Computing life-cycle emissions from transitioning the electricity sector using a discrete numerical approach

Accepted Manuscript Computing life-cycle emissions from transitioning the electricity sector using a discrete numerical approach Nicholas E. Hamilton, Bahareh Sara Howard, Mark Diesendorf, Thomas Wiedmann PII:

S0360-5442(17)31170-2

DOI:

10.1016/j.energy.2017.06.175

Reference:

EGY 11183

To appear in:

Energy

Received Date: 2 February 2017 Revised Date:

21 April 2017

Accepted Date: 30 June 2017

Please cite this article as: Hamilton NE, Howard BS, Diesendorf M, Wiedmann T, Computing life-cycle emissions from transitioning the electricity sector using a discrete numerical approach, Energy (2017), doi: 10.1016/j.energy.2017.06.175. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Computing Life-Cycle Emissions from Transitioning the Electricity Sector Using a Discrete Numerical Approach Nicholas E. Hamiltona,d,∗, Bahareh Sara Howardb,d , Mark Diesendorfc,d , Thomas Wiedmannb,d a School of Materials Science & Engineering Assessment Program, School of Civil & Environmental Engineering c Environmental Humanities Group, School of Humanities & Languages d UNSW Sydney, NSW 2052, Australia

RI PT

b Sustainability

SC

Abstract

M AN U

We present a discrete numerical computational approach for modelling the CO2 eq emissions when transitioning from existing legacy electricity production technologies based on fossil fuels, to new and potentially sustainable alternatives based on renewable energy. This approach addresses the dynamic nature of the transition, where the degree of transition has an ongoing, beneficial and compounding effect on future technological deployments. In other words, as the energy system evolves, renewable energy technologies are made increasingly with renewable energy, thus becoming renewable energy ’breeders’. We apply this routine to four previously published scenarios for the transition of the Australian electricity sector, which at present accounts for about one-third of the country’s annual CO2 eq emissions. We find that three of the four scenarios fail to satisfy the electricity sector’s proportion of Australia’s share of the 2.0o C/66% IPCC carbon budget, and none of them achieves the 1.5o C budget. Only the High Carbon Price scenario could be deemed to have made any meaningful impact. An urgent, rapid transition to 100% renewable energy must be made in the whole energy sector, not just electricity, if the 1.5o C budget is to be satisfied.

1. Introduction

TE D

Keywords: Renewable Energy, Scenario, Life Cycle Analysis, Discrete Numerical Computation, Dynamic Model

AC C

EP

As the major economies of the world face increasing pressure to meet the various international accords and agreements with regard to acceptable cumulative greenhouse gas (GHG) emissions over the forward century, challenges arise when attempting to model an energy system, which, almost certainly will need to be turned completely on its head if these agreements are to be satisfied. Presently ratified by just-under two-thirds of all parties to the United Nations Framework Convention on Climate Change and coming into effect on 4 November 2016, the Paris Agreement[28] has set a global goal to limit mean global temperature to well below 2o C, if possible to 1.5o C, relative to pre-industrial levels. The parties that have currently ratified the agreement in total contribute over 80% of global emissions so perhaps the immediate half-century will witness the single largest transition in energy supply that post-industrial human society experiences. Under the assumption that the above agreements are necessary and valid, in order to facilitate the radical changes required, almost all nations throughout the globe will be required to modify how they produce their energy. The question addressed in his paper is the calculation of GHG emissions resulting from transitioning the electricity sector from fossil fuels to renewable energy, taking into account both life-cycle emissions and dynamic effects. Life-Cycle Assessment (LCA)[15, 16, 17, 32] and Input-Output (IO) modeling[26, 27] remain powerful tools to understand, respectively, the cradle-to-grave emissions and the resources required per unit of production. However, from a macro perspective, when modeling such a drastic shift in the paradigm of how energy is produced, this problem becomes inherently dynamic in nature. In 20-50 years, the ensemble of technologies may have changed from ∗ Corresponding

Author Email address: [email protected] (Nicholas E. Hamilton)

Preprint submitted to Energy

July 1, 2017

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

presently being largely provided by fossil-fuel technologies to, for instance, a mix of renewables. The majority of emissions from renewable technologies can be generally classed as upstream emissions (ie emissions obtained from resource extraction through to manufacturing and installation) and so at present renewable technologies largely rely on fossil-fuelled energy sources for coming into existence. However, as the macro-state of transition progresses towards completion, renewable energy technologies will increasingly be made with renewable energy. This trend is already commencing, as witnessed by the powering of the Tesla’s ’gigafactory’ for manufacturing Lithium batteries entirely by solar PV[31]. In addition to the above, in the mathematical sense, the deployment of new technologies and the decommissioning of existing technologies is not a continuous problem. This is because, in the real world, technologies are installed in discrete ’chunks’ of typical nameplate capacity. For example, coal-fired power plants have been installed as large plants, typically of a generating capacity of 1-2 gigawatts (GW), while large-scale wind and solar farms are generally 50-200 megawatts and residential solar systems are generally rated at 1-5 kilowatts (kW). Hence this problem lends itself to a method that is discrete in nature. Therefore this paper presents a discrete numerical solution of modelling the emissions that would result from the transition of one or more existing technologies in their current mix of contributions into one or more new or existing technologies, such that each technology satisfies its respective pre-defined specific transition scenario, and where the set of technology transition scenarios aggregate to satisfy the total demand. We take the approach that each technology is represented by a functional unit of that technology, which includes the key information from typical LCA modeling to represent a schedule of emissions year-to-year from the typical life-cycle. This schedule of emissions within each functional unit includes the process of resource extraction, construction, operation and decommissioning. This process captures the total life-cycle emissions and addresses the dynamic nature of this problem, where at any given year, the degree of existing transition may have a beneficial (or negative) compounding effect. Many existing works do not account for the compounding (breeding) effect of transition, instead utilizing emissionsfactors derived from static LCA[20, 32, 3]. Some works offer emissions-factors which is somewhat time-dynamic, however dynamic for reasons to do with technological advancement alone. One such example is Hondo et. al[17] who lists ’present’ and ’future’ cases on a per-technology basis, that differ due to the anticipated efficiency gains and/or changes to the respective manufacturing processes – neither of which are related to the breeding effect. On the other hand, there are few works that do take into account changes in origin of electricity supply, for example, Hertwich et. al[16] utilizes a vintage capital based hybrid LCA model, where ”new technologies become part of the electricity mix and thus the life cycle of the same and other new technologies”. Similarly, Arvesen et. al[1] take into account changes in the electricity source (a function of time) during manufacture when modelling the global large-scale adoption of wind technologies. Wolfram et. al[33] also employs a hybrid LCA methodology, re-calculating the respective supply-and-use tables every five years. We apply our method to demonstrate the outputs of modeling a transition of the Australian electricity sector, via re-modeling the four transition scenarios offered by ACIL Allen[25], in response to the emissions target pathways produced by the Australian Federal Government’s Climate Change Authority (CCA)[7]. This work does not provide an alternate or new form of LCA (in adherence with ISO14040-2006[19]), but instead provides an interpretation of existing LCA data to the specific problem of scenario analysis when modeling the transition of one mix of technologies, to another mix at a future date. In essence, existing LCA data is used as the input boundary conditions for providing the required data for definition of the various technology templates. The practitioner should be mindful to select pertinent LCA datasets relevant to the geography and particular modeling circumstance.

