Вы находитесь на странице: 1из 10

The Electricity Journal 32 (2019) 15–24

Contents lists available at ScienceDirect

The Electricity Journal


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

Efficient pathways to reduce carbon emissions in the electric sector T


Alan D. Lamont
Engineering Lawrence Livermore National Laboratory (ret), 13160 Skyline Blvd, Oakland, CA, 94619, United States

ARTICLE INFO ABSTRACT

Keywords: The study develops a 2-dimensional map of the feasible design space for an electric generation system. Cost and
Electric generation CO2 emissions contours are plotted across the map. The principles of constrained optimization are used to
Efficient carbon reduction identify the efficient pathways eliminating emissions and the technology drivers that determine them. The CO2
Technologies for carbon reduction intensity of the baseload is shown to be the primary driver. Further, the need for large, seasonal storage depends
Energy storage
strongly on the level of baseload deployed.

1. Introduction particularly storage.


This analysis illustrates the underlying economic principles that
This paper takes the position that elimination of carbon emissions determine the efficient approaches to carbon reduction, and it shows
from the energy system is one of society’s most important goals for the the effects that the characteristics of the technologies have on the ef-
foreseeable future. It is important to understand the factors that de- ficient paths.
termine cost effective ways to eliminate carbon emissions from electric
generation, since a large portion of emissions comes from electric 2. Literature review
generation, and electric generation will play a large role in replacing
fossil based transportation fuels. Many studies have explored ways that the future electricity system
Future electric generation systems will primarily rely on non-carbon could be structured to reduce or eliminate carbon emissions. The results
emitting generators. These include intermittent renewables (wind, found in any given analysis depend on its objectives, its mathematical
solar, waves, …), and non-carbon baseload technologies (nuclear, fossil approach, and constraining assumptions. Existing studies roughly fall
with carbon capture and sequestration, geothermal). However, these into three classes depending on whether or not they are optimizing, and
non-carbon technologies are not sufficiently flexible to follow the load depending on the a priori assumptions about the degree of carbon re-
from hour to hour. Intermittent generation depends on the time of day duction sought and the deployment of technologies, particularly non-
and the weather. Baseload generators are operated a fairly steady rate carbon baseload (nuclear and geothermal). Several studies are men-
for both economic and engineering reasons. These inflexible technolo- tioned briefly in this review. Later in the paper, the results and insights
gies need to be complemented by flexible generation such as natural gas obtained by these studies are compared to the results obtained in this
cycling generators, non-carbon cycling generation (eg. hydro) and by analysis.
storage. Flexibility can also be introduced through demand response One class finds plausible ways to structure the energy system to
and load shifting. reduce or eliminate carbon through renewables. These studies are not
There is, in fact, a wide range of ways that a system can be struc- optimizing and do not necessarily claim that they find the most efficient
tured to reliably meet the loads. The totality of all the ways that a ways to eliminate carbon. They are generally structured for finding
system can be structured constitutes the “design space” for the electric pathways to reducing carbon to a specified level, and/or for deploying
generation system. This paper uses a “ternary diagram” to provide a renewables to a specified level. Although they are not optimizing, they
two-dimensional map of the design space. The map can show the provide an understanding of the operational and cost issues that will
contours of the carbon emissions and system costs across the space. arise with high renewable systems and suggest approaches to dealing
Using the insights from economic optimization, these contours identify with them. Some examples include: Becker et al (2014), Hart and
the most cost efficient pathways to eliminating carbon. More important, Jacobson, (2012), Williams et al, (2012); E3 (2014); Shaner et al
the map shows the way that these pathways change depending on the (2018); Converse (2012), and Weitemeyer et al. (2015). These studies
nature of the technologies, and changes in their costs. We can also generally find that a system fully driven by renewables is feasible. They
determine the need for, and feasibility of, other technologies, often find that some form of large storage and/or load shifting is

E-mail address: alandlamont@earthlink.net.

https://doi.org/10.1016/j.tej.2019.02.005

Available online 28 February 2019


1040-6190/ © 2019 Elsevier Inc. All rights reserved.
A.D. Lamont The Electricity Journal 32 (2019) 15–24

