Beyond the Death Spiral
Transitioning to Renewable Energy in Western Australia
William Grace
Australian Urban Design Research Centre
University of Westem Australia
PO Box 2729
Cloisters Square
Perth WA 6850
Australia
T: 461 8 6318 6200
E: bill.grace@uwa.eduau
Abstract
Model based projections for the rapid uptake of rooftop solar photovoltaics in Western
Australia indicate that private capacity will be so large that the centralised network based
electricity system will become disrupted in the 2020s. By 2050 private systems may produce
around 85- 90% of projected electricity demands. In the interim period it may be more
econonically viable to avoid introducing large scale renewable energy to the network while
Planning for a completely renewable system by 2050 when rooftop solar approaches
saturation levels.
By 2050 it is projected that only around 2,250 MW of large scale renewable energy will be
needed to complement private solar PV, optimally in the form of wind energy, or a
combination of wind and wave energy. In order to avoid very large storages it will be
to retain fast response thermal generation, most likely using state of the art open
cycle gas turbines fuelled by renewable sources such as biogas from organic wastes.
The network will likely require around 32,000 MWh of energy storage to complement the
private storage by 2050, with pumped hydro utilising Perth's water supply dams, a potential
source.
Keywords
Solar, electricity, energy, battery, storage, wind, wave, network
Introduction
Ina previous study (Grace, 2015) I described the use of a system dynamics model to explore
the impact of private solar PV and battery storage on the electricity network in Westem
Australia, known as the South West Interconnected System (SWIS). They key findings of
that work were that falling costs of solar PV systems will drive exponential growth that will
eventually disrupt base-load generation. In this article I examine the broader implications of
these findings in respect of transitioning the network to 100% renewable energy.
The SWIS serves Perth, the capital city of Westem Australia and the south west which is the
most populous region of the state. Both peak and annual demand (Figure 1) has plateaued in
recent years(Australian Energy Market Operator, 2016) due to a combination of factors
including the uptake of private solar PV, energy efficiency measures and slower population
growth. Changes in the nature of the economy (particularly decline in manufacturing have
seen a decline in energy intensity, ie. electricity demand per unit of Gross State Product
(GSP).
4,000 =
10.0
3,500
&
oO
3
8
3,000 30 3
2,500 ~
,! =
3 -60 &
2,000 ——— z
3
1,500 —Peak demand (LHS) 40 8
1,000 — Average demand (LHS) Loo &
500
——Energy intensity (RHS)
Sourve: Australian Energy Market Operator (AEMO)! and author's calculations
Figure 1 Electricity demand on the SWIS
The SWIS remains very much a fossil fuel dominated network, with coal providing 30% of
capacity, gas a further 25% and hyrid gas / diesel peaking plants another 28%. In 2014-15
only 3% of the energy generated came from renewable sources, mainly wind which
accounted for 94% of renewable generation.
Over the same period there has been a rapid growth in the uptake of private solar installations
in the state, mirroring the situation in other states of Australia and across the world more
broadly. Westem Australia currently has some 650 MW of rooftop solar (Figure 2). In the
area served by the SWIS some 23% of customers have rooftop PV. The average size of
systems is 2.5 kW, although new installations in 2015-16 averaged 4.5 kW. Since 2010-11,
the average annual growth rate of installations has been 25% per annum, and system sizes
have grown at 19% per annum.
300,000
250,000 2016 capacity 647 MWp
Average array 3 kW
its
200,000
150,000
Number of uni
100,000
_ cam
0
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
Source: Clean Energy Regulator”
Figure 2 Rooftop solar PV installations in Westem Australia
4 https://www.aemo.com.au/Electricity/Wholesale-Electricity-Market-WEM/Planning-and-forecasting/WEM-
Electricity-Statement-of-Opportunities
* http://www. cleaneneryyregulator. gov.au/RET/Forms-and-resources/Postcode-data-for-small-scale-installations
2
The major factor in the growth of private solar PV is the unit cost, which has decreased by
40% over the period from 2012 to 2016 (Figure 3). These costs include the benefit of
Australia’s Renewable Energy Target under which small scale solar PV systems receive a
rebate based on an estimate of the amount of generation. A 1 kW system in the SWIS area
receives small scale technology certificates (STCs) which presently realise a discount of
about $775. The majority of these system are on residential rooftops, as the take-up in
commercial premises has been limited to date, partly impeded by restrictions placed on
connections by the goverment owned network operator, Westem Power °.
Average $IW (all system sizes)
$2.40
$1.80 eee
$1.20
$0.60
$0.00
Ck Xo 99 10h NE Oh 9K WX? 0K W'S go?
