 65,937 SING 1.06 228,300 SIC

14.3(3) 12.6(3)

Abbreviations: SedEx e Sedimentary Exhalative, VMS e Volcanic Massive Sulfide, IOCG e Iron Oxide Interconnected System), SIC e Sistema Inteconnectado del Central (Chilean Central Interconnected System). Process Key: M e Mine, C e Concentrator, S e Smelter, R e Refinery, L e Leaching & SXeEW. The number of years that data were available is shown in brackets; individual weightings were used for each year. Sources: (AA, Various; BHPB, Various; Codelco, Various; Inmet Mining Corporation, Various; MMG, Various; Newcrest Mining Limited, Various; OTML, Various; Oxiana Limited, Various; OZM, Various; Rio Tinto, Various; Teck Resources Limited, Various; WMC, Various; Xstrata, Various).

Copper Gold, OGC e Orogenic gold-copper,

SING e Sistema Interconnectado del Norte Grande (Chilean Norte Grande

122

S. Northey et al. / Journal of Cleaner Production 40 (2013) 118e128

Cu

Cu-Au Cu-Mo Cu-Zn Cu-Au-AgU Cu-Mo-Au-Ag Cu-Zn-Pb-Ag

1,200 1,000 800 600 400 200 0

Fig. 1. Average copper production as a function of average ore grade.

4.2. Energy

There was variation both between and within companies relating to how energy consumption was reported. The GRI guidelines recommend reporting direct and indirect energy consumption data by primary energy source (GRI, 2011). Direct energy accounts for fuels burnt on-site while indirect energy accounts for energy (typically electricity) used on-site or purchased from national grid, but generally produced off-site. Reporting in this way was popular across operations regardless of whether they reported to GRI standards. However, for some operations only total energy consumption was reported. Where possible reported energy consumption totals were cross-checked against stated materials consumption information. In several instances it was determined that either incorrect units or values were stated and consequently these numbers were revised for this analysis. For instance in 2010 Xstrata's Southern Peru Division consistently reported energy consumption in terms of kilowatts, kW, a unit of power (Xstrata, 2010, Southern Peru Division Sustainability Report, pp. 63e66). However for our analysis we have assumed that the correct unit was in fact kilowatt-hours, kWh, a unit of electrical energy consumed. It is not uncommon for mining operations to produce a proportion of their electricity from fuel burnt on-site. Under the GRI guidelines (GRI, 2011) electricity produced on site should be classified as direct energy. However for companies using their own reporting scheme it is not usually stated if the electricity consumed was purchased from the grid or if it includes electricity generated on-site. It was assumed that all reported electricity consumption be analysed as though it is indirect energy consumption expressed in terms of primary energy. Therefore the values for indirect energy consumption stated in this paper provide merely a good approximation of electricity consumption given that mines typically purchase the vast majority of their electricity from off-site generators. Although the actual magnitude of specific energy consumption is debatable based on this assumption, the relationship between the energy intensity and ore grade is still valid.

Even though about 50% of the mining companies of 100 Australian Stock Exchange listed companies reported on material, energy and water as indicators of sustainability accounting reports (Yonvanich and Guthries, 2006), the actual data are not very useful for LCA because of lack of detail information and clarification to improve environmental performance. The companies should separate their energy consumption by type as either thermal heat or electrical energy, diesel input for machineries in detail for each

machine type, source of electricity as produced on-site and imported from an off-site grid, the mix of fuel or sources of elec- tricity in tabulated form for improvement of sustainability report- ing and also to find specific opportunities for improvement measures to be implemented. The evolution of sustainability reporting processes for various mining companies have been ana- lysed by Jenkins and Yakovleva (2006). They have identified leaders and laggards of the industry based on the reporting standards. As a measure for improvement, the laggards can learn from the top quality reporting companies. Peck and Sinding (2003) reported the wide variation in sustainability reporting practices and sincerity of the reporting can have consequences for improvement of sustain- ability of the industry.

The energy intensity (Gigajoules per tonne of copper; GJ/t Cu) of mine sites is plotted against ore grade in Fig. 2. Each data point represents one year of production from a mine site for which energy data were available. Energy intensities were weighted to represent the proportion of revenue produced by the copper product and were grouped based on the major processes present on site.