required to meet the loads, at least at intermittent penetration above


roughly 50%.
Another class of studies employs optimization models to find least
cost structures for the system. However, this class assumes some sort of
constraint on baseload—either they assume it will be retired altogether,
or they assume that it will only be deployed at a low level. These
constraints may reflect the analyst’s view of what is technically or po-
litically feasible (e.g. deployment of nuclear). The assumption that
there will be little to no baseload drives the solution to some combi-
nation of intermittents and cycling generation (natural gas and hydro).
Examples of these studies include: Mai et al (2012); Haller et al (2012);
Wei et al (2012) investigates paths to emissions reductions of 80% from
the entire economy.
The final class of studies finds cost minimizing approaches to re-
ducing carbon while making very few assumptions about technology Fig. 1. Ternary map showing possible percentages of generation technologies
constraints. Examples include Jenkins and Thernstrom (2017); De for the electric system.
Sisternes et al (2016); Hirth (2015); McCollum et al (2012), and Nelson
et al (2012) Each of these finds that non-carbon baseload (nuclear and
Each point in the map specifies a mixture of primary energy. As an
geothermal) is part of the efficient mix for reducing carbon emissions
example, the highlighted point in Fig. 1 represents a system where 30%
of the energy is provided by baseload generators, 40% by cycling
3. Overview of the paper
generators, and 30% by intermittent generators. Note that the percen-
tages in the map refer to total generation including over-generation.
The following sections develop the ternary map and examine the
A model with hourly loads and hourly intermittent generation is
most efficient pathways to reducing carbon.
solved at each point in the map to determine the capacities and hourly

• Section 4 introduces the construction and use of a “ternary” map to generation required. The model’s equations specify that the load must
be met in every hour, and that the proportions of energy from each type
illustrate the design space.
• Section 5 explores the operations of the system across the map to of generator must match the specified proportions. Dispatch of tech-
nologies uses the following loading order: The intermittents and base
provide insight into the over- and under-generation across the map
are must-run and are dispatched first. Then cycling is dispatched up to
and the value of seasonal storage.
• Section 6 Illustrates the economic principles that determine the ef-
its capacity. Any additional loads are met from storage. The operation
of storage is discussed below. The model assumes that the transmission
ficient pathways to eliminating carbon. This section demonstrates
system is a copper plate and does not include transmission costs or
the impact of assumptions about the carbon intensity of the baseload
losses.
generation.
• Section 7 discusses the conclusions from the study.
This stage of the analysis simply finds the feasible solutions at each
point in the map. It is not concerned with the costs of the technologies.
Economics and emissions are introduced at a later stage to illustrate the
4. Representation of the electric generation design space and the
way that costs and emissions change across the design space, depending
system model
on our assumptions about technology costs and emissions rates.
The hourly load data and the intermittent generation are based on
An electric generating system is designed to meet a pattern of loads
data from California, CAISO (2011), so that loads and intermittent
over the year using a set of generators with different operational
generation are realistically correlated. However, this is not literally a
characteristics. This analysis recognizes that there are three basic
model of the California electricity system. For example, it does not
classes of generators:
include imports and exports, both of which are large components of the

• Baseload: These run at full capacity all the time and typically are not California electricity supply.
The pattern of intermittent generation depends on the proportion of
cycled. These include nuclear, coal, gas, geothermal, fossil with
wind and solar generation in the system, so the proportions of wind and
carbon capture and sequestration (CCS), biomass
• Cycling: These can be started and stopped, and ramped up and down
solar generation must be specified ahead of time. In the analyses below
the intermittent generation is assumed to be 50% wind and 50% solar.
to meet the varying load. Principally these are natural gas turbines
Sensitivity cases were run with 100% wind and with 100% solar. The
and hydro.
• Intermittent: wind, solar, wave, tidal. These are all carbon free. They
case of 100% wind produced results similar to those shown here. The
case of 100% solar showed costs substantially higher than these results.
operate on their own schedule that does not necessarily match the
The economic and technology parameters of the model and their
load (although solar does).
sources are given in Appendix B. The primary results shown here are
developed using the set of parameters for fuel costs, and capital costs of
A combination of these three types of generators will provide 100%
technologies shown in the appendix.
of the primary energy. When 100% of something is provided by three
sources, the results can be plotted in a 2-dimensional “ternary” dia-
gram. 4.2. Implementation of storage in the model

4.1. Ternary map and system model At higher levels of intermittents and lower levels of cycling gen-
eration there are hours with over- or under-generation. In that case, the
The ternary diagram provides a graphical map of the design space model adds storage to the system to capture excess generation and use it
for the electric generating system—all of the basic combinations of to cover the under-generation
ways that electric generators can be configured to meet the electric Storage charges when the sum of baseload plus intermittents ex-
load. Fig. 1 illustrates a ternary diagram, while Appendix A illustrates ceeds the load and discharges when the capacities of the generators are
the derivation of the diagram. not sufficient to cover the load. The model is constrained so that storage