DO BOE pI ge” por pI Qe” POT pI gee po!" pak
po
Source: Solar Choice *
Figure 3 Unit cost of solar PV in Australia
The rapid growth of rooftop solar is a worldwide phenomenon. In their Q2/Q3 2016 Solar
Industry Update (Feldman, 2016), the US Department of Energy provide an insight into the
growth of rooftop solar PV in the US (Figure 4). These figures exclude utility scale solar PV
and represent annual growth rates of 50% per annum for the period 2013-15 and 30% per
annum for the projected period.
US residential and non-residential PV
demand
(installed since 2013)
0 —
2012 2014 2016 2018 2020 2022
Historic —e@—Projection
Source: U.S. Department of Energy
Figure 4 U.S. Solar PV installations
> http;//reneweconomy.comau/regulations-cause-dead-spot-in-wa-solar-market-91167/
* hhttps://www.solarchoice.net.au/blog/news/residential-solar-pv-system-prices-december-2016
Objectives of the study
Findings from the previous study identified that by the early 2020's the so-called duck curve
will be in play on the SWIS (Figure 5), i.e. there could be an intermittent over-generation
problem when minimum network loads fall below the normal operating capacity of baseload
coal generation, which is intended to nu consistently and cannot be readily cycled down and
up in a matter of hours. Steep ramping of generation is required in the hours of declining
solar generation A similar situation has been identified by the Califomia Independent System
Operator”.
Typical March day
—2015
—2018
—2021
—2023
—2026
1 4 7 10 13 «#416 ©6«©619 «22
Figure 5 The SWIS duck curve
This raises the question of both the scale and role of legacy networks in the future given:
e the likely ongoing cost reductions in solar PV and battery storage; and.
e the necessity of eliminating fossil fuel based generation by mid-century.
The work presented here uses projections of the take-up of private solar and storage to
explore what the SWIS (and by analogy all similar centralised networks) might look like by
2050, and what policies should be in place to progress an orderly transition.
The results reported here are derived from 2 separate system dynamics models, as described
below.
Models
A monthly time step model described in the previous paper®, uses arrays to model the hourly
behaviour of a typical day in each month. It has been modified for this study to run from
2015 until 2050, ie. 420 months. This model establishes electricity demand and the likely
take-up of solar PV and storage by households and businesses.
A companion hourly time step model that simulates each hour of the year, again by assuming
a typical day for each month, is used to model network storage for both private generation
and that of large scale renewable energy generation that will be added to the system as it
transitions away from fossil fuels.
Both models use Version 6.3 of the Vensim Professional software.
The monthly model determines the electricity demand separately arising from:
* http://www.caiso.com/documents/flexibleresourceshelprenewables. fastfacts.pdf
° A full explanation of the model structure, together with its documentation, is included in a detailed report on
the research at https://www.audre.org/exploring-the-death-spiral/
e residential houses; and
¢ commercial and industrial facilities.
The existing residential demand and commercial demand profiles have been determined from
historical half hourly reports of total network load, and presentations of the previous
independent market operator (IMO WA) on residential and commercial loads.
The model calculates the contribution of household scale solar energy generation, with and
without energy storage. The model calculates the payback period for a household arising
from:
¢ avoided electricity imports from the network at the residential tariff; plus
e electricity exports to the network at the residential feed-in tariff (renewable energy
buyback scheme); and.
e the installed cost of solar energy and battery storage.
The unit cost of solar PV is modelled as a stock with an initial value reflecting present unit
costs (A$2,200 /kW) which is the approximate installed cost of systems in Australia
presently, excluding the benefit of the small scale technology certificates (STCs). The model
assumes that the unit cost transitions to a final unit cost of A$1,000 /kW (Figure 6).
2,500
2,000 +
1,500 +
$/kW installed
Bb
500
o+ r ;
2010 2015 2020 2025 2030 2035 2040
Figure 6 Solar PV cost curve
The model sets an optimum storage capacity based on the capacity of the solar array, which
avoids multiple combinations of solar capacity and storage capacity. The model calculates the
additional benefit to the householder from adding storage to their solar array. It is assumed
that storage operates simply on the basis that:
e solar generation in excess of demand is stored (up to the limit of the storage capacity);
e the storage discharges to meet demand that cannot be met by solar generation; and
e remaining demand is met by the SWIS network.
The incentive to add solar PV and storage is determined by a payback period (Figure 7)
calculated from the benefits noted above and the unit cost of solar storage. The latter assumes
that the present storage costs of approximately $1,000 / kWh will drop to around $200 / kWh
(Climate Council of Australia, 2015) (Figure 8).