2. Method

In the present work, we have defined R6[5] object-oriented classes in the R statistical programming language[30]. The primary classes can be described as follows: 1. Template – The base functional unit to describe the lifetime emissions and energy production schedule for a single functional generation unit, such as a coal-fired power station or a wind farm. 2

ACCEPTED MANUSCRIPT

Table 1: Summary of parameters for technology templates, including the capacity (C), capacity factors (F), annual energy production (E), lifetime energy output (ET ), construction (tc ), operational (to ), post-refurbishment (tr ) and decommissioning (td ) time in years. Category FF. Coal FF. Oil + NGas

RE. Solar + Wind

RE. Other

Name Coal Coal CCS Cogen Oil NGas Gas CCS Solar Solar Thermal Wind Hydro Geo Wave Bio

C

F

ET

tc

to

td

CO2,d

CO2 eq

MW

None

MWh

MWh

Yr

Yr

Yr

Yr







T/MWh

2e+03 2e+03 0.1 4e+02 2e+02 2e+02 0.1 0.1 5 10 0.1 50 10

0.65 0.65 0.31 0.33 0.31 0.31 0.21 0.22 0.34 0.50 0.91 0.29 0.35

1.1e+07 1.1e+07 2.7e+02 1.2e+06 5.4e+05 5.4e+05 1.8e+02 1.9e+02 1.5e+04 4.4e+04 8e+02 1.3e+05 3.1e+04

E

5.7e+08 5.7e+08 8.2e+03 3.5e+07 1.6e+07 1.6e+07 4.3e+03 4.5e+03 3.7e+05 2.2e+06 3.2e+04 1.9e+06 9.2e+05

4 4 2 4 2 2 2 2 2 4 3 4 4

50 50 30 30 30 30 25 25 25 50 40 15 30

2 2 2 2 2 2 2 2 2 2 2 2 2

15 15 30 30 30 30 25 25 25 50 40 15 30

tr

CO2,c 315g/W 315g/W 178g/W 0.0028g/Wh 160g/W 160g/W 1630g/W 1630g/W 619g/W 0.0199g/Wh 836g/W 0.00706g/Wh 258g/W

CO2,o 0.0623g/Wh + 0.964g/Wh 0.101g/Wh 0.476g/Wh 0.697g/Wh 0.0744g/Wh + 0.373g/Wh 0.11g/Wh

15.2g/W 15.2g/W 15g/W

1.03 0.10 0.48 0.70 0.45 0.11 0.04 0.04 0.01 0.02 0.01 0.02 0.06

0.00141g/Wh 0.00413g/Wh 0.0097g/Wh 0.0129g/Wh 0.0602g/Wh

6.39g/W 6.39g/W 37.8g/W 37.8g/W 22.4g/W

Mod.

gτ 0.06 0.06 0.07 0.07 0.07 0.07 1.00 1.00 1.00 0.99 0.97 1.00 0.07

RI PT

1 2 3 4 5 6 7 8 9 10 11 12 13

0.12g/W

-0.5%/pa -0.5%/pa

Rank 13 8 11 12 10 9 6 5 1 4 2 3 7

SC

Note: This table also includes the various emissions contributors during the typical life cycle, ie, during construction (CO2,c ), operation (CO2,o ) and decommissioning (CO2,c ), which produces an average emissions per unit of energy output (CO2 eq). Finally, where relevant, the output modifiers have been listed (Solar degrades at ∼0.5%pa). The technologies have been ranked (Rank) in terms of their emissions per unit of production. Sources: Where possible, these figures represent an interpretation of the data provided by NREL[14], and, for technologies not defined by the NREL report, average LCA emissions were used from various other sources[25, 17, 22]

M AN U

2. Implementation – A set of multiple commissionings and/or decommissionings of a template that achieve a pre-defined technological scenario. 3. Technology scenario – A postulated time sequence of commissioning and decommissioning of each template. Multiple implementations, one per technology, describe the transition from an initial mix of technologies towards a different mix of technologies as time progresses. 4. Ensemble – A set of technology scenarios that have different emission target pathways or demonstrate different rates and/or degrees of transition utilizing the same family of templates.

AC C

EP

TE D

In the Introduction, it was outlined that in the real-world, technologies are implemented in discrete ’chunks’, which are one or more multiples of the functional unit for each generation technology. In this paper the latter are referred to as a ’technology template’. A template defines the various parameters for the commissioning of a given technology; examples of such parameters have been listed in Table 1. A technology template can be thought of as being a single power station, or a single unit of power generation, and contains information on its year of commissioning, viable lifetime, whether it can be refurbished or not, its nameplate capacity, capacity factor, and a dictionary of emissions sources that result during commissioning (upstream), operation, and decommissioning (downstream). For each template there is a pathway, which may have multiple commissionings and decommissionings of subunits within the basic functional unit. Emission quantities, being typical output from LCA, can be defined as absolute amounts (eg kg), amounts per capacity (eg kg/kW), or amounts per unit of energy (eg kg/kWh). One or more of these emission quantities in either of the above formats can be assigned to the technology template during any of the construction, operational and decommissioning stages, in order to describe the profile of emissions throughout the life of the template. Firstly with respect to a single functional unit (template), consider the amount of energy produced during each year of operation and in total over the lifetime of the template: E(t) = h · C · F · m(t) to X ET = E(t)

(1) (2)

t=1

where E(t) and ET are the amounts of energy produced during each year of operation and over the lifetime of the template respectively, whilst C, F and h = 8760 represent the capacity, capacity factor and number of hours in a year, respectively. The number of operational years for the template is given by to , and m(t) is a modifier function of t producing a scalar coefficient that serves to alter the expected annual emissions as a function of system age – useful for handling certain technologies, such as solar PV, which are known to diminish in performance over time[23], or, even perhaps introducing intentional statistical variations to performance year-to-year. 3