16
A.D. Lamont The Electricity Journal 32 (2019) 15–24

does not go below zero and that the level of storage at the start and end there is little to no over- or under- generation. The upper boundary of
of the year are equal. In the runs shown here, energy storage capacity is this region in indicated in Fig. 3A by a dotted line. Note that the exact
allowed to be as large as needed to cover all of the under-generation location of this boundary is nominal. It is drawn here along the contour
and to absorb nearly all the excess generation. where the over generation is approximately 1% of the total load.
The sections below show that under a broad range of conditions Seasonal storage is important once we are outside of the balanced
seasonal storage is needed for a practical system. To optimize a system, region. Fig. 4 shows the energy storage capacity used across the map.
the marginal value of a technology should be deployed such that its Within the balanced region the level of storage required is essentially
marginal value is equal to its marginal cost. The marginal value of zero. Outside of the balanced region the requirement for storage in-
energy storage is a function of the number of times it fully cycles during creases rapidly.
a year and the price differential between the marginal charging and Although this analysis uses storage whenever there is under- or
discharging prices each time it cycles (Lamont (2013)). over-generation, it is possible to structure the system to operate with
In this case, seasonal storage is required. Since seasonal storage is limited storage outside of the Balanced Region. When storage is limited,
only cycled once a year, the cost per unit storage must be low to be the model scales up the capacities to cover the loads in all hours. This
economically viable. This analysis assumes ammonia as a storage can meet the loads, but at the cost of high capacity investment and
medium. Ammonia has the advantage that it is liquid at room tem- substantial over-generation.
perature and fairly manageable pressures (about 128 psi) so it is easily The finding that large levels of storage are needed as the system
stored in inexpensive pressurized tanks. The cost per unit energy stored moves to larger shares of intermittents is consistent with other studies.
is very low, about 13¢/kWh, Bartels and Pate (2008). It can be trans- Hart and Jacobson (2012) study the CO2 reduction that results from
ported easily using existing pipelines. There is considerable experience several different scenarios of high renewable penetration. Their ana-
in storing and transporting ammonia, Ahlgren (2012). lysis finds that up to 20,000 MW h storage (about 8% of annual load) is
Ammonia is currently produced using the Haber-Bosch synthesis needed for California. Jenkins and Thernstrom (2017) survey several
process, which is relatively capital intensive and has an energy effi- studies on high renewable systems that find that seasonal storage is
ciency of 0.5, Wang et al. (2018). In this analysis a newer process needed for a workable, reliable system with high intermittents. Haller
termed the Solid State Ammonia Synthesis process (SSAS) (also termed et al (2012) modeled scenarios to decarbonize Europe and the Medi-
the electro-catalyst process), is assumed. This technology is still in de- terranean region. They assume that nuclear and coal will be retired.
velopment, but it promises substantially lower capital cost and higher Their analysis finds that storage is needed when renewables reach 20 to
efficiency (around 0.71), Bartels and Pate (2008); Leighty (2011); Wang 50% of generation. The maximum stored energy reaches 10% to 15% of
et al. (2018). Assuming the SSAS process for storage, rather than Haber- the total annual energy. Converse (2012) examines the need for sea-
Bosch, lowers the cost of over- and under-generation and provides more sonal storage over the entire U. S. The results indicate that seasonal
optimistic results for implementing highly intermittent systems. storage needs to be 5% to 27% of annual consumption, depending on
The ammonia is converted to electricity using a turbine assumed to the mix of intermittents. Weitemeyer et al. (2015) evaluates storage
be similar in cost to a combined cycle natural gas turbine with a requirements in Germany with different levels of renewables with no
somewhat lower efficiency (about 40% is assumed here). This yields a baseload. Storage is needed at about 50% penetration of wind and solar.
round-trip efficiency of 28%. This is in line with current research, Shaner et al (2018) studies mixes of wind and solar needed to meet up
Brown (2017); Michie et al. (2010). Although such turbines have not to 80% of electric demands. They do not include baseload. The results
been used commercially, they have been demonstrated in research indicate that ≈32 days of storage are required. Becker et al (2014)
settings, Uenami, (2017), and Japan for Sustainability (2016). evaluate the seasonal storage requirements across the U.S. for systems
that are 100% renewable but with different mixtures of wind and solar.
5. The drivers of costs and emissions across the map The storage requirement varies substantially with the proportion of
wind and solar. For the wind/solar mix that minimizes the storage re-
This section reviews the changes in structure and operations over quirement, the required storage is around two to three months of the
the map to better understand the drivers of costs and emissions. annual load.