The model structure for commercial solar is identical to the residential model in all respects,
except for the initial conditions. The model assumes that at the outset there is no commercial
solar or storage. It is assumed that a maximum of 60% of residential dwellings could
5
accommodate solar PV and storage, and 50% of business premises. A maximum array of 7.2
kW is assumed for houses and 150 kW for businesses.
120%
100% A
80%
% uptaeke
40%
0% T T
oO 5 10 15 20 25
Payback in years
Figure 7 Uptake of solar and storage based on payback period
1200
$/ kWh
8
to)
2015 2020 2025 2030 2035
Figure 8 Cost curve for household and commercial scale battery storage
Electricity demand is based on assumptions of the market operator and manifests as a
doubling of total demand over the period 2015 to 2050 (average annual growth of 2% per
annum).
The model incorporates changes in tariff over the simulation period by assessing the
operating costs of the network as it is affected by private solar / storage. Take-up is however
dominated by the falling unit cost of those technologies rather than increases in tariff,
although these could be very significant, as previously reported.
However the take-up of private battery storage will be very much influenced by tariffs. The
payback period for private battery storage is dependent on the relative costs of imported and
exported energy. Currently in Westem Australia (for most customers) the former is around
three times the latter (the solar feed-in-tariff), meaning there is a strong business case for
storing energy for later use, thus reducing the amount of more expensive imported energy. If
however, the solar feed-in-tariff is the same as the import tariff (so-called net metering) there
is no business case for private storage. Accordingly future policy changes will have a large
impact on the take-up of private storage which is still in its infancy.
Various tariff policy scenarios that affect the likely take-up rates of solar PV and storage
were simulated including assuming the feed-in-tariff remains as it is, doubling and tripling
the feed-in-tariff. This analysis shows that to have a major influence on the take-up of private
storage, the feed-in-tariff would need to be tripled. Given the cost of such a measure to a
network that is already operating with revenues below cost, there is little likelihood of this
occuring. Accordingly, the following assumes that private storage take-up occurs in line with
the model projections using the existing feed-in-tariff, escalating network tariffs and assumed
solar PV and storage cost reductions referred to above.
Monthly modelling results
The result of this modelling (Figure 9) indicates that by 2050:
e Nearly 60% of houses and 50% of businesses have onsite solar PV systems, with output
of 6,000 MW and 15,000 MW respectively; and
e Nearly 50% of houses and 45% of businesses have onsite eneryy storage systems, with
capacity of 10,000 MWh and 20,000 MWh respectively
Although the addition of storage lags initially, it increases significantly after 2025 when unit
costs are starting to fall significantly. However even at 2050 private storage in aggregate is
still only 1.5 hours of nameplate solar capacity.
35,000
30,000 -- Solar PV capacity (MW)
25,000 + =——=Storage capacity (MWh)
20,000 ZA
15,000 a Pa
10,000 Le
5,000 ra
al
(0) — T T T T ; T
2010 2015 2020 2025 2030 2035 2040 2045 2050 2055
Figure 9 Simulated take-up of private solar PV and storage
By 2050, total electricity demand is projected to be 37,000 GWh per annum. Rooftop solar
PV in Westem Australia has a capacity factor of around 0.183, meaning that 20,000 MW of
Private solar is capable of generating some 32,000 GWh per annum, i.e. 85% of total
electricity demand. Although counter intuitive at first glance, these projections are based on
modelling the rational behaviour of electricity users who respond to their own cost-benefit
assessments of the options, i.e. purchasing electricity from the network versus a capital
investment in solar PV that has virtually no operating costs and a payback period of less than
5 years (within a decade). Unless the cost projections of solar PV tum out to be wildly
optimistic, or the cost of network generated electricity reduces significantly, it is difficult to
come to any other conclusion than, in the future, most energy will be generated onsite by
solar PV. Although it could be argued that network scale solar or wind energy could
eventually be cheaper than small scale systems on a per MW basis, the cost of distributing it
will always be a necessary on-cost that is avoided by onsite generation.
Of course to realise the private solar scenario described above, all privately generated power
would have to be utilised. In the previous study, I assumed that when private solar energy
generation exceeded total demand in the middle of the day (commencing in the early 2020s)
7
Private solar PV generation would be curtailed by the network operator (assuming it is
possible to do so), because there is currently no network storage capacity. However if this
was to occur, zero marginal cost, emission free power would be substituted by highly
polluting, expensive fossil fuel based generation, simply to enable the network to operate as it
has historically. This would be a perverse outcome.
This conclusion has very significant implications for policy formulation. Much work is going
on in Australia (Blakers, 2017)and elsewhere to identify how the national network (National
Electricity Market) and the local SWIS can transition away from fossil fuels. However most
of these studies assume that the network will be of similar scale, and operate in a similar way
to the current situation.