ACCEPTED MANUSCRIPT

Example Technology Template Schedules A, Coal, 1 x 2000MW Capacity

B, NGas, 1 x 200MW Capacity

9.0 × 10+9

2.0 × 10+8

3.0 × 10+9

5.0 × 10+7

Avg. CO2e/Unit = 1027.4590 kg/MWh

+0

4 × 10+5 2 × 10+5

0 × 10+0

0 × 10+0 Y0

Y10

Y20

Y30

Y40

Y50

Y60

Y70

Y0

C, Wind, 1 x 5MW Capacity

Y20

Y30

Y40

Y50

Y60

Y70

D, Solar, 1 x 0.1MW Capacity 8 × 10+4

1.5 × 10+6

M AN U

CO2e / kg

6 × 10+4

CO2e / kg

1.0 × 10+6 5.0 × 10+5 Avg. CO2e/Unit = 9.9929 kg/MWh

0.0 × 10+0 1.5 × 10+4

Y10

Energy / MWh

Energy / MWh

+6

3 × 10+6

4 × 10+4 2 × 10+4

Avg. CO2e/Unit = 38.5254 kg/MWh

0 × 10+0

+3

0.0 × 10+0

Energy / MWh

1.5 × 10+2

Energy / MWh

1.0 × 10+4 5.0 × 10

Avg. CO2e/Unit = 449.4410 kg/MWh

0.0 × 10+0

9 × 10+6 6 × 10

1.0 × 10+8

SC

0.0 × 10

1.5 × 10+8

RI PT

6.0 × 10+9

CO2e / kg

2.5 × 10+8 CO2e / kg

1.2 × 10+10

1.0 × 10+2 5.0 × 10+1 0.0 × 10+0

Y10

Y20

Y30

Category

Y40

Y50

Y60

Y70

Y0

TE D

Y0

Construction

Operation

Refurbished

Y10

Y20

Decommission

Y30

Y40

Y50

Y60

Y70

Termination

Figure 1: Example emissions and production schedules for a number of technology templates.

EP

Note: The typical schedules of Fossil-Fuel based technologies (Coal and Natural Gas) can be compared to the typical schedule profile for renewables (wind and solar). Specifically, the former typically demonstrates low upstream emissions, but high (and sustained) emissions during operation. Renewables, however, show high upstream emissions when compared to operation. The effect of the output modifier function can be seen with the Solar template, where the output is diminished at a rate of 0.5%pa[23].

AC C

For some technologies, the NREL data[14] give two contributions to the emissions, e.g., a non-combustion and a combustion-related contribution for coal in Table 1. We handle this by allowing each phase within the life-cycle of a template, from commissioning (c), operation (o) and decommissioning (d), to contain a set (s x ) of emissions sources of n emission contributions, which is to say: s x = {e x1 , e x2 , ..., e xn } ;

x ∈ c, o, d

(3)

where e xi , i ∈ 1 . . . n, is an individual emission source contributing to s x . Given each emission source can be defined in one of three formats, i.e. absolute amount (α), amount per unit of capacity (β) or amount per unit of energy (γ), we can determine the total emissions per year that we expect within these individual life-cycle phases according to the following:

4

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 2: Schematic framework for the modeling approach used in the present publication. For a more detailed programming flow-chart, refer to the publication supplementary documentation.

5

ACCEPTED MANUSCRIPT

n

1X k xi e xi ; where t x i=1    1 e xi ∈ α       e xi ∈ β C k xi =    ET e xi ∈ γ, x ∈ c, d      E(t)t x e xi ∈ γ, x ∈ o

e x (t) =

RI PT

(4)

(5)

TE D

M AN U

SC

where e x (t) and t x are the total emissions per year and number of years within phase x, x ∈ c, o, d respectively. C is the capacity, ET is the lifetime energy and E(t) is the annual energy output of the template at time t. Given Equations (1) and (4), each technology template can therefore provide a schedule of its energy output, and CO2 eq emissions as a function of time, from commissioning, through operation and to decommissioning. Typical schedules for coal, natural gas, solar and wind have been illustrated in Figure 1 using the parameters listed in Table 1. A functional unit can be retired early, provided the current year is within the operational window, in which case its operational emissions are truncated and the downstream emissions are brought forward to the point of early retirement – a process that will affect the average emissions per unit of energy produced over the lifetime of the template. In Table 1, also indicated in Figure 1, we have included an additional period to the operational period, referred to as the ’refurbished’ period, where for certain technologies the standard operational life can be extended, thereby avoided additional upstream emissions that would be associated with a fresh technology replacement. Further development to the modelling described in this work may consider performance enhancements to the template at the point of refurbishment, particularly for technologies that degrade over time, such as solar PV, however this feature is not incorporated in the present model. Continuing with the next class used within this work, an implementation is an intended pathway of desired energy output produced by one or more (de)commissionings of a single template as a function of time. In order to satisfy the required energy output, it contains a dictionary (or list) of template (de)commissionings, which have each been either commissioned or decommissioned at different years throughout the time envelope of the implementation. If the implementation contains a list of n template (de)commissionings, then the energy and emissions produced at time t within the implementation is as follows: Ei (t) =

n X

E j (t)

(6)

j=1

n h X i e j,c (t)λ j,c (t) + e j,o (t)λ j,o (t) + e j,d (t)λ j,d (t)

EP

ei (t) =

(7)

j=1

AC C

given that

λ j,x

   t−t j0   δ b c  t     t−t j0 −tc c  = δ b to c       δ b t−t j0 −tc −to c td

x∈c x∈o x∈d

(8)

where bc denotes the floor operator. Ei (t) and ei (t) represent the aggregate energy and emissions at time t within the implementation i. The year of commissioning for template (de)commissioning j is t j0 , whilst δ(x) takes on the form of the Kronecker delta function, that is to say, δ(x) is equal to 1 if x is zero else 0. Accounting for construction time tc relative to t, that is to say time t + tc within pathway p, the difference between the amount of energy that we seek to have delivered (by an unknown quantity of template (de)commissionings) and the energy anticipated to be delivered by the members of the list of currently implemented template (de)commissionings is: ∆E(t + tc ) = E p (t + tc ) − Ei (t + tc ) 6

(9)

ACCEPTED MANUSCRIPT

where E p (t + tc ) represents the energy that is sought within the pathway, Ei (t + tc ) represents the energy that is currently produced by the list of template (de)commissionings and ∆E(t + tc ) represents the excess (or shortfall) in energy that is required to supply demand at time t + tc . If the amount of energy that each functional unit is able to produce at age 0 is E(0), then the quantity q of functional units requiring commissioning (or decommissioning) is as follows: '