5.1. Seasonal operations and requirements for storage at selected locations 5.3. Cost contours across the map
across the map
Up to this point, the discussion has not included cost considerations.
Fig. 2 shows the hourly generation for several days during winter The capacities and the dispatch of different components needed to
and summer at different levels of intermittent penetration. It also shows make the system feasible and function reliably are determined by the
the levels of storage over the year for each example. patterns of generation and loads, and by the efficiencies of the storage.
Panel A shows the generation with 30% intermittent generation, This section addresses the costs and cost drivers across the map.
40% base, and 30% cycling. There is very little to no over- or under- In the cases below, the baseload is modeled with different ratios of
generation and essentially no storage is required. Panel B shows a case coal and nuclear. The proportions of baseload coal and nuclear gen-
with 50% intermittents. Panel C shows the fully intermittent case. eration affect the shapes of both the cost and the emissions contours.
At higher penetration of intermittents there is a large seasonal im- The discussion considers three cases with 50% coal / 50% nuclear, 25%
balance, requiring a large seasonal storage capacity. In the100% in- coal / 75% nuclear, and 100% nuclear. The operations of the generators
termittent case, the maximum energy storage capacity requirement is and storage are not dependent on the coal/nuclear mix—baseload is
16% of the total annual energy load. In the case of 50% intermittents baseload. However, the observations about costs and emissions do de-
and 30% base the maximum energy in the storage amounts to about 2% pend on these assumptions.
of the total annual generation. However, when the level of intermittents Fig. 5 shows the generation cost contours for the three cases. The
decreases to 30%, essentially no storage is required. costs are presented in terms of cost per MWh of load. In Fig. 5A
(baseload is 50% coal and 50% nuclear) the minimum cost system is
5.2. Storage and over- and under-generation across the map similar to conventional systems in place today. It uses about 60%
baseload energy and the rest is cycling. No intermittents are used.
Fig. 3 shows the total over-generation (panel A) and under-gen- Fig. 5B assumes the baseload is 25% coal and 75% nuclear. Under the
eration (panel B) across the map. We note that there is a broad, roughly assumptions of this analysis, nuclear generation has a relatively high
triangular region in the lower left where the system is balanced and capital cost. Using such a large component of nuclear generation means

17
A.D. Lamont The Electricity Journal 32 (2019) 15–24

Fig. 2. Generation in winter an summer and the corresponding requirements for storage.

that the minimum cost system is in the lower left corner—a system that imposes capital costs directly and indirectly. Direct costs include the
is entirely based on natural gas generation. This is true in spite of the cost of the storage system itself. Indirect costs result from the fact that
fact that this analysis assumes a relatively high price for natural gas the imperfect efficiency of storage requires significant additional in-
($10/mmBTU). The final case, Fig. 5C, uses 100% nuclear baseload. vestment in generation capacity to provide enough energy to meet the
Again, the minimum cost system is all cycling technology. loads. We can lower the indirect costs of storage by using a more effi-
For all of these cases the costs within the balanced region are re- cient technology (eg. batteries), but the direct costs would increase
latively low and change very little over a broad range of system designs. substantially.
Outside of the balanced region, costs increase quickly as the system
moves to higher levels of intermittent generation or reduces the cycling
generation. The costs increase due to the need for storage. Storage

Fig. 2. (continued)

18
A.D. Lamont The Electricity Journal 32 (2019) 15–24

Fig. 3. Total over generation and under generation (fraction of total annual load). Dashed line in Panel A is the nominal upper boundary of the balanced region.

5.4. Emission contours across the map

Along with the cost contours, the shape and orientation of the
emission contours determine the efficient path to follow to reduce
carbon emissions. The emissions contours depend very strongly on the
carbon intensity of the base load generation. The three cases are shown
in Fig. 6.
Fig. 6A shows the contours for high carbon intensity baseload—50%
coal and 50% nuclear. The contours slope up to the right. Carbon
emissions can only be reduced by moving up and to the left in the map
to reduce baseload generation.
Fig. 4. Seasonal storage energy capacity used (fraction of total annual load). As the carbon intensity of the baseload is reduced, the emissions
contours rotate in a clock-wise direction. Fig. 6B shows the case where
the baseload is assumed to be 25% coal and 75% nuclear. In this case
carbon is reduced by moving up and to the right, reducing cycling
generation.

Fig. 5. Levelized costs of energy ($/MWh) across the map for different baseload carbon intensities.

19
A.D. Lamont The Electricity Journal 32 (2019) 15–24

Fig. 6. Emissions contours for baseload that is 50%, 75%, and 100% nuclear (Kg CO2/MWh).

