Greenhouse gas emissions from aluminium production – a life cycle approach
T. E. Norgate and W.J. Rankin
CSIRO Minerals, Box 312, Clayton South Vic 3169, Australia
ABSTRACT
A
Life Cycle Assessment (LCA) of aluminium metal production from bauxite mining
through to aluminium smelting has been carried out. The process routes
considered were:
· the Bayer process for alumina production
· the Hall-Heroult process for aluminium production
· the carbothermic reduction process for aluminium production
A number of developing technologies associated with these processes were also included in the study, namely:
· drained cathodes, inert anodes and low temperature electrolytes for the Hall-Heroult process
· direct and solvent-based processing options for carbothermic reduction of alumina and also of dross
The effect of using the latest electricity generation technology for black coal, natural gas and hydroelectricity was also investigated in the study. The greenhouse gas emissions from each process are compared on a life cycle basis and the potential for reducing these emissions is discussed.
INTRODUCTION
Aluminium production is a major industry in many industrialised countries and is a significant contributor to industrial greenhouse emissions due to the relatively high energy consumption of the process, particularly in the smelting stage. For example, aluminium production contributed about 30% to the total greenhouse gas emissions from all industrial processes in Australia in 1998 (National Greenhouse Gas Inventory (1)). Despite cell emissions being reduced considerably in recent years, the Hall-Heroult smelting process still causes pollution problems, and indeed a strong driving force behind continued research to develop an alternative aluminium process derives from the environmental effects of the existing process.
The growing global concern on environmental issues has resulted in sanctions being increasingly imposed on companies for their environmental emissions and it is now critical that some form of environmental impact assessment of new and existing processes be carried out in addition to the customary techno-economic analysis. However, because of the large number of feed streams, by-product streams, waste streams and energy inputs that occur during a multi-step manufacturing process, it can be difficult to analyse alternative options for particular stages of the process to determine which will have the least environmental impact in isolation from the rest of the process. Life Cycle Assessment (LCA) provides a framework for such analysis.
Life Cycle Assessment (LCA) is a technique for assessing the potential environmental impacts associated with a material or product from “cradle to grave”, that is from raw material acquisition to the production, use and disposal of the material or product. It involves the compilation of an inventory of relevant environmental exchanges of the material or product throughout its entire life cycle and evaluating the potential environmental impacts associated with those exchanges. A brief description of the LCA methodology is given in a previous paper (Norgate and Rankin (2)) and a more detailed description is given by Weidema (3).
This
paper describes a LCA of aluminium production by a number of process routes and
incorporates a number of developing technologies associated with these
processes with particular emphasis on greenhouse gas emissions.
SCOPE OF WORK
Because
of the complexity of the strict
“cradle-to-grave” approach of LCA and the lack of suitable data, the system
boundary for the study described here was restricted to “cradle-to-gate”; i.e., the processes have only been
considered to the point where refined metallic aluminium is available to the
secondary manufacturing sector. The
processes investigated are shown in Table I. All of the smelting processes
considered were based on alumina
feedstock with the exception of one case (Case H) which used an aluminium dross
feedstock. Several options for the Hall-Heroult process were investigated and
the various case studies are listed in Table I. Spreadsheet models of each
process were developed using the MS EXCELTM based LCA-PRO software
package (Norgate and Rankin (2)) with each flowsheet consisting of three
process steps; mining, alumina refining and aluminium smelting.
The environmental impact categories
considered in the LCA included solid, liquid and gaseous emissions, but only
greenhouse gas emissions are reported here.
The functional unit (ie. the measure of performance of the functional
outputs of the system) was chosen as one kilogram of primary aluminium metal.
The total process energy (also called resource energy consumption or Gross
Energy Requirement (GER)) was also estimated for each process route.
ASSUMPTIONS
AND INVENTORY
The
data used for input to each process model were derived solely from the
literature, with the data being cross-checked with more than one source where
possible. Emissions of the perfluoronated carbon compounds (PFCs)
tetrafluoromethane (or carbon tetra fluoride) CF4 and
hexafluoroethane (or carbon hexa fluoride) C2F6 that are
generated during the so-called “anode effect” in the electrolysis stage of
aluminium smelting were included in the inventory because of their significant
greenhouse effects.