RI PT

∆E(t + tc ) q= E(0) $

(10)

M AN U

SC

where be denotes the closest integer operator. If q > 0, then this means q units of the template should be commissioned at time t so that they are commissioned with sufficient time to be producing the necessary output at time t + tc , whereas, if q < 0, then this means |q| existing units should be decommissioned at time t + tc , removing their contribution from that point forward. This is achieved by iterating over the list from oldest to newest, and retiring available quantity qi until |q| has been exceeded. Technologies that reach the end of their life become barren in terms of energy production and therefore, in this case, adding new capacity is synonymous with generating replacement capacity. A scenario contains numerous templates each handled by an individual implementation which represents the desired energy output to be contributed by a given technology. If the scenario contains k implementations, then the total energy and emissions produced by the scenario are as follows: E s (t) =

k X

Ei (t)

(11)

ei (t)

(12)

i=1

e s (t) =

k X i=1

TE D

where E s (t) and e s (t) are the energy and emissions for a scenario at time t within scenario s. Each scenario comprises a family of technological transitions, each representing a time-series of the commissioning and decommissioning details for each template. Scenarios have been set up in a way that allows us to consider the compounding effect of the overall transition, that is, the increasing use of renewable electricity to manufacture, construct and operate each technology. As more renewable generation capacity is installed over time, all parts of the energy system become less carbon intensive. Using IO techniques[33, 9] and life-cycle carbon inventories we can assign a factor, g(τ), to each technology, τ, using: T I Mτ − DI Mτ ; 0 ≤ g(τ) ≤ 1 (13) T I Mτ where T I Mτ and DI Mτ represent the total and direct emissions intensity multipliers respectively for technology τ. When DI Mτ is equal to T I Mτ , then g(τ) is equal to zero, which would be the case for the most imperfect technology that is totally impartial to the degree of green electrification and where all emissions come from the power plant. When DI Mτ is equal to zero, then g(τ) is equal to unity, the case for the most ideal technology where no direct emissions come from the power plant. Values for g(τ) used in this work have been included in Table 1 where g(τ) is approximately equal to 0.06-0.07 for fossil-fuel type technologies and 0.97-1.00 for renewables. Logically these values makes sense, since fossil fuel power plants produce their output from the burning of fuel, which is almost independent of the degree of renewable electrification, whereas, renewables have zero direct emissions but high upstream emissions associated with manufacturing and installation, which are electricity-intensive operations and hence strongly dependent upon the means of producing electricity. Within a scenario s containing k implementations, we can use the above g(τ) factor to determine the ’degree of transition’, φ(t), throughout the transition envelope:

AC C

EP

g(τ) =

k P

φ(t) =

Ei (t)g(i)

i=1

E s (t) 7

(14)

ACCEPTED MANUSCRIPT

This degree of transition can be used to ’adjust’ the annual emissions for each implementation if we firstly split the emissions into a fixed (static) component and a variable (dynamic) component which are allocated based on the overall degree of transition within the scenario: ei (t) = ei,s (t) + ei,d (t)

RI PT

(15)

where ei,s and ei,d are the static and dynamic components expressed as follows:   ei,s (t) = ei (t) · 1 − φ(t)

and

ei,d (t) = ei (t) · φ(t)

(16) (17)

SC

Therefore if the static component remains unchanged whilst the dynamic component described within Equation (17) is scaled by the technology specific factor provided by Equation (13), then the adjusted annual emissions per implementation become: 0

M AN U

ei (t) = ei,s (t) + ei,d (t) · g(i) 0

(18)

AC C

EP

TE D

where ei (t) is the adjusted annual emissions at time t within implementation i. In the above, Equation (18), for example, if the degree of transition is zero, then the dynamic component will be zero and the adjusted emissions will be unmodified, regardless of the technology. However, if the degree of transition is 0.5, then this will produce a 50/50 split between the dynamic and static proportions, and a ’perfect’ technology with g(τ) = 1 will have its overall annual emissions diminished by 50%, whilst a less ’perfect’ technology, with say g(τ) = 0.5, will have its emissions diminished by 25%. The implications are that we are now able to adjust the aggregated emissions for each technology, in each scenario, across the simulation envelope, to determine the compounding effect over time that the family of transition paths has based on the dynamic nature of the degree of transition and the suite of technology templates being implemented. Finally, an ensemble contains one or more scenarios, to be processed and compared between each other. Scenarios in each ensemble share the same technology templates, however, have different intended greenhouse target schedules. In the results presented below, the selected greenhouse target schedules of the Australian Climate Change Authority[7] have been processed as an ensemble. A concise schematic framework for the calculation methodology has been provided in Figure 2, whilst a detailed programming flowchart to describe the processing of the above equations in an iterative manner has been provided in the publication supplementary documentation.. In the next section, we compare the results of executing the above simulations with previously published technology scenarios by ACIL Allen, which do not take account of the dynamic compounding effect of using renewable electricity to manufacture renewable energy technologies. The results of the original modeling by ACIL Allen[25] do not make inference to any kind of emissions budget allocation for Australia moving forward, and so we take this opportunity to compare he re-modeled results against a conservative allocation for the Australian electricity sector. Where references and comparisons have been made to Australia’s emissions budget, we have assumed that Australia’s ’fair-share’ of global emissions is 1%, a charitable assumption[24]. The electricity sector within Australia is responsible for 33% of Australia’s annual emissions[10, 11], however in the past has been as high as 38%[4] to 44%[8]. Given the above, for the purposes of this work, Australia’s budget is deemed to be 33% of 1% of the values stipulated by the IPCC[29, tab. 2.2]. 3. Results and Discussion We have modeled the four Australian technology scenarios provided by ACIL Allen[7, 25], for which the comprehensive generation and operational emissions data has been made available online by the Climate Change Authority[6]. The technology shares for the four scenarios have been included in Figure 3. The Central Policy scenario has a carbon price intermediate between those of the Low Carbon Price and High Carbon Price scenarios. This data provides the 8

ACCEPTED MANUSCRIPT

Annual Demand and Shares by Technology for Scenarios Total and stacked relative shares per technology, historical and projected No Carbon Price

Central Policy

High Carbon Price

Scenario

400

Scenario

Scenario

Scenario

300

200

Historical

Historical

Historical

● ●● ● ●●●●● ● ● ● ● ●● ●● ●●●

● ●● ● ●●●●● ● ● ● ● ●● ●● ●●●

● ●● ● ●●●●● ● ● ● ● ●● ●● ●●●

RI PT

Annual Demand ( TWh / year)

Low Carbon Price

Historical

● ●● ● ●●●●● ● ● ● ● ●● ●● ●●●

100

2000

2020

2040

2000

Source

Coal

2020 CoalCCS

2040 NGas

2000

GasCCS

Oil

SC

0 2020

Bio

Geo

2040

Hydro

Solar

2000

2020

2040

Wind

M AN U

Figure 3: Australia’s electricity consumption, historical[18] and projected under ACIL Allen’s[25]) various carbon price scenarios. Note: Indicated by the vertical dashed line is the year 2011, as reported by the IPCC, where global emissions budgets subsequent to this year deemed to limit warning to 1.5, 2.0 and 3.0o C (with 33%, 50% and 66% certainty) have been clearly defined[29, Table 2.2]. The stacked area charts indicating the relative technology shares consistent with the transition pathways considered within the present work.