Many factors contribute to variability between sites for the production of copper. Generally, underground mines are more energy intensive than open pit mines (per tonne of rock or ore mined). This is due to a variety of factors such as mine depth, smaller ore throughputs and the need for support services such as ventilation (Brake, 2002). Ventilation typically contributes 30e40% of the total operating costs of underground mining due to the energy requirements of this task.

For copper operations that only have a mine and concentrator, there is a strong correlation between energy intensity and ore grade (0.71 correlation coefficient). For these operations, consumption of diesel in the mine site increases with declining ore grade due to the need to process more ore per unit of copper. The energy requirements for the concentration stage also increase, primarily due to the need to crush and mill more ore. The config- uration of the mill and flotation circuit depends on the mineralogy of the ore and site specific impurities. For example, ore at New- crest's Telfer AueCu mine contains significant amounts of pyrite, so the flotation circuit is more complex in order to separate the pyrite before gold is recovered via carbon-in-leach processes (Zheng et al., 2010). This added complexity increases the energy requirements of flotation for this operation.

Large variability in the data exists independent of ore grade for sites which include a smelter. Factors which contribute to these are differences in smelting technologies, the minerals present in the feed concentrate, and the proportion of this feed concentrate which is imported into the site from other copper operations. Of the operations considered, two use Outokumpu flash furnaces, three use Teniente furnaces, one uses Mitsubishi furnaces and one uses the Isasmelt process (Kapusta, 2004). The final product varies between sites with copper matte, blister copper and fire-refined copper anode being produced from different smelting operations. Smelters and converters require energy from fossil fuels to main- tain optimal operating temperatures. Fuel sources for these purposes can include fuel oil, natural gas and coke of various compositions (Schlesinger et al., 2011). Reactions inside sulfide smelting furnaces are typically exothermic and the amount of energy released depends on the composition of the concentrate being smelted. Consequently, large variations exist between sites, not only in the types of smelting and converting furnaces used but also in the fuel input required to maintain operating temperature. Anglo-American reports sustainability data for Chagres smelter in Chile, which operates an Outokumpu process (USGS, 2004). Over the period 2003e2010 the energy intensity of Chagres smelter averaged 6.7 GJ/t Cu or 1.9 GJ/t of concentrate.

01234 Average Ore Grade (%Cu)

Average Production (kilotonne Cu)

S. Northey et al. / Journal of Cleaner Production 40 (2013) 118e128 123

70 60 50 40 30 20 10

0 0 0.5 1 1.5 2 2.5 3 3.5 4

Ore Grade (% Cu)

Fig. 2. Energy intensity as a function of ore grade for 31 copper operations, with each data point representing a year of production.

y = 36.529x-0.351 R2 = 0.40

Mine + Leaching, SX-EW (LSE) Mine + Concentrator Mine + Conc. + LSE Mine + Conc. + Smelter

Mine + Conc. + Smelter + Refinery LSE Mine + Concentrator Mine + Conc. + Smelter + Refinery LSE

y = 15.697x-0.573 R2 = 0.71

There is a low correlation (0.40 correlation coefficient) between energy intensity and ore grade when considering operations that include both a smelter and a refinery with or without hydromet- allurgical processing. In these cases, the variability is increased due to the varying contributions of pyrometallurgical and hydromet- allurgical processes. The inclusion of a refinery within an operation is a significant investment for a company, so very high smelter throughputs are required to justify it. Accordingly, pyrometallur- gical processes typically are the dominant processing route for these operations. However it is uncommon for copper mines to have an on-site refinery, so there are only a limited number of data points available for this category of operation. The energy intensity of refined copper cathode produced from some of these sites is likely to be higher than the industry average due to the nature of the operations included. Olympic Dam produces electro-refined copper cathode but operates an unusual process of leaching its tailings and concentrate prior to smelting, due to the uranium present in the ore (Woodcock and Hamilton, 1993). At Palabora, magnetite concentrate and vermiculite are produced in addition to copper. The site produces not only copper cathode but also casts a proportion of this to produce copper rods (Rio Tinto, Various). Variability is present in the energy requirements for electro- refining between sites and can range from 285 to 390 kWh/t Cu depending on operation (Schlesinger et al., 2011). Energy consumption for Xstrata Copper's Townsville refinery averaged 2.5 GJ/t Cu from 2005 to 2010. Electricity represented 60% of this with the remainder being primarily LPG (Xstrata, Various).