The reality is that the future network will be much smaller and structured to accommodate
large scale renewables, but only to complement onsite generation. Its major role will be to
store energy to balance generation with demand, which requires storage well beyond the 1.5
hours likely to be procured privately.
The second part of this study seeks to identify how this new network might develop over
time.
Renewable energy targets
Various dates for the transition to a fully renewable energy electricity system have been
suggested in Australia, and in Westem Australia. Although the national govemment has made
commitments under the Paris accord for reductions in greenhouse gas emissions in aggregate,
these commitments only require Australia to reduce emissions by 26-28 per cent (on 2005
levels) by 2030. Australia is the fifteenth largest emitter of greenhouse gases in the world
(Climate Change Authority, 2015). There is currently no meaningful approach to reducing
economy wide emissions since the current govemment abolished Australia’s emission trading
scheme in July 2014. However the Renewable Energy Target (RET) has been in place
nationally since 2001. Under this scheme some 20 per cent or 41,000GWh electricity was to
be generated by renewable sources by 2020. The current government has recently reduced
this to 33,000 GWh by 2020.
The current national opposition party has committed to 45% emissions reduction on 2005
levels by 2030, and pledged to ensure that 50% of Australia’s electricity is sourced from
renewable energy by then. The Australian Greens policy is to increase the RET to achieve
90% renewables by 2030.
The modelling results set out above provide some insights into the selection of an appropriate
target. Three scenarios have been examined for achieving 100% renewable energy in the
SWIS area: by 2030, 2040 and 2050. This is simply done by calculating the renewable energy
that would have to be generated at utility scale to complement private solar generation. The
result of these calculations is set out in Figures 10 a) to c).
40,000,000
35,000,000
30,000,000 Demand
25,000,000 = Met by FF generation
20,000,000
MWh
15,000,000 tm Met by network
— renewables
10,000,000 i
1m Met by onsite solar
5,000,000
2020 2030 2040 2050
Figure 10a) Generation mix to achieve 100% renewable energy by 2030
40,000,000 +
35,000,000 -
30,000,000 -
25,000,000 - l= Met by FF generation
£
2 20,000,000 + Met by network
renewables
15,000,000 +
lm Met by onsite solar
10,000,000 +
5,000,000 =
2020 2030 2040 2050
Figure 10b) Generation mix to achieve 100% renewable energy by 2040
40,000,000
35,000,000
30,000,000
25,000,000 lm Met by FF generation
£
2 20,000,000 = Met by network
renewables
15,000,000
m Met by onsite solar
10,000,000
5,000,000
2020 2030 2040 2050
Figure 10c) Generation mix to achieve 100% renewable energy by 2050
All scenarios assume that network renewables increase to achieve the RET target of 23.5%
by 2020, and then fill the gap between the renewable target and private solar generation
thereafter. Aggregate electricity demand increases beyond 2050 would need to be shared in
similar proportions between network renewables and onsite generation under the modelling
reported here, i.e. that private solar has reached near saturation by 2050 as a fraction of
housing and business stock.
As Figures 10a) and b) indicate, an early transition to network renewables may be
problematical in that the required capacity first increases and then decreases over time as the
cost-benefit of private onsite solar PV improves (see Figure 9) and take-up approaches
saturation. As the design life of wind turbines is 20-30 years and for solar farms is 25 years, it
will not be commercially viable to annually decrease the generation output of renewable
power stations in this way. If the real world uptake of private solar reflects the projections
here, setting a target of 2050 appears to be the most practical and economical way of
introducing large scale renewable energy to the SWIS, while eliminating fossil fuel based
generation in a gradual and orderly way.
This scenario would require renewable generation on the SWIS to increase from the currently
modest 500 GWh pa to 2,100 GWh pa by 2020 and then double over the following 30 years.
Such a scenario would also benefit from the likely decrease in unit cost of large scale
renewable energy systems over that time period. However the major focus for the network
will be to store energy rather than generate it.
The Future SWIS
An hourly model that represents a typical day in each month of the year was used to examine
in more detail the scenario identified in Figure 10c) above, i.e the situation at 2020, 2030,
2040 and on achievement of a fully renewable energy system by 2050.
The model calculates the amount of energy generated by private solar systems with battery
storage and accordingly the residual load on the local network. When the aggregate amount
of solar PV generation exceeds demand, energy is exported to the network, where it is seen as
a negative load. According to the results reported above this situation will likely commence
in the early 2020s. The model simulates the diumal network demand pattem across the year,
and matches generation and network energy storage at either (or both of) the local substation
scale or transmission scale.