AC C

EP

TE D

input target scenario for the respective technology implementations where the template parameters have been listed in Table 1. Comparison of the target and modeled energy outputs have been listed in Table 2 and a direct comparison between the previously reported annual emissions found by ACIL Allen and the results of the present work have been listed in Figure 4, which shows agreement in the overall pattern of emission evolution vs time for the scenarios under investigation. In making this comparison, note that we have used the emissions-factors from NREL and other sources (as summarized in Table 1), which are not the same as the highly localized emissions factors used by ACIL Allen as part of their PowerMark LT modeling suite. ACIL Allen also handles the addition of new capacity in ’any optimal increment’, whereas, the modeling as part of the present work permits addition of capacity in discrete ’chunks’ of typical nameplate capacity as outlined previously in the methods. This work obtains slightly higher emission than ACIL Allen. However, unlike ACIL Allen, we have incorporated into the technology templates the full upstream emissions, decreasing over time as renewable electricity is used increasingly to produce renewable energy technologies. The latter effect gives our ’adjusted’ emissions. A typical scenario, comparing the target vs modeled, piecewise in nature, is indicated in Figure 5. A ramp period of 50 years was utilized, to deploy technology linearly from zero to the commencement capacities at the beginning of the pathway definitions, being 1990, the earliest data available via the International Energy Agency (IEA)[18] with respect to Australia. This ramp period prevents harmonic/cyclical emissions patterns (a mathematical anomaly) if the target energy required within any of the implementations commences with a value significantly greater than zero resulting in the deployment of many units instantaneously within the first year. The implications of this instantaneous deployment is that an equivalent number of units in x years (x = tc + to + td ) would require replacement, creating a regular ’spike’ in emissions every x years, particularly for technologies that demonstrate high upstream emissions. The use of the ramp period can be avoided completely by providing pathways that have sufficient history where the target energy commences from zero. The time evolution of the degree of transition has been illustrated in Figure 6, which indicates that the High Carbon Price scenario only achieves a 70% degree of transition, largely due to the assumption by ACIL Allen of Coal-CCS technology playing a role instead of a fully renewable system. To be clear, achieving only a 70% degree of transition is not because we are modelling the electricity sector only, rather because 106.1 out of a total 239.75TWh (30.7%) is still contributed by fossil fuel sources in 2050. The transition here is largely completed by around 2030, where the degree of transition plateaus and makes no significant gains beyond that point. The Central Policy and Low Carbon Price scenarios only achieve a 54 and 50% degree of transition respectively, whilst the No Carbon Price scenario is 9

ACCEPTED MANUSCRIPT

Table 2: Summary of the results from processing the four scenarios as an application of the present work.

RE. Solar + Wind

RE. Other

High Carbon Price

FF. Coal FF. Oil + NGas RE. Solar + Wind

RE. Other

FF. Coal

TE D

Low Carbon Price

Et /T Wh 169.84 173.18 104.58 41.11 47.77 80.16 5.29 54.88 150.27 17.65 21.97 46.57 169.84 38.72 56.28 41.11 42.30 49.59 5.29 92.75 144.34 17.65 90.48 95.44 169.84 180.17 103.67 41.11 41.09 93.99 5.29 54.09 151.52 17.65 21.63 30.77 169.84 227.83 329.29 41.11 33.67 49.97 5.29 51.56 69.61 17.65 20.08 20.22

FF. Oil + NGas

EP

RE. Solar + Wind

RE. Other

No Carbon Price

FF. Coal

FF. Oil + NGas RE. Solar + Wind

RE. Other

Em /T Wh 170.94 170.94 102.56 41.45 48.04 80.72 5.30 54.88 150.27 17.65 21.97 46.58 170.94 34.19 56.98 41.45 42.60 49.12 5.30 92.74 144.33 17.65 90.46 95.42 170.94 182.33 102.56 41.45 41.52 94.30 5.30 54.08 151.52 17.65 21.64 30.78 170.94 227.92 330.48 41.45 33.91 49.74 5.30 51.56 69.61 17.65 20.06 20.23

CO2 eq/MT 175.62 175.59 84.33 19.20 22.36 37.18 0.88 1.10 2.33 1.52 0.41 0.83 175.62 25.04 5.75 19.20 19.97 17.86 0.88 1.06 2.75 1.52 1.14 1.09 175.62 187.29 84.33 19.20 19.62 42.73 0.88 1.15 2.36 1.52 0.38 0.45 175.62 234.23 339.64 19.19 16.09 23.47 0.88 0.99 0.95 1.27 0.32 0.32

CO2,a eq/MT 174.05 172.35 81.48 19.01 21.89 35.75 0.75 0.78 1.08 1.34 0.36 0.52 174.05 23.95 5.50 19.01 18.98 16.97 0.75 0.33 0.84 1.34 0.53 0.51 174.05 183.86 81.65 19.01 19.21 41.18 0.75 0.82 1.17 1.34 0.34 0.34 174.05 230.48 334.61 18.99 15.79 23.08 0.75 0.74 0.73 1.12 0.30 0.30

∆/% -0.9% -1.8% -3.4% -1.0% -2.1% -3.8% -14.2% -29.3% -53.6% -12.1% -13.0% -37.9% -0.9% -4.4% -4.4% -1.0% -5.0% -5.0% -14.2% -69.2% -69.7% -12.1% -53.8% -53.7% -0.9% -1.8% -3.2% -1.0% -2.1% -3.6% -14.2% -29.0% -50.4% -12.1% -11.6% -24.2% -0.9% -1.6% -1.5% -1.0% -1.8% -1.7% -14.2% -25.4% -23.5% -11.7% -7.4% -6.9%

RI PT

FF. Oil + NGas

Year 2010 2030 2050 2010 2030 2050 2010 2030 2050 2010 2030 2050 2010 2030 2050 2010 2030 2050 2010 2030 2050 2010 2030 2050 2010 2030 2050 2010 2030 2050 2010 2030 2050 2010 2030 2050 2010 2030 2050 2010 2030 2050 2010 2030 2050 2010 2030 2050

SC

Series FF. Coal

M AN U

Scenario Central Policy

AC C

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

Note: Here we have listed the target energy output (Et /T Wh), modelled energy output (Em /T Wh), annual emissions (CO2 eq/MT ), adjusted annual emissions (CO2,a eq/MT ) and the difference between the unadjusted and adjusted annual emissions (∆/%).