The energy requirements for hydrometallurgical processes are more difficult to determine, as the process is less modular when compared to pyrometallurgical processing. Electricity consumption for the SXeEW process at Lomas Bayas was reported by Xstrata

over the period from 2008 to 2010 (Xstrata, Various). The contri- bution of the electro-winning process to cathode production averaged 1968 kWh/t Cu over this period. Energy consumed by electrowinning processes at other sites range from 1800 to 2000 kWh/t Cu (Schlesinger et al., 2011).

Norgate and Rankin (2000) examined the production of copper cathode from a 3% Cu sulfide ore via pyrometallurgical processing. Calculation based on the trend-line equation for mining, concen- trating, smelting and refining shown in Fig. 5 results in an energy intensity of 24.8 GJ/t Cu for a 3%Cu ore. This is lower than the 33 GJ/ t Cu obtained in the Norgate and Rankin (2000) study, as expected due to the reduced scope of company reporting.

To gain a more accurate assessment of the average energy intensity of copper cathode produced across the industry, it is necessary to consider the full chain of operations rather than individual operations by themselves. Many companies report aggregate or group data that represents the combined data of all their copper operations. Average aggregate data for Anglo- American Chile and Codelco are shown in Table 3. The combined copper production for both companies was approximately 12e13% of global production with average annual copper grades varying between 0.8 and 1.0%. The average energy intensity of 22.5 GJ/t Cu is much lower than for copper operations that include an on-site refinery. Therefore greater efficiencies appear to be possible by having centralised refinery and smelting operations.

The energy intensity can change significantly over the life of individual operations as shown in Fig. 3. El Soldado has seen a constant increase in energy intensity due to a decline in ore grades in both its' underground and open pit mines. Although Olympic Dam has seen a gradual decline in ore grades the energy intensity appears more dependent on other factors at the site.

Fig. 3. Energy intensity and ore grade over time for several operations.

Energy Intensity (GJ/t Cu)

124

S. Northey et al. / Journal of Cleaner Production 40 (2013) 118e128

Mine + Leaching, SX-EW (LSE) Mine + Concentrator Mine + Conc. + LSE Mine + Conc. + Smelter

Mine + Conc. + Smelter + Refinery LSE Mine + Concentrator

y = 1.5548x-0.606 R2 = 0.28

10 8 6 4 2 0

Fig. 4. GHG intensity as a function of ore grade for 28 copper operations, with each data point representing a year of production.

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Ore Grade (% Cu)

There was a major expansion of the mine between 1997 and 1999 which accounts for the large peak in energy intensity at this time. The smaller peak in 2003 is the result of a fire in the SXeEW plant in 2001 and maintenance in the smelter which resulted in 20% less throughput. Palabora saw a major spike in energy intensity from 2001 to 2004 as the mine transitioned from being an open pit mine to an underground mine. The sharp decline in 2006 is due to a decrease in the percentage of copper for the sites revenue for that year, resulting in more energy being allocated to the other products from the site. The sharp increase in energy intensity at Salvador during 2007 and 2008 is the result of industrial action at the site reducing mine production.

Below ore grades of approximately 0.5% Cu, fuel consumption is higher than electricity consumption. However, above this grade, electricity consumption becomes the dominant source of energy consumption. Ore grade primarily influences energy consumption in the mining and beneficiation stages of copper production. Once a concentrate has been produced the energy requirements of the smelting and refining stages are dependent on the composition of the concentrate but not on the original ore grade. The various minerals and impurities present in the concentrate determine the most suitable type of smelter and refining configuration. Diesel for

7 6 5 4 3 2 1 0

0 2 4 6 8 10 kWh/kg Cu

Fig. 5. Indirect energy intensity as a function of scope 2 GHG emissions intensity for 19 copper producing operations.

most operations is used primarily to operate heavy machinery in the mine. Consequently copper specific diesel consumption is closely correlated with ore grade and can range from 3 GJ/t Cu for high grade ores to 40 GJ/t Cu for low grade ores.