The following assumptions were made in carrying out the study:
General
· bauxite is transported a distance of 1400 km by sea to the alumina refinery and alumina is transported a distance of 1600 km by sea to the aluminium smelter (average Australian figures on a tonnage weighted basis)
· 1 t of aluminium requires 1.95 t of alumina and 5.75 t of bauxite which corresponds to a bauxite ore grade of 17.4% Al, ie. 32.9% Al2O3
· diesel fuel for mining equipment is transported by road (diesel) a nominal distance of 500 km to the mine
· coke and coal are transported by road (diesel) a nominal distance of 500 km to the refinery or smelter
Table I-Process Routes and Technologies Included in
the LCA Study
|
Aluminium production |
Case study |
|
|
Hall-Heroult process for alumina -
Australian average power consumption -
limiting power consumption - drained cathodes - inert anodes - inert anodes / drained cathodes & low temperature electrolyte Carbothermic reduction (direct) of alumina Carbothermic reduction (metal solvent) of alumina Carbothermic reduction (metal solvent) of dross |
A B C D E F G H |
· electricity is produced using black coal which is transported a nominal distance of 200 km to the power station (56% of Australian electricity was produced from black coal in 1997/1998, 30% from brown coal)
· electrical power station efficiency is 35% (Australian power grid average in 1995)
· electrical power is transmitted a nominal distance of 200 km from generating plant to mine, refinery or smelter
Hall-Heroult process
· consumption rates reported for carbon anodes in the Hall-Heroult process include petroleum coke and coal tar pitch
· carbon emitted from combustion of carbon anodes is all considered as CO2 (National Greenhouse Gas Inventory (1))
Carbothermic reduction
process
· offgas from carbothermic reduction is virtually all CO which is combusted within the process to cover the fuel requirements and offset the electricity requirement of the process
· gaseous emissions associated with electricity generated from the process offgas are similar to those associated with electricity generated from black coal at the power plant
· thermal and electric losses in the electric arc furnace (EAF) for carbothermic reduction total 15%
· aluminium dross contains 50% metallic aluminium and 50% Al2O3
· recovery of metallic aluminium from dross in the dross press is 40%
· metal solvent (tin) to metallic aluminium ratio is 10:1
· the power consumption of the dross press was estimated from various sources to be in the order of 15 kWh/t dross (ie. 20 kWh/t Al product)
PROCESS DESCRIPTIONS
A considerable amount of research has been carried out to improve the existing Hall-Heroult process and also to develop alternative processes for aluminium smelting. These developments are described below, although none have been implemented commercially yet and are unlikely to be for some time according to Keniry (4).
Hall-Heroult Process
Primary aluminium smelting by the
Hall-Heroult process involves electrolysis of alumina in electric reduction
furnaces, commonly referred to as pots or cells. The overall cell reaction is:
2Al2O3 + 3C = 4Al + 3CO2 (1)
with the carbon being in the form of pre-baked carbon anodes.
The
reported electrical energy consumption of the electrolysis process varies from
12.9 to 18.3 kWh/kg Al depending on the process technology and plant operation
methods (Briem et. al. (5)). The average Australian consumption is 14.3 kWh/kg
Al (DISR (6)), while the lower limit of the process is realistically expected
to be about 12.5 kWh/kg Al (Briem et. al. (5)). The remainder of the smelting
site (excluding the anode plant) consumes about 0.88 kWh/kg Al (DISR (6)).
These two cases for the Hall-Heroult process are listed as cases A and B
respectively in Table I, with case A being considered as the base case. The
total smelting site electricity consumptions are 15.18 and 13.38 kWh/kg Al for cases A and B respectively.
Drained
Cathodes
Stability of the aluminium-bath interface
is one of the major factors affecting current efficiency in the Hall-Heroult
cell. The use of aluminium-wetted and drained cathodes should allow ultimate
stability of this interface to be achieved. This will provide the maximum
current efficiency obtainable with a given bath composition, temperature and
cathode current density. A stable aluminium-bath interface will also allow the
anode-to-cathode distance (ACD) to be reduced, thereby producing a lower cell
voltage. The lower cell voltage and higher current efficiency will reduce the
power consumption of the cell.
The
energy saving resulting from the use of drained cathodes has been estimated by
various authors (Zhang et. al. (7), Brown et. al. (8), Tabereaux et. al. (9))
to be in the order of 9% compared to the conventional Hall-Heroult cell.