10

ACCEPTED MANUSCRIPT

Annual Emissions for Scenarios Comparison between this and existing work, as reported

300

Central Policy

Low Carbon Price

High Carbon Price

No Carbon Price

Source This Work (Adjusted)

SC

ACIL Allen

RI PT

400

200

100

0 2010

2020

M AN U

Annual CO2eq Emissions (MT / year)

Scenario

2030

2040

2050

TE D

Year

Figure 4: Comparison of the adjusted annual emissions (MT), across the years 2010 through to 2050, between the results of the present work, and those published previously by ACIL Allen[25, figs. ES1,ES4] when modelling the four (4) greenhouse target scenarios by the Climate Change Authority[7].

AC C

EP

the most poorly performing scenario in terms of emissions mitigation, terminating with a 24% degree of transition. These progressive reductions in degree of transition are reflected in higher cumulative emissions with reduced degree of transition, as indicated within Figure 8. The signature of the emissions profiles can be observed in the relative correlation matrix shown in Figure 7, where the majority of the savings are attributed to sustained reductions in fossil-fuel emissions in the High Carbon Price scenario where the 2030-2050 annual emissions are around 75-85% below what was experienced in 2011. By comparison, the Central Policy scenario only demonstrates a relatively meagre reduction in annual emissions of around 40% by 2050, after passing through a mild increase (+6% at 2030) between 2025-2035. It is also interesting to note the apparent periodic nature of the emissions in the renewable class of technologies. Given the majority of their emissions occur during manufacturing, one might initially think that this can be attributed to replacement technologies being issued. However, upon reflection on the transition pathways for the High Carbon Price scenario, the apparent periodicity can be explained by an accelerated period of deployment for solar technologies around the year 2025, and again when approaching 2040. Figure 8, compares the cumulative emissions over time since 2011. Here we can see that all scenarios with the exception of the High Carbon Price scenario fail to satisfy the 66% probability 2.0o C emissions budget. The adjusted emissions can be seen in the same figure, which equates to a reduction in cumulative emissions (since 2010) of about 2.6% in the Central Policy and 4.5% in the High Carbon Price scenarios. In absolute terms, these percentage reductions correspond to a reduction in the cumulative emissions from 7.47 to 7.28GT and 3.51 to 3.36GT for the Central Policy and High Carbon Price scenarios respectively. 11

ACCEPTED MANUSCRIPT

Example Pathway High Carbon Price Scenario, Coal Techology Implementation, Target vs Modelled

Series

●

● ●

● ●

● ● ●

●

Model

● ●

●

●

●

●

Target

●

●

● ● ● ● ● ●

Region

● ● ● ● ●

Ramp Period

●

● ●

100

● ●

●

SC

● ● ●

●

●

●

● ●

●

● ●

●

● ●

●

●

50

● ● ● ● ● ● ● ● ● ● ● ● ●

0 1975

M AN U

Annual Energy (TWh / year)

●

150

RI PT

●

● ●

2000

●

●

●

● ●

●

● ●

●

●

●

●

●

●

●

●

2025

●

●

●

●

●

●

●

●

2050

TE D

Year

Figure 5: Example implementation, showing the piecewise nature of the modeled pathway overlayed against the target pathway. Note: The stepped nature of the modeled pathway is a result of the incremental deployment and decommissioning of functional units comprising the scenario. Here we have demonstrated the modeled pathway for coal-fired power plant transition, where post year ∼2010, contributions are gradually diminished.

AC C

EP

One of the principal objectives of this research was to better understand the compounding effect of the renewable electricity transition, where installed renewable capacity essentially becomes ’a renewable energy breeder’ for subsequent transitions when heading towards a complete degree of transition. As shown previously, each annual emissions can be adjusted to produce an adjusted subsequent annual emissions that accounts for the present mix of technologies which cumulatively satisfy the required total electricity demand. In order to understand the effect, consider Figure 8, which describes the cumulative emissions contributed by the fossil and renewable class of technologies, as well as in total for each scenario. When considering the cumulative emission since 2011, the best case in terms of the compounding effect is the High Carbon Price scenario, which implies a 4.5% reduction in the expected cumulative emissions after accounting for the degree of transition. Although this seems almost negligible at first glance, we must consider that firstly this scenario only experiences a 70% degree of transition, secondly the majority of the transition takes place between the years 2020 and 2030, and thirdly we are considering the cumulative emissions, not the annual emissions. Therefore, the beneficial effect of the renewable transition is simply being ’overshadowed’ by the fossil-fuel-related emissions already released prior to 2030. To illustrate the above point, in Figure 8, we have included the cumulative emission profiles starting at several commencement years (2011, 2020, 2030, 2040 and 2045). Within the same High Carbon Price scenario, the overall compounding effect is elevated to above 12.5% if the cumulative emissions are measured from 2030 and beyond. This too may seem low, however despite considering the renewable technologies alone, whose emissions are reduced in excess of 69% by the transition process, the contributions from the renewable classes are still overshadowed by the 12

ACCEPTED MANUSCRIPT

Degree of Transition for Scenarios Historical and Projected

Scenario

High Carbon Price, 70%

Low Carbon Price

60

Central Policy

Central Policy, 54%

High Carbon Price

SC

Low Carbon Price, 50%

40

20

Historical Degree of Transition ●●●●●●●●

1990

● ●●●●●●●●●

2000

●●●

●

2010

M AN U

Degree of Transition (%)

RI PT

No Carbon Price

2020

2030

2040

No Carbon Price, 24%

2050

TE D

Year

Figure 6: The degree of transition for the scenarios under investigation, calculated as the weighted sum of the energy contributions of each technology. Note: Each energy contribution is weighted by the individual ’g’ factor divided by the total energy produced within the respective scenario, as described within Equation (13).