If the baseload is entirely non-carbon (e.g. nuclear, geothermal, or nuclear. As noted above, the minimum cost system configuration
fossil with CCS), the emissions are determined by the cycling genera- (without carbon constraints) is a conventional system where the bulk of
tion. This is shown in Fig. 6C. In this case the emissions contours slope the energy is provided by baseload and the rest by cycling. There is no
down to the right. Emissions are reduced by moving to the right side of intermittent generation. However, to reduce carbon efficiently the
the map. In fact, any system structure on the right-hand boundary of the system structure moves upward and to the left in the diagram, following
map has zero emissions. Consequently, we can reach zero emissions at the points of tangency. Along this pathway the system eliminates the
any level of intermittent generation. The question is what is the most coal baseload (along with the nuclear since in this example the baseload
efficient level of intermittent and base generation? is assumed to be 50% nuclear at all points). Eventually the path reaches
the left hand side of the map where baseload is eliminated and gen-
eration is provided by a combination of 55% cycling and 45% inter-
6. Conditions that determine the efficient pathway to reduce mittents. From this point on, reductions in carbon emissions require the
emissions reduction of cycling generation and an increase in intermittent gen-
eration. This is accomplished by increasing the storage capacity (see
To find the optimal pathway for carbon reduction, we essentially Fig. 4) to capture the increasing over-generation from the intermittents.
need to determine the optimal system structure needed to meet a re- In the case of 25% coal and 75% nuclear (Fig. 7B) baseload power
quired level of carbon emissions. As the required level of emissions is becomes more expensive, given the relatively high cost of nuclear as-
reduced, the sequence of system structures defines the efficient sumed in this analysis. The least cost system is entirely powered by
pathway. The mathematical conditions leading to a cost minimizing natural gas (lower left hand corner). To eliminate even a small amount
solutions are derived from Lagrangian optimization. These conditions of carbon, the system structure moves to a system with roughly 50%
optimize a function—in this case minimizing cost—subject to a con- baseload and 50% cycling. Further reduction in carbon initially moves
straint. In this case the constraint is that CO2 emissions equal a speci- the system up and to the right, following the points of tangency. In-
fied level. The basic theory can be found in a wide range of sources, for itially, this increases the intermittent generation to about 25% of total
example Chiang (1984) and Osborne (2016). The conditions for con- generation while reducing the cycling to about 30% of generation.
strained optimization are termed the Karush-Kuhn-Tucker conditions. Further reduction of carbon shifts the system toward the left hand
These state that the cost minimizing solution to meet a given emissions boundary, eliminating the baseload almost entirely. From then on the
constraint occurs at the point where the constraint contour is tangent to path is similar to the case with 50% coal and 50% nuclear—cycling
a cost contour. Or, if there is no point of tangency (ie. the emissions generation is replaced by intermittent generation through the increase
contour cuts across the cost contours), the minimum occurs at the in storage capacity.
boundary of the space. The cases with any coal baseload are similar in the end. Eliminating
When the cost contours and emissions contours are plotted together carbon requires a system with 100% intermittents. However, the case
on the map, the points of tangency and the optimal pathway are readily with 100% nuclear baseload, Fig. 7C, is quite different. We note that all
identified. Fig. 7 shows these pathways for the three different levels of of the carbon emissions come from the cycling generation. Eliminating
nuclear baseload and carbon intensity. carbon requires that the system move to the right, across the emissions
Fig. 7A illustrates the case with baseload that is 50% coal and 50%

20
A.D. Lamont The Electricity Journal 32 (2019) 15–24

Fig. 7. Efficient pathways follow the tangents between the cost and emissions contours. The cost values are from Fig. 5, and emissions values are from Fig. 6.

contours. Initially, the minimum cost system is 100% cycling, again due European system. Carbon is reduced by imposing a carbon tax in the
to the high cost of nuclear generation. However, reducing carbon model. The analysis finds that as the carbon price rises and carbon is
emissions moves the system in an arc across the lower part of the map. reduced, the share of wind initially increases and then decreases similar
Initially, the efficient path adds some intermittents, increases non- to Fig. 7C. As carbon is reduced, nuclear baseload increases.
carbon baseload, and decreases cycling. Midway through the path, the Nelson et al. (2012) evaluate cases to partially reduce carbon
share of intermittents begins to decline as the pathway approaches the emissions in WECC, down to a level of 54% of 1990 emissions. The
boundary of the balanced region. At the far right in the map, reliance on parameters of the analysis allow very little to no coal baseload. Base-
intermittents reduces to a small amount. load is provided by non-carbon technologies including nuclear, geo-
The pathway for the 100% nuclear case stays almost entirely in the thermal, and biomass. The solutions in this study also fall within the
balanced region. In this region there is little to no over- or under-gen- balanced region with 15–30% intermittents and 20–50% baseload.
eration and little storage is required. It is only at the extreme right, McCollum et al (2012) finds least cost pathways to reduce CO2 from
where cycling generation is completely eliminated that there is a sig- entire California energy system to reduce carbon emissions 90% below
nificant mismatch between generation and load requiring some storage. the 1990 levels. The solution found for the electric system in 2050 falls
It is reasonable to assume that if future systems include baseload close to the border of the Balanced Region with about 36% non-carbon
generation, it will be a combination of non-carbon technologies—pos- baseload (27% nuclear and 9% geothermal). Wind and solar account for
sibly fossil with carbon capture and sequestration (CCS), nuclear, and a little less than 50% of electric generation.
geothermal. For the rest of this discussion it is be assumed that the Wei et al (2012) find structures to reduce emissions by 80% from
baseload is nuclear. the entire energy system. The level of baseload generation is con-
Other results from the literature also find that the optimal path to strained and only non-carbon baseload is allowed in the solutions. The
reducing carbon (as opposed to promoting renewables) makes heavy solutions use 20 to 30% cycling with 12 to 36% baseload. These results
use of nuclear with a modest level of intermittent generation. stay within or close to the Balanced Region.
The key issue is the extent to which an analysis allows, or assumes,
the deployment of non-carbon baseload. As is noted earlier, many
studies make the a priori assumptions that baseload will not be used, 7. Conclusions and policy implications
driving the model solutions to the left hand side of the map. A few
studies include nuclear baseload and allow the model to determine the This study started from the premise that elimination of carbon from
optimal baseload capacity. the electric system is our primary goal. As a corollary, technologies
de Sisternes et al. (2016) allow nuclear baseload in the system. They should be chosen to the extent that they help meet that goal. The
find a efficient pathway similar to the path in Fig. 7C. question is how can we do this most efficiently. This analysis leads to
Hirth (2015) uses the EMMA optimizing model to evaluate the pe- two primary conclusions:
netration of intermittent generation technologies in the northern The efficient pathway to reducing carbon depends very strongly on the
carbon intensity of baseload generation. The carbon intensity determines