Beneficiation typically places the largest electricity demand on copper operations. As an example, the concentrator at Palabora accounts for 60% of the total electricity consumption of this site (Rio Tinto, 2003). Lower ore grades increase the energy requirements for beneficiation as more ore has to be treated to produce the same amount of copper concentrate of a particular grade. Even for a particular ore grade, large variability exists in the energy consumed at this stage of copper production. Communition and flotation circuits vary between mine sites due to different ore mineralogy and hardness. Differences such as the effect of grind size on the energy intensity are amplified as ore grade declines (Norgate and Jahanshahi, 2010). Therefore improvements in grinding efficiency and optimisation of grind size have potential for significant energy reductions in the future. However, offsetting this is the possibility that future ore deposits may be finer-grained and will require grinding to finer sizes.

The primary source of electricity for copper mines varies between regions. In Chile, electricity to mine sites is supplied primarily by two different electrical networks (Central Energia, 2012), the 'Sistema Interconnectado del Norte Grande' (SING) and the 'Sistema Interconnectado del Central' (SIC). Approximately 73% of the SING network was used to supply electricity to copper mines in 2008 (Pimentel, 2009a). This network supplies electricity to six of the mines considered in this report. Installed capacity on the SING is largely composed of coal and natural gas in equal proportions. However, over time, the mix of energy production has ranged from being natural gas dominated to being coal and diesel dominated. In contrast, installed capacity on the SIC network, that supplies six of the mines considered in this report, is comprised of hydropower (46%) with the remainder coming from natural gas (24%), oil (15%), coal (11%), biomass (3%) and wind (1%) (Central Energia, 2012).

y = 0.802x R2 = 0.70

Australia Canada Chile - SIC Chile - SING Finland

Laos

South Africa Turkey Australia Chile - SIC Chile - SING

y = 0.6529x R2 = 0.59

y = 0.2923x R2 = 0.75

Table 3

Group energy data. Company/Group

Anglo-American Chile

Codelco

t Cu/year

441,916 1,642,636

Fuel Electricity GJ/t Cu GJ/t Cu

10.4 13.0 8.5 13.8

Total GJ/t Cu

23.4 22.3

Period

2003e2009 2003e2008

Sources: (AA, Various; Codelco, Various).

kg CO2-e/kg Cu

GHG Intensity (t CO2-e/t Cu)

4.3. GHG

GHG data was collected in terms of scope 1, scope 2, scope 3 and total GHG emissions. Calculated total GHG emissions are the sum of scope 1 and scope 2, plus scope 3 where available. It should be noted that scope 3 emissions are rarely reported by mining companies.

GHG emissions during the production of copper are largely a result of the use of fossil fuel energy. Therefore the major trends present for energy consumption are present also for GHG emis- sions, as shown in Fig. 4. As with energy, a decline in ore grade generally results in higher GHG emissions intensity. However GHG emissions display much more variability with respect to ore grade as shown by the low R2 value. Additional processing stages at a site further add to this variability.

The different methodologies used to report GHG emissions give rise to significant variation, both over time and between sites. For Olympic Dam's most recent expansion environmental impact statement (see BHPB, 2009), future GHG emissions were estimated using Australia's National Greenhouse Emission Reporting (NGER) Act (2007) method, as well as BHP Billiton's own estimation method that included emissions not covered by the Act. These include the release of carbon dioxide from the leaching of carbonate minerals with sulphuric acid (BHPB, 2009). The calcu- lated GHG emissions for the initial stages of expansion were 26% higher using BHP Billiton's internal method compared to using the NGER Act's method. Note that these projected emissions scenarios are not included in Fig. 4. Another example of this variability can be seen at Mt. Isa which previously reported under the Australian Government's Greenhouse Challenge Plus and Energy Efficiency Programs but commenced reporting under the NGER Act in July 2009 (Xstrata, 2009). The 2010 Mt. Isa Mines and Xstrata North Queensland sustainability reports restated GHG figures for the period 2007e2010 using the NGER Act methodology resulting in a 10% increase in reported GHG emissions across those years (Xstrata, 2010). These two examples highlight the influence that reporting methodology can have on GHG emission estimates.