AC C

EP

cumulative emissions from fossil-fuel sources. In fact, even with a 70% degree of transition, the annual emissions from fossil fuel sources terminate at a level which is 16.75 times higher (22.47 vs 1.34MTpa) than what is contributed by renewable sources. The bottom line is that the accumulated carbon emissions from installing renewable energy capacity to replace existing fossil-fuel capacity are almost negligible compared with accumulated coal emissions and the most rapid transition from fossil fuel sources as early as possible will have the most meaningful effect. As a final discussion point, the method outlined in this publication is relatively easy to implement, and therefore provides a valuable tool for policy and decision makers for the optimization of proposed nonlinear transition pathways. Given that we are now at a point where time is of the essence if meaningful impact is to be realized, having a rapidlydeployable computational tool that is able to process and compare hundreds or thousands of subtley different scenarios will provide valuable information when deciding the most suitable path of least resistance in the migration towards an energy network supplied by 100% renewable technologies. 4. Conclusions This research has developed a discrete numerical approach for calculating the life-cycle greenhouse gas emissions from transitioning an electricity sector from fossil fuels to renewable energy and/or other low carbon technologies. The method handles the repeated commissioning and/or decommissioning of technologies in order to satisfy a given scenario or family of transition scenarios. Our software is based on R6 object oriented classes for the R statistical environment. 13

ACCEPTED MANUSCRIPT

Relative Emissions Between Years Relative Emissions, Year to Year (R/C) Central Policy

High Carbon Price

2040

SC

2020 2010

2040 2030 2020 2010 2010

2020

2030

2040

M AN U

2050

2050 2010

2020

Renewable

Comparison Year (C)

2030

Fossil

RI PT

2050

2030

2040

2050

% Change

TE D

Reference Year (R)

−84% −75% −60% −37%

0%

58% 151% 298% 531%

EP

Figure 7: The relative changes over time when comparing two years are highlighted in the above emissions correlation matrix, where the colour gradient is mapped to the emissions emitted in one year along the x-axis, relative to a second year along the y-axis.

AC C

Note: If a point on the surface resides within the region to the lower-right of the black line, this represents a percentage difference in emissions at one year t relative to a previous year, t − ∆t, while if the point resides in the region above and to the left of the black line, this represents the inverse, being the percentage difference between emissions at one year t relative to a future year, t + ∆t. Points along the black line have a percentage change of exactly 0%, since the difference in emissions between one year and the same year is always zero.

Our method creates representations of the simplest functional unit of a given technology that can be deployed and/or retired (i.e. implemented) sequentially in order to achieve a desired energy output pathway as defined by the user. Multiple sets of implementations make up a scenario, and multiple scenarios make up an ensemble, which can be used for comparing scenarios within the ensemble. We have applied this method to the execution of several scenarios, previously defined and reported elsewhere, for various degrees of transition for the electricity sector in Australia. We have shown correlation between the output of our simulations and the anticipated annual time-series emissions as reported previously[25]. We have accounted for the dynamic nature of the transition process, where the degree of transition is used to adjust the expected emissions, serving to illustrate the compounding effect of rapid transition towards sustainable alternatives. This comes into effect by diminishing the expected cumulative emissions in the scenarios considered in this work, by up to 4.5% when considering cumulative emissions since 2011; however, this reduction increases to over 14

ACCEPTED MANUSCRIPT

Cumulative Emissions for Scenarios Emissions vs Adjusted Emissions, Various Commencement Years Low Carbon Price

−33.7% −36.2% −38.1%

−43.1%

−44.4%

−44.9%

−47.4%

Emissions Adjusted

Emissions Budgets

Since

B1.5C

2011

2030

2020

2040

2050

−12.5% −13.5% −13.1%

2040

2050

2040

2020

2010

2030

2020

2010 Emissions

−4.5% −7.5%

B2.0C

2045

TE D

Source

2050

2040

2030

2020

2010

0.1

B1.5C

−2.6% −3.0% −3.5% −4.1% −4.4%

M AN U

B1.5C

B2.0C

Total

B1.5C

1.0

−2.5% −2.9% −3.4% −4.0% −4.0%

B2.0C

2050

−1.6% −1.7% −1.6% −1.6% −1.6%

B2.0C

2040

10.0

2030

0.001

2030

0.010

−52.2% −57.0% −65.0% −65.8% −65.3%

SC

−19.4% −20.4% −19.9% −20.0% −19.6%

−4.8% −4.9% −4.9%

RI PT

−32.8% −35.3% −37.1%

0.1

0.100

−2.4% −3.5%

Renewable

Cumulative CO2eq Emissions (GT)

High Carbon Price −2.0% −2.3% −2.6% −3.1% −3.3%

Fossil

1.0

Central Policy

−2.0% −2.3% −2.5% −3.0% −3.2%

2020

−1.5% −1.6% −1.5% −1.5% −1.5%

2010

No Carbon Price 10.0

Figure 8: The cumulative emissions for the four ACIL Allen scenarios.

Note: Also shown are the IPCC[29] remaining budgets for 1.5 and 2.0o C with 66% probability. The IPCC values have been scaled by 1% being Australia’s fair-share allocation relative to the globe[24], and by a further 33%[10] being the proportion of Australia’s emissions attributed to the electricity sector.

AC C

EP

12.5% if the aggregation is commenced after 2030, being the time when the majority of the transition commences for the High Carbon Price scenario. Over this period, the adjusted annual life-cycle emissions contribution from renewable sources in the High Carbon Price, High Renewable Energy scenario are 69% below the unadjusted annual emissions Out of the four ACIL Allen scenarios, three of them failed to satisfy Australia’s share of the 2.0o C/66% IPCC budget[29], and none of them achieved the 1.5o C budget. Only the High Carbon Price scenario could be deemed to have made any meaningful impact. It is very clear that a reduction in fossil-fuel sources through to complete abolition is imperative if the 1.5o C budget is to be satisfied. Furthermore, our results indicate that an early, aggressive shift away from fossil fuels creates the most benefit. Several hourly computer simulation models have shown that 100% renewable electricity is technically feasible, reliable and affordable for Australia [13, 12, 2], the USA[21] several other countries and regions, and the whole world[22]. Future research is needed to apply our method to calculating the CO2 eq emissions from transitioning the whole energy sector, not just electricity, from one that is predominantly fossil fueled to one that is predominantly or entirely renewable. References [1] Anders Arvesen and Edgar G Hertwich. Environmental implications of large-scale adoption of wind power: a scenario-based life cycle assessment. Environmental Research Letters, 6(4):045102, 2011. doi: 10.1088/1748-9326/6/4/045102.