Therefore the electrical energy consumption of the electrolysis component of
the total Hall-Heroult electricity requirement for the limiting-energy case
(viz. 12.5 kWh/kg Al) was reduced by 9% to 11.375 kWh/kg Al, giving a total
smelting site electricity consumption of 12.255 kWh/kg Al for this case (Case
C).
Inert
anodes
Inert or non-consumable anodes are chemically and electrochemically non-reactive. With inert anodes the overall cell reaction is:
2Al2O3 = 4Al + 3O2 (2)
While the
total cell voltage would be increased by this reaction, it is believed that
inert anodes would allow the ACD to be reduced by about 57% (Kvande (10),
Jarrett (11)), giving a net reduction in cell voltage and energy consumption of
10% compared to the conventional cell. The electrical energy consumption of the
electrolysis component for this case (Case D) was reduced to 11.250 kWh/kg Al,
giving a total smelting site electricity consumption of 12.13 kWh/kg Al.
Low temperature electrolytes
The bath temperature in conventional Hall-Heroult cells is typically in the order of 945 – 965 °C. A lower bath temperature (eg. 750 °C) can be expected to give a lower metal solubility in the bath and hence a higher current efficiency and a lower power consumption. Energy savings in the order of 2% have been reported for low temperature baths by Grjotheim et. al. (12).
Rather than investigate the effect of low temperature electrolytes alone, given this relatively small energy saving, this process option was combined with the two previous ones, drained cathodes and inert anodes, for this particular case study (Case E).
Given
that the reduction in cell power consumption for both the drained cathode and
inert anode cases are largely attributable to the reduced ACD, the energy
saving for the combined effects is likely to be less than the sum of the two
individual effects (viz. 9% + 10% = 19%). Zhang et. al. (7) have reported
energy savings for these combined process options to be in the order of 16%. A
further 1% was added to account for the additional effect of the low
temperature electrolyte, giving a total energy saving of 17% over conventional
cells for this case study. The electrical energy consumption of the
electrolysis component for this case was reduced to 10.375 kWh/t Al), giving a
total smelting site electricity consumption of 11.255 kWh/kg Al.
Carbothermic Reduction
For this processing route the mining and alumina refining stages are the same as for the Hall-Heroult process. There are several variations of this process for smelting aluminium and they have been described in detail by Rajakumar and Saunders (13) and Motzfeld et. al. (14), some of which are considered below.
Direct
carbothermic reduction
This process has been described by Motzfeldt et. al. (14) and Kibby and Saavedra (15). The process uses three furnaces, in the first of which (the “primary” furnace) alumina and carbon react according to the following reactions:
2Al2O3 + 9C = Al4C3 + 6CO (3)
Al2O3 + Al4C3 = 6Al + 3CO (4)
under conditions such that aluminium metal plus an oxide melt with about 9.5% w/w Al4C3 is produced.
The gas formed here, which contains aluminium-bearing species, is passed through the incoming charge and subsequent gas cleaning equipment for condensation of the aluminium for recycling back to the furnace. Aluminium is then tapped to a secondary furnace (the “decarbonization” furnace) where it is decarbonized by either extraction of the carbide into the alumina to form oxycarbide or reduction to the metal according to equation (4). When the carbide content of the metal decreases to about 2% w/w Al4C3, aluminium is tapped into a third furnace where fluxes are used to float the oxide and carbide to the surface of the metal, to be skimmed and recycled.
The theoretical energy requirement at 2000 °C for reactions (3) and (4) combined was estimated at 8.68 kWh/kg Al using the METSIM (Bartlett (16)) flowsheeting package. The total process energy requirement for this case (Case F) was estimated at 10.05 kWh/kg Al (Norgate and Rankin (17)).
Carbothermic
reduction using metal solvent
This process has been described in some detail by Kusik et. al (18). After pelletising or briquetting, the alumina and coke feed are charged into an EAF and the following reaction takes place at a temperature of about 2000 °C using tin as a solvent for the aluminium.
Al2O3 + 3C = 2Al + 3CO (5)
The
aluminium-tin alloy is cooled to crystallise the aluminium and the aluminium
crystals are separated from the tin at a temperature of about 320 °C (some 90 °C above the melting point of tin) using presses. Two stages
of crystallisation are considered likely. Separated tin is recycled to the EAF
after first recuperating some heat from the aluminium-tin alloy discharging
from the EAF. The aluminium crystals are re-melted and then purified (possibly
using molten sodium) at a temperature of 700 °C before being cast into ingots.