Regional differences in electricity generation are a major factor which adds to variability in GHG emissions between sites. Fig. 5 was plotted to compare the indirect energy intensity with scope 2 GHG emission intensity. There is a moderate relationship between the two intensities on a regional basis, particularly for Australia, the Chilean SING grid, and the Chilean SIC grid (correlation coefficients: 0.59, 0.70 and 0.75 respectively). The gradient of the trend lines between the two intensities provides an approximation of the electricity GHG emissions factor of those regions (i.e. the kg of CO2- e emitted per kilowatt-hour of electricity supplied).

The Chilean SING electricity grid generates its electricity from fossil fuel sources. In 2001 around 55.8% of SING electricity came from natural gas power plants. However by 2007 coal power had grown to 55% and diesel power grew to 26.8% of electricity production, resulting in an increase of the grids emissions factor from 0.63 kg CO2-e/kWh in 2001 to 0.92 kg CO2-e/kWh in 2007 due to this change. The emissions factor for the Chilean SIC electricity grid is lower at 0.2 kg CO2-e/kWh in 2001 but has increased to 0.36 kg CO2-e/kWh in 2007 due to growth in diesel and coal elec- tricity generation (Pimentel, 2009a, 2009b). The trend line equa- tions in Fig. 5 for the SING and SIC grids give emissions factors of 0.80 kg CO2-e/kWh and 0.29 kg CO2-e/kWh respectively, which correspond well with reported factors.

Emissions factors for Australia are generally higher than Chilean emissions factors. Australia generates a large proportion of its electricity from coal and has a fairly stagnant national emissions factor of 1.04e1.06 kg CO2-e/kWh over the period from 2006 to 2010 (DCCEE, 2011). The average emission factor for Australia is

estimated to be 0.65 kg CO2-e/kWh which is lower than the national average and suggests some variability in electricity production. This variability is evident when emissions factors are considered on a state by state basis. South Australia's electricity grid, which supplies electricity to Prominent Hill and Olympic Dam, has declined from 1.00 kg CO2-e/kWh in 2006 to 0.81 kg CO2-e/ kWh in 2010 due to significant investments in wind turbines (DCCEE, 2011; Cutler et al., 2011). The Tasmanian grid supplies electricity to Rosebery and is dominated by hydropower plants. Consequently, Tasmania's emissions factor is much lower than the country's average. However, over the period from 2000 to 2009 the emissions factor increased from 0.10 kg CO2-e/kWh to 0.34 kg CO2- e/kWh (DCCEE, 2011). This increase is due to the connection of Tasmania to the National Electricity Market in 2005 which has enabled Victoria's brown coal powered electricity to be exported to Tasmania.

The key trend present in Fig. 5 is that regions with large hydroelectricity capacity, such as Canada, Finland and the Chilean SIC grid, have lower emissions factors. In contrast, Australia, South Africa and the Chilean SING grid have much higher emissions factors due to their dependence on fossil fuels. An increase in electricity generated from natural gas provides one avenue of lowering these regions' factors but ultimately a transition to carbon capture and storage or renewable energy technologies will be required to reduce these factors further.

Mining companies rarely report scope 3 GHG emissions. During 2009, 2010 Xstrata Copper reported scope 1, scope 2 and scope 3 GHG emissions (Xstrata, Various, refer to 2009e2010 Xstrata sustainability reports). Scope 3 emissions constituted approxi- mately two thirds of their total GHG emissions for this period. Transportation of goods to and from copper operations by land comprised 98% of the reported scope 3 emissions. A value for the GHG intensity of copper cathode produced by Xstrata was calcu- lated to be 12.7 t CO2-e/t Cu for 1,440,000 tonnes of copper cathode production, when scope 3 emissions were considered. This is three times higher than when only scope 1 and scope 2 emissions were considered. This reveals the significant impact that scope 3 emis- sions can have on the total GHG emissions of copper production. Other factors such as acid production on-site compared with importing from another site can influence GHG intensity of copper. The consumption of sulphuric acid during leaching stage was re- ported to be 2.4 t/t Cu (Ayres et al., 2003). The GHG footprint of sulphuric acid produced in Australia is found to be 0.13 kg CO2-e/t acid (Ecoinvent Database, 2012 in SimaPro). Thus, it is possible that (2.4 0.13) or 0.31 kg CO2-e/t of Cu should be considered if acid is imported from another site. Norgate and Rankin's (2000) LCA study of copper assumed an acid neutral process with the assumption that a pyrometallurgical plant would supply acid to the adjacent concentration plant. The majority of the copper production pyro- metallurgical sites around the world have an acid plant onsite (Davenport et al., 2002). If this acid is not used within the process boundary, this should be regarded as a by-product with due credit given to the GHG of copper metal. This may reduce the embodied GHG of copper metal production and a full LCA is recommended if all these related factors are needed to be taken into account for a specific case study.