21
A.D. Lamont The Electricity Journal 32 (2019) 15–24

the orientation of the carbon emissions contours. If there is no non- ployment of new nuclear technologies. Second, society can start to
carbon baseload available, the efficient pathway moves toward the left implement of large, seasonal storage capacity. From the discussion
side of the map, eliminating baseload and relying on intermittent above, a high degree of nuclear deployment would eliminate the need
generation with natural gas cycling. With non-carbon baseload, the for large seasonal storage. But such a high level of nuclear capacity may
strategy is a larger reliance on baseload with moderate amounts of not occur due to social and political issues. Without a high level of
intermittents. nuclear deployment, the system will still rely to a significant extent on
Under and over generation are major issues in relying on inflexible intermittent generation with at least some over- and under generation,
generation. There is a Balanced Region in the design space of generation so seasonal storage will be needed.
systems that has little to no over- and under-generation and little need
for storage. Staying within, or close to, this region substantially sim-
plifies the operation of the electric system and reduces cost. Funding
These observations highlight the choices that society has in dec-
arbonizing the electric system. There are several technology choices This research did not receive any specific grant from funding
that can be made: first, society can promote the development and de- agencies in the public, commercial, or not-for-profit sectors.

Appendix A. Derivation of Ternary Diagram

Fig. A1

Fig. A1. Derivation of the ternary diagram.

Appendix B. Parameters of model

The analysis is done using the assumptions and parameters listed below.
System loads and intermittent generation
Loads and generation are taken from CAISO (2011). These data give California loads, wind generation and PV generation for the 33% renewable
penetration analysis for the year 2020. The original data is given in one minute time steps. The data were averaged over each hour for this study.
Generation technology data
The technology data is taken from the Energy Information Agency Annual Energy Outlook from EIA (2016a) This data is reproduced in Table A1,
below. The last column in the table, “Efficiency (computed from heat rate)”, was not part of the original data and is computed in this study.
Fuel costs
The following fuel costs were used:

22
A.D. Lamont The Electricity Journal 32 (2019) 15–24

Table A1
Costs and efficiencies of generation technologies from EIA (2016a).
Technology Nominal Capacity Heat Rate (Btu/ Overnight Capital Cost Fixed O&M Cost Variable O&M Cost Efficiency (computed from
(MW) kWh) ($/kW) ($/kW-yr) ($/MWh) heat rate)

Adv Nuclear 2,234 10,449 6,108 98.11 2.25 1.0a


Wind 100 9,541 1,644 49.98 0.00 na
Photovoltaic 150 9,541 2,480 21.33 0 na
Adv Comb. Cycle 429 6,200 1,080 9.78 1.96 0.55
Dual Unit Adv Coal PC 1,300 8,800 2,934 31.18 4.47 0.38747

a
The price of uranium fuel is given in terms of the cost per unit of electricity produced ($/kWhe). Consequently, this study uses an efficiency of 1.0 for the nuclear
generators.