The existing residential demand and commercial demand profiles have been determined from
historical half hourly reports of total network load, and presentations of the system operator
on residential and commercial loads (Australian Energy Market Operator, 2016). The annual
residential demand has been derived from the reported network loads, modified to include the
demand met by private solar. Generation pattems for household solar energy are based on
analysis of data from the network manager Westem Power (Jones, 2012). Generation pattems
and capacity factors for wind energy and large scale solar energy are derived from actual data
from existing facilities on the SWIS (Rose, 2016)’, and wave energy data for Westem
Australia derived from (Hughes & Heap, 2010). Demand and generation pattems are read
into the model via Excel lookup functions.
A screen shot of the network generation and storage element of the model is shown in
Attachment A.
” Courtesy of Sustainable Energy Now’s SIREN model http://www.sen.asn.au/
10
By 2020
By 2020 it is projected that there is 1,550 MW of private solar PV generating some 2,500
GWh of electricity. The largest single renewable energy facility on the SWIS presently is the
Collgar wind farm at 206 MW. This facility operates at a capacity factor of 0.38, generating
some 686 GWh pa. Adding a further 1,400 GWh to the network by 2020 is therefore not a
major impediment. Renewable energy in the form of onsite solar and network wind
generation are meeting about 22% of network demands.
Figures 11a) illustrates the impact that private solar exports (Figure 11b) have on network
loads. Although there is 135 MWh of private storage, this is insufficient to affect electricity
demand at the local substation scale. However at this stage baseload thermal generation
potentially becomes disrupted as minimum loads approach 1,000 MW in Autumn.
Figures 11 c)-e) show how total demand is met on an annual basis and for typical January and
July days.
By 2030
By 2030 it is projected that there is 6,000 MW of private solar generating nearly 10,000 GWh
of electricity (Figures 12b), now having a significant impact on network loads (Figure 12a).
There is also nearly 5,000 MWh of private storage, although this is insufficient to avoid net
exports of power to the network in Autumn and Spring when demands are lowest (Figure
12a).
A small amount of additional renewable energy (now 750 MW of wind) is required at
network scale to achieve nearly 50% of demand from renewable energy. Network demand
drops to zero in the middle of the day in most months of the year (Figure 12a) and very steep
ramping of the thermal network occurs, meaning that baseload coal generation will need to be
eliminated well before 2030. The system will become more dependent on fast response open
cycle gas turbines and energy storage during the 2020s.
Figures 12 c)-e) show how total demand is met on an annual basis and for typical January and
July days. In this scenario storage at the network scale required to absorb the net exports from.
Private solar, is avoided by curtailing a minor amount of wind energy.
By 2040
By 2040 it is projected that there may be 13,000 MW of private solar capacity generating
some 20,500 GWh of electricity (Figure 13b). This is likely to be supplemented with nearly
18,000 MWh of private storage which will play a more significant role in balancing supply
and demand behind the meter. However this is still insufficient storage to avoid large exports
to the network (Figure 13a).
No additional renewable energy is required at network scale, as onsite solar and 750 MW
wind is sufficient to reach nearly 75% of demand from renewable energy. Fast response
thermal generation is still required throughout the year (Figure 13c-e), although is sparsely
used in the summer months.
About 22,000 MWh of network storage is now required to absorb the net exports from private
solar, and to absorb wind generation without curtailment (Figure 13f). Storage is effective in
meeting network loads in all but the winter months.
At this stage private solar and storage is providing for about 65% of demand, wind energy for
8% of demand and the balance (27%) by thermal generation.
11
By 2050
Private solar PV systems are now nearing saturation in the system at about 60% penetration
in residential premises and 40% of businesses. Further demand growth beyond 2050 will be
met pro-rata by onsite and network generation. Accordingly the 2050 scenario provides a
picture of how the system may operate into the future.
By then the total capacity of solar generation is about 20,000 MW, and is accompanied by
32,000 MWh of private storage (Figure 14b). However this is insufficient to avoid very large
exports of solar energy to the network in the middle of the day (Figure 14a), exceeding
system demand in the summer months.
Wind energy is increased during the previous decade and its capacity now is 2,250 MW. The
implications of this level of solar and wind generation is significant for the network (Figure
14c-e). Extremely large storages of around 32,000 MWh are required to meet demand and
avoid excessive over-generation. Even then it will be necessary to retain fast response
thermal generation to avoid massive storages. In this scenario, 2,500 MW of thennal
generation is retained, although it is only operational in the winter / early spring and
generates only 1,600 GWh per year (less than 5% of demand). In order to maintain storages
at economically viable levels, the amount of wind energy is somewhat greater than necessary
for energy balance, leading to about 4,000 GWh of curtailment of wind energy in the early
and latter parts of the year. At such high penetration of renewables the challenge will be to
optimise the quantity of rapid response thermal generation and network scale storage.