4.4. Water

Water is used in copper operations at all stages of production. At the mine site, water is used in various ways such as to control moisture content in slopes, dust suppression and to cool drilling and hauling equipment. Water is required in the beneficiation stage to ensure optimal grinding of ore, to allow transfer of concentrate slurry between unit processes and for flotation processes.

S. Northey et al. / Journal of Cleaner Production 40 (2013) 118e128 125

126 S. Northey et al. / Journal of Cleaner Production 40 (2013) 118e128

Additionally, water may be used to cool heavy machinery and furnaces. A modest amount of water is also required for potable uses.

Fig. 6 shows the relationship between the specific water intensity of copper and the ore grade. There appears to be only a limited relationship between ore grade and water intensity for copper, with large degrees of variability apparent independent of ore grade.

Comparison of the broad mineral processing technologies provides some limited scope to explain this variability. More vari- ability is present for operations that do not include a smelter or refinery. The inclusion of a smelter on site is generally associated with higher throughputs of ore and concentrate. Prior work by Mudd (2008) suggests that there is limited degree of 'economies of scale' regarding ore throughput (i.e. higher ore throughput leads to a reduction of water required per tonne of ore milled). Copper production plotted against water intensity is shown in Fig. 7. Limited economies of scale are shown to be present in the industry when operations are considered on a regional basis.

Fig. 7 shows that regional variability is a key factor in deter- mining water intensity of copper production. Water intensity is increased when copper operations are located in arid regions such as in parts of Chile and Australia. Conversely operations in regions with cooler climates such as in Canada and Finland display decreased water intensity.

Higher temperatures in arid regions results in more water being lost throughout the site via evaporation. Evaporation reduces the amount of water available to be recovered through tailings dew- atering activities (Castillo et al., 2001; Wels and Robertson, 2003).This requires more external water to be input into the copper extraction process to ensure optimum process efficiency. Water required for dust suppression activities is also increased due to decreased moisture content in soils (Gambatese and James, 2001).

The availability of water in the environment is reduced when operating in these areas. Access to rainwater and surface water is limited and so mines have the option of either extracting ground- water from underground aquifers or transporting water to the site via a pipeline. Groundwater levels are in decline in many regions where copper mines operate (Houston, 2002) and so access to water can represent a major blockade to expansion of mining activities in arid areas (Romero, 2001).

The impacts from increased water use when operating in remote desert regions combined with a reduction in the availability of water in these areas make the industry acutely susceptible to the

400 350 300 250 200 150 100

50

impacts of increasing water scarcity. This can have large financial ramifications on companies operating in these areas and is re- flected by significant investments in water related infrastructure as companies attempt to secure their water resources (Caceres et al., 1992).

Potential exists for large reductions in water consumption at copper operations as demonstrated at Olympic Dam (Torrisi and Trotta, 2009). Reductions of 0.20 kL/t of ore milled were achieved over the period 2004e2009. Much of this was efficiency driven rather than production driven. Gunson et al. (2012) provides an overview of strategies to reduce the water requirements of mine sites. Strategies presented include evaporation prevention, paste tailings disposal, filtered tailings disposal and ore pre-sorting. Their findings suggest that for some sites reductions of up to 74% of water consumption may be possible via ore pre-concentration and filtered tailings disposal.

There are large degrees of inconsistency within the mining industry in regards to the reporting of water consumption, management and disposal. Water reporting under the GRI frame- work is less robust when applied to mine sites and so improve- ments need to be made in this area. In particular, the clarity of information regarding sources of water extraction (especially the extent of recycling), water quality issues, and site water balances needs to be improved. Importantly, an industry wide standard of calculating key factors needs to be developed (Cote et al., 2009).