Coal: 0.0058 $/kWh Price for 2005 based on EIA (2010)


Uranium: The Nuclear Energy Institute reports fuel costs of $0.0058 $/kWh to $0.0078$/kWh, Nuclear Energy Institute (2017). A study by MIT
(Yangbo and Parsons, 2009) gives a fuel cost of $0.00697/kWhe. The present study uses a value of $0.0075/kWhe. These prices are given in terms of
the cost per unit of electricity produced, so the efficiency of the nuclear generators is set to 1.0.
Natural gas: Was set at 10 $/mmBTU. This is at the high side of the EIA past data and future projections (EIA, 2016b).
Emissions factors
Emissions from the technologies were as follows:
Coal emissions: In pounds of carbon dioxide per million Btu, U.S. average factors are 227.4 for anthracite, 216.3 for lignite, 211.9 for sub-
bituminous coal, and 205.3 for bituminous coal, Hong and Slatick (1994). A value of 205.3 lbs/million Btu was used in here.
Natural gas emissions: Value for natural gas is 116.39 lbs CO2/MMBtu EIA (2004)
Ammonia production and storage costs
Efficiency of converting electricity to ammonia (Kwh of energy in ammonia per kWh of electricity) is calculated from Bartels and Pate (2008) as
0.71 for the Solid State Ammonia Synthesis (SSAS) process. This is similar to Wang et al. (2018).
The capital costs of the SSAS process is calculated from Bartels and Pate (2008) and from Leighty (2011) as $844.8/kWh NH3-h. That is $844 of
capital cost for a plant that can produce 1 kW h of NH3 energy per hour.
The capital cost of liquid ammonia storage is calculated from Bartels and Pate (2008) as $0.13/kWh NH3.

References Report, January-april 1994, DOE/EIA-0121(94/Q1) (Washington, DC, August 1994).