Assuming that the thermal generation can be supplied by renewable sources, this scenario
represents a fully fossil free electricity system served by 90% private solar and storage, 6%
by large scale wind and 4% by thermal generation.
Technologies
Large scale renewable energy
In the foregoing it is assumed that wind will be the future source of renewable energy at
network scale. Various combinations of wind, large scale solar and wave energy were tested
using the model.
Figure 15 illustrates the assumed capacity factors for each resource across the year, depicted
as a fraction of their annual hourly average. This illustrates that while performance of both
solar and wind decline during the middle months of the year, wave energy is at its highest.
g
Fd
B
=
fe}
R
100%
% of average hourly capacity factor
oo
g
&
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Wind Solar Wave
Figure 15 Capacity factors of renewable energy sources
12
However the diumal performance of each source is also important. In order to understand the
impact of each it is useful to see the performance of each source at the most important time,
when residual network demands are high at the end of the day (i.e. when the influence of
onsite solar is at its lowest). Figures 16a) and b) illustrate the capacity factors of wind, solar
and wave for typical January and July aftemoons and evenings in 2050°. In the summertime
evenings, wind energy is increasing and operating at about 150% of its average annual
performance, while wave energy is operating at about its average annual level. Although at its
most powerful early in the day, solar is declining rapidly during this period. In the
wintertime, wind enerpy is relatively constant across the day but only operating at about 80%
of its annual potential. On the other hand wave is operating at about 150% of its average
annual level.
For these reasons, meeting the residual network loads using large scale solar PV requires
double the amount of storage as is required for a wind resource that is available throughout
the day and night. However wave energy could be a valuable complement to wind energy, as
its constancy is very high and its performance during the winter is superior to other sources
(Hughes & Heap, 2010).
Other sources that could be economically feasible by 2050 such as concentrated solar, which
can potentially incorporate storage, have not been considered in this study.
-
ain
NS
88
88
5 By 0
£55 -2,000
zB _ 4,000 2
1s 6,000
a | 8,000
05 — 2 | 10,000
0 - -12,000
12 13 14 15 16 17 18 19 20 21 22 23 24
Hour of the day
—Wind capacity factor |= ——Solar farm capacity factor
—Wave capacity factor «mmmNetwork load
Figure 16a) 2050 January day - network load vs capacity factors of renewable energy sources
* In the absence of more detailed hourly data, capacity factors for wave energy have been assumed to be
constant across the day of each month, but vary from month to month.
13
ee P 3/000
p15 ee
S / umes - 2,000
€ =
Bl 1,000 Fa
&
& 0
§ 05
1,000
0 2,000
12 13 14 15 16 17 18 19 20 21 22 23 24
Hour of the day
—Wind capacity factor | ——Solar farm capacity factor
—Wave capacity factor mmNetwork load
Figure 16b) 2050 July day - network load vs capacity factors of renewable energy sources
Thermal generation
Tt is assumed here that by 2050 the 2,500 MW of thermal generation is provided by open
cycle gas turbines, which are prevalent as peaking plants in the current SWIS, fuelled by
natural gas. A recent study (Rose, 2016) proposes state of the art ‘aero-derivative’ turbines
with the capability to run on bio-diesel derived from oil mallee. Another promising source is
biogas derived from anaerobic digestion of organic wastes. By 2050 Perth households will be
producing around 4.5m tonnes of organic wastes(Zhang, Su, Baeyens, & Tan, 2014) , which
can produce enough biogas to generate around 1,000 GWh of electricity, which is about 60%
of the projected thermal energy requirement of the SWIS. If organic wastes from wastewater
treatment plants and animal manure are added, it is likely that biogas could provide most or
all of the feedstock necessary for thermal generation, simultaneously reducing greenhouse
gas emissions from the natural decomposition of organics in the environment.
Network energy storage
Storage at the network scale could potentially be pumped hydro as assumed in a recent report
on renewables in the National Electricity Market (Blakers, 2017) and in a report by local
researchers (Rose, 2016). The state has potential sites for pumped hydro storage between
cliff-top ponds and the ocean, and the now under-utilised dams that previously contributed
most of Perth's water supply, could also play a part. It is estimated that approximately 40% of
the capacity of just five of the water supply dams adjacent to the metropolitan area could
provide the 32,000 MWh required by 2050. This would save a considerable portion of the
capital costs of completely new systems.
Summary
Two complementary system dynamics models were used to analyse the performance of
renewable energy sources on the SWIS:
e¢ amonthly model that rms to 2050 which projects demand, and the take-up of private
solar and energy storage at household and business premises; and
e an hourly model that operates across a given year to ascertain the performance of the
projected combinations of private solar and storage with large scale renewable energy
sources and energy storage.