15

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

[2] Australian Energy Market Operator. 100 per cent renewables study–modelling outcomes. Melbourne, Australia, 2013. URL https: //www.environment.gov.au/climate-change/publications/aemo-modelling-outcomes. [3] RJ Barthelmie and SC Pryor. Potential contribution of wind energy to climate change mitigation. Nature Climate Change, 4(8):684–688, 2014. doi: 10.1038/nclimate2269. [4] Liam Byrnes, Colin Brown, John Foster, and Liam D. Wagner. Australian renewable energy policy: Barriers and challenges. Renewable Energy, 60:711–721, dec 2013. doi: 10.1016/j.renene.2013.06.024. [5] Winston Chang. R6: Classes with Reference Semantics, 2016. URL https://CRAN.R-project.org/package=R6. R package version 2.2.0. [6] Climate Change Authority. Targets and Progress Review, Electricity Sector Emissions to 2050, Underlying Data for Figures. URL http: //www.climatechangeauthority.gov.au/targets-and-progress-review-5. Accessed 2017-01-21. [7] Climate Change Authority. Reducing Australia’s Greenhouse Gas Emissions: Targets and Progress Review—Final Report. Climate Change Authority, Melbourne, 2014. [8] Climate Council. Australia’s Electricity Sector: Ageing, Inefficient and Unprepared, 2014. [9] ClimateWorks Australia. Pathways to Deep Decarbonisation in 2050: How Australia Can Prosper in a Low Carbon World. Technical report, ClimateWorks Australia, 2014. [10] Department of the Environment. Tracking to 2020 an interim update of Australia’s greenhouse gas emissions projections. Technical report, Australian Government, 2015. [11] Department of the Environment. Australia’s emissions projections 2016. Technical report, Australian Government, 2015. [12] Ben Elliston, Iain Macgill, and Mark Diesendorf. Least cost 100% renewable electricity scenarios in the Australian National Electricity Market. Energy Policy, 59:270–282, 2013. doi: 10.1016/j.enpol.2013.03.038. [13] Ben Elliston, Iain MacGill, and Mark Diesendorf. Comparing least cost scenarios for 100% renewable electricity with low emission fossil fuel scenarios in the Australian National Electricity Market. Renewable Energy, 66:196–204, 2014. doi: 10.1016/j.renene.2013.12.010. [14] MM Hand, S Baldwin, E DeMeo, JM Reilly, T Mai, D Arent, G Porro, M Meshek, and D Sandor. Renewable Electricity Futures Study. Volume 1: Exploration of High-Penetration Renewable Electricity Futures. Technical report, NREL, MIT, 2012. [15] Stefanie Hellweg and Llorenc¸ Mil`a i Canals. Emerging approaches, challenges and opportunities in life cycle assessment. Science, 344 (6188):1109–1113, 2014. doi: 10.1126/science.1248361. [16] Edgar G Hertwich, Thomas Gibon, Evert A Bouman, Anders Arvesen, Sangwon Suh, Garvin A Heath, Joseph D Bergesen, Andrea Ramirez, Mabel I Vega, and Lei Shi. Integrated life-cycle assessment of electricity-supply scenarios confirms global environmental benefit of lowcarbon technologies. Proceedings of the National Academy of Sciences, 112(20):6277–6282, 2015. doi: 10.1073/pnas.1312753111. [17] Hiroki Hondo. Life cycle GHG emission analysis of power generation systems: Japanese case. Energy, 30(11):2042–2056, 2005. doi: 10.1016/j.energy.2004.07.020. [18] International Energy Agency. Statistics Search. Website. URL https://www.iea.org/statistics/statisticssearch/. [19] EN ISO. 14040: 2006. Environmental management-Life cycle assessment-Principles and framework. European Committee for Standardization, 2006. [20] Mark Z Jacobson. Review of solutions to global warming, air pollution, and energy security. Energy & Environmental Science, 2(2):148–173, 2009. doi: 10.1039/B809990C. [21] Mark Z Jacobson, Mark A Delucchi, Mary A Cameron, and Bethany A Frew. Low-cost solution to the grid reliability problem with 100% penetration of intermittent wind, water, and solar for all purposes. Proceedings of the National Academy of Sciences, 112(49):15060–15065, 2015. doi: 10.1073/pnas.1510028112. [22] MZ Jacobson, MA Delucchi, ZAF Bauer, C Savannah, WE Chapman, MA Cameron, A Cedric, L Chobadi, JR Erwin, SN Fobi, et al. 100% Clean and Renewable Wind, Water, and Sunlight (WWS) All-Sector Energy Roadmaps for 139 Countries of the World By 2050. DRAFT, 2015. URL http://web.stanford.edu/group/efmh/jacobson/Articles/I/WWS-50-USState-plans.html. [23] Dirk C Jordan and Sarah R Kurtz. Photovoltaic degradation rates—an analytical review. Progress in photovoltaics: Research and Applications, 21(1):12–29, 2013. doi: 10.1002/pip.1182. [24] Sivan Kartha. Implications for Australia of a 1.5°C Future, Working Paper. Technical report, Stockholm Environment Institute, 2016. [25] Owen Kelp and Guy Dundas. Electricity Sector Emissions, Modelling of the Australian Electricity Generation Sector. techreport, Acil Allen Consulting, 2013. [26] Ronald E Miller and Peter D Blair. Input-output analysis: foundations and extensions. Cambridge University Press, 2009. [27] Shinichiro Nakamura and Keisuke Nansai. Input–output and hybrid LCA. In Special Types of Life Cycle Assessment, pages 219–291. Springer, 2016. doi: 10.1007/978-94-017-7610-3 6. [28] United Nations. Framework Convention on Climate Change (2015) Adoption of the Paris Agreement. In 21st Conference of the Parties, Paris: United Nations, 2015. [29] R.K. Pachauri, L.A. Meyer, and Core Writing Team. Climate Change 2014: Synthesis Report. Contribution of Working groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. IPCC, 2014. [30] R Core Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria, 2016. URL https://www.R-project.org/. [31] Tesla Motors. Tesla Gigafactory, 2017. URL https://www.tesla.com/en_AU/gigafactory. Accessed 19th January 2017. [32] Daniel Weisser. A guide to life-cycle greenhouse gas (GHG) emissions from electric supply technologies. Energy, 32(9):1543–1559, 2007. doi: 10.1016/j.energy.2007.01.008. [33] Paul Wolfram, Thomas Wiedmann, and Mark Diesendorf. Carbon footprint scenarios for renewable electricity in Australia. Journal of Cleaner Production, 124:236–245, 2016. doi: 10.1016/j.jclepro.2016.02.080.

16

ACCEPTED MANUSCRIPT * Computational method for modelling CO2 emissions transitioning the electricity sector * Convenient method that considers the dynamic compounding / breeding effect * The act of transition has a compounding effect on future technological deployments

AC C

EP

TE D

M AN U

SC

RI PT

* Method demonstrated using scenarios applicable to the Australian electricity sector