pp. 1–8. accessed 26Aug13. http://www.eia.gov/coal/production/quarterly/co2_
article/co2.html.
Ahlgren, William L., 2012. The Dual-Fuel Strategy: An Energy Transition Plan. Proc. IEEE Japan for Sustainability, 2016. Researchers Succeed in Generating Electricity Using
100 (November (11)), 2012. Ammonia-Fired Gas Turbine: One Step Closer to Hydrogen Power. accessed
Bartels, Jeffrey R., Pate, Micheal B., 2008. A Feasibility Study Of Implementing An 17May18. https://www.japanfs.org/en/news/archives/news_id035522.html.
Ammonia Economy. Iowa State University Dec. 2008. Jenkins, Jesse D., Thernstrom, Samuel, 2017. Deep Decarbonization of the Electric Power
Becker, Sarah, Frew, Bethany A., Andresen, Gorm B., Zeyer, Timo, Schramm, Stefan, Sector Insights From Recent Literature. Energy Innovation Reform Project March.
Greiner, Martin, Jacobson, Mark Z., 2014. Features of a fully renewable US electricity Lamont, Alan D., 2013. Assessing the economic value and optimal structure of large-scale
system: optimized mixes of wind and solar PV and transmission grid extensions. electricity storage. Ieee Trans. Power Syst. 28 (2) May 2013.
Energy 72 (2014), 443–458. Leighty, William C., 2011. Solid State Ammonia Synthesis (SSAS) Pilot Plant
Brown, Trevor, 2017. Round-Trip Efficiency of Ammonia as A Renewable Energy Demonstration System for Renewable Energy (RE) Firming Storage, Transmission,
Transportation Media. Ammonia Energy. October 20, 2017 http://www. and Export. Alaska Applied Sciences, Inc., pp. 2011.
ammoniaenergy.org/round-trip-efficiency-of-ammonia-as-a-renewable-energy- Mai, Trieu, Wiser, Ryan, Sandor, Debra, Brinkman, Gregory, Heath, Garvin, Denholm,
transportation-media/ accessed 17May18. Paul, Hostick, Donna J., Darghouth, Naim, Schlosser, Adam, Strzepek, Ken, 2012.
CAISO (California Independent System Operator), 2011. One-Minute Data for Load, Exploration of high-penetration renewable electricity futures, national renewable
Wind, and Solar 3/14/2011. accessed from. http://www.caiso.com/Pages/ energy laboratory/TP-6A20-52409. Renew. Electr. Futures Study 1. http://www.
documentsbygroup.aspx?GroupID=FDE2E73E-EE90-4B43-8168-467221BB808C. nrel.gov/analysis/re_futures/.NREL/TP-6A20-52409.
Chiang, Alpha C., 1984. Fundamental Methods of Mathematical Economics, McGraw-Hill McCollum, David, Yang, Christopher, Yeh, Sonia, Ogden, Joan, 2012. Deep greenhouse
International Editions. gas reduction scenarios for California strategic implications from the CA-TIMES en-
Converse, Alvin O., 2012. Seasonal energy storage in a renewable energy system. Proc. ergy-economic systems model. Energy Strategy Rev. 1, 19–32.
Ieee 100 (February (2)), 2012. Michie, Preston, Holbrook, John, Robertson, Jack, Field, Pam, 2016. Preliminary
de Sisternes, Fernando J., Jemkins, Jesse D., Botterud, Audun, 2016. The value of energy Feasibility Study of Hydrogen Hubs: An Energy Storage Device Using Anhydrous
storage in decarbonizing the electricity sector. Appl. Energy 175, 368–379. Ammonia and A Storage Medium, Version 2.0. The Northwest Hydrogen Alliance, pp.
E3 (Energy and Environmental Economics), 2014. Investigating a Higher Renewables 2010.
Portfolio Standard in California. . https://ethree.com/documents/E3_Final_RPS_ Nelson, James, Johnston, Josiah, Mileva, Ana, Fripp, Mattias, Hoffman, Isn, Petros-Godd,
Report_2014_01_06_with_appendices.pdf. Autumn, Blanco, Christian, Kammen, Daniel M., 2012. High resolution modeling of
EIA (Energy Information Agency), 2004. Unit Conversions Emissions Factors, And Other the western north american power system demonstrates low-cost and low carbon
Reference Data, November 2004. Downloaded from http://www.epa.gov/cpd/pdf/ futures. Energy Policy 43, 436–447.
brochure.pdf accessed 26Aug13. . Nuclear Energy Institute, 2017. Nuclear Costs in Context. Nuclear Energy Institute
EIA (Energy Information Agency), 2010. Annual Energy Outlook 2010, Report #:DOE/ August.
EIA-0383(2010), Release Date: May 11, 2010. Downloaded from http://www.eia. Osborne, Martin J., 2016. Mathematical Methods For Economic Theory. available at
doe.gov/oiaf/aeo/aeoref_tab.html, accessed 25Sep10. . http://mjo.osborne.economics.utoronto.ca/index.php/tutorial/index/1/men/t
EIA (Energy Information Agency), 2016a. Cost and Performance Characteristics of New Accessed May 29, 2016. .
Generating Technologies, Annual Energy Outlook 2016. Shaner, Matthew R., Davis, Steven J., Lewis, Nathan S., Caldeira, Ken, 2018. Geophysical
EIA (Energy Information Agency), 2016b. Annual Energy Outlook 2016 With Projections constraints on the reliability of solar and wind power in the United States. Energy
To 2040, DOE/EIA-0383 (2016). Environ. Sci. 4, 2018.
Haller, Markus, Ludig, Sylvie, Bauer, Nico, 2012. Decarbonization scenarios for the EU Uenami, Harumi, 2017. Ammonia-Fueled Gas Turbine Power Generation. Ammonia
and MENA power system: considering spatial distribution and short term dynamics of Energy News February 16.
renewable generation. Energy Policy 47 (2012), 282–290. Wang, Lu, Xia, Meikun, Wang, Hong, Huang, Kefeng, Qian, Chenxi, Maravelias, Christos
Hart, Elaine K., Jacobson, Mark Z., 2012. The Carbon Abatement Potential of High T., Ozin, Geoffrey A., 2018. Greening ammonia toward the solar ammonia refinery.
Penetration Intermittent Renewables. Energy Environ. Sci. 5, 6592–6601. Joule 2 (June (20)), 1–20. https://doi.org/10.1016/j.joule.2018.04.017.
Hirth, Lion, 2015. The optimal share of variable renewables: how the variability of wind Wei, Max, Nelson, James H., Ting, Michael, Yang, Christopher, Greenblatt, Jeff B.,
and solar power affects their welfare-optimal deployment. Energy J. 36 (1). McMahon, James E., Kammen, Daniel M., Jones, Chris, Mileva, Ana, Johnston,
Hong, B.D., Slatick, E.R., 1994. Carbon Dioxide Emission Factors for Coal, Note: This Article Josiah, Bharvirkar, Ranjit, 2012. California’s Carbon Challenge: Scenarios for
Was Originally Published in Energy Information Administration, Quarterly Coal Achieving 80% Emissions Reduction in 2050. Lawrence Berkeley National Laboratory

23
A.D. Lamont The Electricity Journal 32 (2019) 15–24

October 31. Energy and Environmental Policy May, 09-004.


Weitemeyer, Stefan, Kleinhansa, David, Vogta, Thomas, Agerta, Carsten, 2015.
Integration of renewable energy sources in future power systems: the role of storage. Dr. Lamont Is a retired researcher from Lawrence Livermore National Laboratory. While
Renew. Energy 75, 14–20. at LLNL he was engaged in studies of system economics. He developed that META-Net
Williams, James H., DeBenedictis, Andrew, Ghanadan, Rebecca, Mahone, Amber, Moore, economic modeling system to analyze long-term market equilibrium solutions for energy
Jack, Morrow III, William R., Price, Snuller, Torn, Margaret S., 2012. The technology systems. His work also included studies of the long-term economic value of intermittent
path to deep greenhouse gas emissions cuts by 2050: the pivotal role of electricity. generation and storage technologies.
Science 335, 53.
Yangbo, Du, Parsons, John E., 2009. Update on Cost of Nuclear Power. MIT Center for

24

Вам также может понравиться