14
The analysis suggests that in 2020 private systems will likely produce only about 12% of
energy demand. However by 2050 some 60% of households and 50% of commercial
premises could have solar PV systems, with around 80-90% of these premises having battery
storage systems. This amounts to some 21,000 MW of nameplate solar capacity and 32,000
MWh of storage. This private generation is sufficient to produce around 85- 90% of projected
electricity demands from the SWIS area This scenario would completely change the
historical arrangements under which electricity is predominantly provided though a
centralised network. The future network will be mainly associated with storing privately
generated solar energy and providing complementary energy sources to match supply with
demand.
Given the strong momentum for private solar, it is questionable whether in the interim period
it is necessary or indeed economically viable to introduce large scale renewable energy to the
network with the objective of eliminating greenhouse gas emissions from electricity. It may
be more appropriate to allow private generation to largely meet this objective while planning
fora completely renewable system by 2050.
By 2050 it is projected that only around 2,250 MW of large scale renewable energy will be
needed to complement private solar PV. This energy could be wind energy, or a combination
of wind and wave energy. Large scale solar energy is obviously also viable as a generation
source, although will require significantly larger energy storages to balance supply and
demand.
In order to avoid very large storages it will be necessary to retain fast response thermal
generation, most likely using state of the art open cycle gas turbines fuelled by renewable
sources such as biogas from organic wastes.
The SWIS will likely require around 32,000 MWh to complement the private storage by
2050, with pumped hydro utilising Perth's water supply dams, a potential source.
References
Australian Energy Market Operator. (2016). Deferred 2015 Electricity Statement of
Opportunities for the WEM: Australian Energy Market Operator.
Blakers, A., Lu, B., Stocks, M. (2017). 100% renewahle electricity in Australia: Australian
National University.
Climate Change Authority. (2015). Comparing countries’ emissions Targets - a practical
guide.
Climate Council of Australia. (2015). Battery storage for Renewable Energy and Electric
Cars.
Feldman, E., Boff, D., Margolis,R. (2016). Q2/Q3 2016 US solar industry update.
Grace, W. (2015). Exploring the Death Spiral: A system dynamics model of the electricity
network in Western Australia. Paper presented at the 33rd Intemational Conference of
the System Dynamics Society, Cambridge, Massachusetts, USA.
Hughes, M. G., & Heap, A. D. (2010). National-scale wave energy resource assessment for
Australia. Renewable Energy, 35(8), 1783-1791. doi: 10.1016/.renene.2009.11.001
Jones, B. (2012). Study on the impact of Photovoltaic (PV) generation on peak demand.
Perth: Westem Power.
Rose, B. (2016). Modelling Renewable Energy Scenarios for the South West Integrated
System. In Sustainable Energy Now (Ed.).
15
Zhang, C., Su, H., Baeyens, J., & Tan, T. (2014). Reviewing the anaerobic digestion of food
waste for biogas production. Renewable and Sustainable Energy Reviews, 38, 383-
392. doi: 10.1016/j.rser.2014.05.038
16
2 VV ARR AAI
sie [AP AIS ENN
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
—Total onsite demand —Total transmission network loads
Figure 11a) 2020 Onsite demand, network loads (typical day)
1,400
1 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
= Total OS demand met by solar mOSS charge Total excess OS generation
Figure 11b) 2020 Performance of onsite solar PV systems (typical day)
mm Exported solar PV lm Demand met by wind generation
Total OS demand met by solar “am Onsite battery discharge
—Total onsite demand
mm Thermal generation
‘lt Network storage discharge
Figure 11c) 2020 System supply (typical day)
ea ete hee ek ea ta eet he ht ee
12345 67 8 9 101112 13141516 171819 2021222324
» Total OS demand met by solar —_™ Onsite battery discharge
@ Exported solar PV
|= Demand met by wind generation
Thermal generation
™ Network storage discharge
12345 67 8 9 10111213 14151617 1819 20 21222324
© Total OS demand met by solar ™ Onsite battery discharge
mw Exported solar PV