Reaction (5) taking
place in this processs is the same as the overall reaction (reactions (3) +
(4)) taking place in the previous process (also at 2000 °C), therefore the theoretical energy requirement is also the same, ie. 8.68 kWh/kg Al. This
value agrees exactly with the value reported by Motzfeldt et. al. (14) for this
process. The total process energy requirement for this case (Case G) was
estimated at 10.60 kWh/kg Al (Norgate
and Rankin (17)), being slightly higher than the direct carbothermic reduction
process as remelting of the aluminium crystals and reheating of the solvent
metal is required here.
Replacement of the
alumina feed to this process with aluminium dross (the undesirable but
unavoidable material generated at the surface of molten aluminium by oxidation)
was also investigated in the study. In this case the dross is first remelted
and some of the metallic aluminium in the dross is removed by hot pressing
prior to the dross undergoing carbothermic reduction in the electric arc
furnace.
The energy required
for reaction (5) at 2000 °C for the dross feed composition and
temperature considered was estimated using METSIM at 8.8 kWh/kg of Al produced
via this reaction. Taking into account the additional metallic aluminium from
the dross in the final product, this value reduces to 3.05 kWh/kg Al product.
On this occasion the off gas credit was insufficient to cover the process fuel
requirement. The total process energy requirement was estimated at 5.00 kWh/kg
Al (electricity) and 0.48 MJ/kg Al (fuel) by Norgate and Rankin (17).
RESULTS
The results of the study
are summarised in Table II. The contributions of the three process stages to
the total energy consumption and GWP are shown in Figure 1 for the base case
(Case A). The contribution of the mining and crushing stage relative to the
other two stages is so small that it is not visible in Figure 1. It is obvious
from this figure that the Hall-Heroult smelting stage is by far the largest
contributor to both total energy and GWP, with the electricity consumption in
this stage contributing 74% of the total process energy requirement. Given this
significant contribution to the total process energy, the effect of different
electricity generation technologies and fuel sources on the LCA results was
included in the study.
Various power generation
technologies and their associated efficiences have been described by Briem et.
al. (5). The best current (2000) efficiencies reported by these authors for
black coal, natural gas (combined cycle) and hydroelectricity are 44, 54 and
80% respectively. Hydroelectricity is often assumed not to be associated with
any greenhouse gas emissions, however
recent studies (BHP Research (19), Carvalho and Bizzo (20)) have suggested that decaying vegetation
submerged by flooding may give off appreciable quantities of greenhouse gases.
The amount of these greenhouse gases emitted from hydroelectric dams varies
greatly, depending on climatic factors and the nature of the land that is
flooded to create the dam. Based on data provided by these authors, a greenhouse
gas emission value of 189 kg CO2-e/MWh (Norgate and Rankin (17)) was
used for hydroelectricity in this study.
Figure
1-Contribution of Process Stages to Total Energy and GWP for the Base Case (A).
The GWP of the various process technologies are compared in Figure 2 for black coal-based power generation at the base case efficiency (35%) and natural gas-based power generation and hydroelectricity at their best current efficiencies (54% and 80% respectively). The GWP result of 22.4 kg CO2-e/kg aluminium for the Bayer/Hall-Heroult process route base case compares with reported values of 18.2 by Grant (21) and 24 by Huglen and Kvande (22) and Kvande (23) for the world average.
For aluminium production from alumina with fossil-fuel based electricity, the process technology having the lowest GWP (7.2 kg CO2-e/kg Al) was the Hall-Heroult process with drained cathodes, inert anodes and a low temperature electrolyte together with natural gas-based power generation (at current best practice generation efficiency of 54% - not tabulated). This represents a 68% reduction in GWP over the base case. Carbothermic reduction of dross was the only other fossil-fuel (natural gas) based case with a lower GWP of 4.2 kg CO2-e/kg Al (not tabulated), however as dross is a byproduct of aluminium processing it can never fully replace alumina as a feedstock.. Hydroelectricity reduces the GWP of these two cases to 4.6 and 3.1 kg CO2-e/kg Al respectively.
The
total energy consumption of the various process technologies are compared in
Figure 3. The total energy result of 211 MJ/kg aluminium for the base case
Bayer/Hall-Heroult process may be compared with results of 207 and 182 – 212
MJ/kg reported by Grant (21) and the
Australian Aluminium Council (24) respectively for the Australian aluminium
industry