4.5. Water, energy and GHG relationship

The need to understand the relationships between water and energy consumption is growing stronger. As water scarcity increases, the amount of energy required to supply a unit of water to a mine site increases as well. Decreases in the quantity of water resources located close to mine sites increases the energy requirements to supply water to the site due to increased pumping requirements. Lower quality water resources may require treat- ment prior to use in many mining and beneficiation processes, the treatment of which also requires energy.

This relationship between water and energy is illustrated in Chile. Many copper operations in Chile are now sourcing their water requirements from sea-water desalination plants built along the coast. Desalination processes such as reverse osmosis are energy intensive. Once desalinated the water then has to be pum- ped along pipelines hundreds of kilometres long to altitudes of up to 4000 m (Global Water Intelligence, 2009), further adding to the embodied energy of the supplied water.

Mine + Leaching, SX-EW (LSE) Mine + Concentrator Mine + Conc. + LSE Mine + Conc. + Smelter

Mine + Conc. + Smelter + Refinery LSE

0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Ore Grade (% Cu)

Fig. 6. Water intensity as a function of ore grade for 31 copper producing operations.

Water Intensity (kL H20/t Cu)

S. Northey et al. / Journal of Cleaner Production 40 (2013) 118e128 127

400 350 300 250 200 150 100

50

0 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500

Copper Production (kilotonnes Cu)

Fig. 7. Water intensity as a function of copper production for 31 copper producing operations.

Australia Argentina Canada Chile Finland Laos Papua New Guinea Peru

South Africa Turkey

Improvements in water efficiency and an associated reduction in water use provide the opportunity for significant energy savings by reducing the need for importing water from energy intensive sources. Energy reductions associated with water use can also arise from improvement of the systems that handle water on-site. Changes in control strategies and equipment layout for mine-mill reticulation systems provide opportunities for significant energy reductions (Vosloo et al., 2012; Gunson et al., 2010).

The energy associated with the use of water on-site also has associated GHG emissions. Consequently, the reductions in energy use mentioned above will also contribute to a reduction in GHG intensity of copper production. Off-site treatment of water and the transportation to sites is not typically reported under mining companies' current energy and GHG reporting methodologies, so it is difficult to ascertain the energy requirements of the industry in this regard.

5. Conclusion

This paper has assessed the current sustainability data available from company reports and provides operational data which can be used in LCA. The paper also provides context for future work into the sustainability of copper production. The typical range of energy intensity was found to be 10e70 GJ/t Cu, with an average of 22.2 GJ/ t Cu. The average GHG intensity was 2.6 t CO2-e/t Cu with the range from 1 to 9 t CO2-e/t Cu. The large variation is due to ore grade, fuel sources for heat and electrical energy and reporting methods and procedures used by the companies. The water footprint can range from several kilolitre to up to 350 kL/t Cu but averaged 74 kL/ t Cu. Variation in water intensity is generally due to the geographical location of the mining operations, limited economies of scale, and the climate type (i.e. arid regions in Australia and Chile or temperate cool climates in Canada or Finland).

Opportunities exist for companies to improve their reporting standards by specifying energy type clearly for each stages of mining, mineral processing and metal production operations. Release of more detailed energy data pertaining to consumption by individual on-site processes is also necessary to gain a more complete understanding of the industry. The quality of GHG emissions reporting could be improved by clearly stating the emissions factors used and the boundaries of the analysis. Addi- tionally, long term consistency in reporting is necessary in order to be able to accurately evaluate the state of the industry and assess key trends and relationships.

Copper ore grades are expected to continue to decline into the future. The decline in ore grade is expected to place significant upwards pressure on the GHG intensity, energy intensity, and water intensity of copper production. The increasing effects of water scarcity, security of energy sources and rising GHG emissions must be monitored to ensure appropriate mitigation measures are taken for sustainable production of primary copper.

Acknowledgements

The authors would like to acknowledge the financial support from Minerals Down Under Flagship in CSIRO for providing finan- cial support to prepare this paper and the internal reviewers in CSIRO and external reviewers to improve the contents and clarity.

Required

a. Write the critical review