= Demand met by wind generation
@ Thermal generation
| Network storage discharge
Figure 11e) 2020 System supply (typical July day)
Figure 11d) 2020 System supply (typical January day)
18
[%
7a
|
W
|
—
—_—,
na/\
n/N
nN
AA | A] AY A NA MA-
4 [D>
Z|
iN >
a
a
2
CT
SS
(+5
Ss
a
= 2
=
Jun Jul Aug Sep Oct Nov Dec
May
—Total onsite demand
Feb Mar
Jan
Total transmission network loads
Figure 12a) 2030 Onsite demand, network loads (typical day)
@ Total excess OS generation
= Total OS demand met by solar OSS charge
Figure 12b) 2030 Performance of onsite solar PV systems (typical day)
19
@m Hourly generation
mmm Exported solar PV
“Total OS demand met by solar mam Onsite battery discharge
—Total onsite demand
lm Thermal generation
| Network storage discharge
Figure 12c) 2030 System supply (typical day)
12345 6 7 8 9 101112131415 1617 1819202122 2324
= Total OS demand met by solar = Onsite battery discharge
m Exported solar PV
Hourly generation
m Thermal generation
m= Network storage discharge
5,000
4,000
3,000
MA
2,000
1,000
123.45 6 7 8 9 101112131415 16171819 2021222324
= Total OS demand met by solar ® Onsite battery discharge
® Hourly generation
@ Exported solar PV
™ Thermal generation
lm Network storage discharge
Figure 12e) 2030 System supply (typical July day)
Figure 12d) 2030 System supply (typical January day)
20
{SX INA
NN ON AN NY ONY ONY
TALIA Y TY TY \E
i yd
———_|
—
Feb
Apr May Jun Jul Aug Sep
—Total onsite demand —Total transmission network loads
Figure 13a) 2040 Onsite demand, network loads (typical day)
12,000
10,000
8,000 -
Feb
Mar
Apr May Jun Jul Aug Sep
= Total OS demand met by solar mOSS charge Total excess OS generation
Figure 13b) 2040 Performance of onsite solar PV systems (typical day)
Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Total OS demand met by solar “am Onsite battery discharge lm Demand met by wind generation
Feb
mm Exported solar PV
—Total onsite demand
mm Thermal generation
‘lt Network storage discharge
Figure 13c) 2040 System supply (typical day)
SVP FR LT
12345 67 8 9 101112 13141516 17181920 2122 23 24
Total OS demand met by solar _—_™ Onsite battery discharge
m= Demand met by wind generation
™ Exported solar PV
mw Thermal generation
Figure 13e) 2040 System supply (typical July day)
m Network storage discharge
12345 67 8 9 10111213 14151617 1819 20 21222324
© Total OS demand met by solar ™ Onsite battery discharge
Exported solar PV
= Demand met by wind generation
@ Thermal generation
@ Network storage discharge
Figure 13d) 2040 System supply (typical January day)
22
oO
Jan Feb Mar Apr Jun Jul
Oct
Figure 13f) 2040 Network storage (typical day)
8,000
6,000
4,000
2,000
(¢)
Mw
-4,000
e———
|
-2,000 |
: |
|
|
ee
<
-6,000
-8,000
-10,000
Dec
-12,000
Jan
Jul Aug Sep
——Total transmission network loads
Apr May Jun
——Total onsite demand
Feb
Oct
Figure 14a) 2050 Onsite demand, network loads (typical day)
23
= Total OS demand met by solar OSS charge Total excess OS generation
Figure 14b) 2050 Performance of onsite solar PV systems (typical day)
Mar Apr May Jun Jul Aug Sep
Feb
‘Total OS demand met by solar mami Onsite battery discharge
mm Demand met by wind generation
mmm Exported solar PV
—Total onsite demand
mm Thermal generation
‘lt Network storage discharge
Figure 14c) 2050 System supply (typical day)
24
(0) I ae a
12345678
T T
© Total OS demand met by solar
@ Exported solar PV
®@ Network storage discharge
LS a
T T ert
9 10111213 14151617 1819 20 21222324
®™ Onsite battery discharge
‘| Demand met by wind generation
@ Thermal generation
oO ica Re a a Ca a
12345678
Total OS demand met by solar
@ Exported solar PV
m Network storage discharge
Tar trad
T Tr Tar
9 101112 13141516 17181920 21222324
T el
m= Onsite battery discharge
| Demand met by wind generation
Thermal generation
Figure 14d) 2050 System supply (typical January day)
Figure 14e) 2050 System supply (typical July day)
35,000
ia
30,000 f\
woo | _{\l_}\
\
fi
|
g 20,000 \
= 15,000 -\
10,000
5,000 Vv
(a)
Jan Feb
Mar May
Oct
Figure 14f) 2050 Network storage (typical day)
Attachment A
Stock and flow structure of hourly model (network portion only)
<tine> Wave lookup
Wind generation LS wind lookup
Wind generation Weve generation
<ime
i at as
<Days pernprth>
<Days pernprth>
‘Amal generation, Cumiative
cntailed Houly OCCT JOCGT gereatin|
Contribution of LS load
exports to load
LS eports> 1S inports>
26
