Chapter 4: Electric Power Generation

INTRODUCTION

The low-carbon emissions and low capital cost

of natural gas generation compared to other
fossil fuel generation, combined with abundant
gas supplies and current relatively low prices,
make natural gas an attractive option in a
carbon-constrained environment, such as that
contemplated in the analysis in Chapter 3. In
addition to its increasingly important role as a
primary fuel for electricity generation, natural
gas will continue to perform a unique function
in the power sector by providing both baseload
power and the system flexibility that is required
to meet variation in power demand and supply
from intermittent sources.

Natural gas provides flexibility to the power system

largely through the three types of generation
technologies: highly efficient natural gas combined cycle (NGCC) units, steam turbines, and
gas turbines. Gas turbines are generally used to
meet peak demand levels and to handle weather,
time of day, seasonal and unexpected changes in
demand. NGCCs and steam turbines can act as
baseload or intermediate-load units, although
the majority of gas capacity in the U.S. now
operates in load-following (intermediate) or
peaking service.
Currently, natural gas is second only to coal in
total generation, fueling 23% of U.S. electricity
production. Natural gas, however, has the highest
percentage of nameplate1 generation capacity of
any fuel, at 41% compared to 31% for coal, which
is the next highest (Figure 4.1). This difference
between nameplate capacity and generation is

The focus of this chapter is on the role of natural

gas in helping to reduce CO2 emissions from the
power sector and the interaction of gas use with
projected growth in wind and solar generation.

Figure 4.1 % Nameplate Capacity Compared to % Net Generation, U.S., 2009*

50
44
41

Percentage (%)

40
30

30
23

20

20

10

<1

<1

0
Natural Gas

*Numbers are rounded

Source: MIT from EIA data

Petroleum

Coal

Hydroelectric

% Nameplate Capacity

Nuclear

Other
Renewables

Other

% Net Generation

Chapter 4: Electricity

73

BOX 4.1 MODELS EMPLOYED TO EXAMINE

THE U.S. ELECTRICITY SYSTEM
The MARKAL (MARKet ALlocation) model of
the U.S. electricity sector enables a granular
understanding of generation technologies,
time-of-day and seasonal variations in electricity demand and the underlying uncertainties
of demand. It was originally developed at
Brookhaven National Laboratory (L.D. Hamilton,
G. Goldstein, J.C. Lee, A. Manne, W. Marcuse,
S.C. Morris, and C-O Wene, MARKAL-MACRO:
An Overview, Brookhaven National Laboratory,
#48377, November 1992). The database for the
U.S. electric sector was developed by the
National Risk Management Laboratory of the
U.S. Environmental Protection Agency (EPA).
The Renewable Energy Deployment System
(ReEDS) model is used to project capacity
expansions of generation, incorporating transmission network impacts, associated reliability
considerations and dispatch of plants as operating reserves. It also captures the stochastic nature
of intermittent generation as well as temporal
and spatial correlations in the generation mix
and demand. It has been developed by the
National Renewable Energy Laboratory (NREL)
(J. Logan, P. Sullivan, W. Short, L. Bird, T.L. James,
M. R. Shah, Evaluating a Proposed 20% National
Renewable Portfolio Standard, 35 pp. NREL
Report No. TP-6A2-45161, 2009).
The Memphis model realistically simulates the
hourly operation of existing generation plants
in the presence of significant volumes of wind
and solar generation. It was developed by the
Institute for Research in Technology of Comillas
University (Madrid, Spain) for the Spanish
Electricity Transmission System Operator (Red
Elctrica de Espaa) to integrate renewable
energies. (A. Ramos, K. Dietrich, J.M. Latorre, L.
Olmos, I.J. Prez-Arriaga, Sequential Stochastic
Unit Commitment for Large-Scale Integration of
RES and Emerging Technologies, 20th International Symposium of Mathematical Programming (ISMP) Chicago, IL, USA, August 2009.
http://www.iit.upcomillas.es/~aramos/ROM.htm

74

MIT STUDY ON THE FUTURE OF NATURAL GAS

explained in part by the overbuilding of

NGCC units in the mid-1990s. It also shows
that NGCC units are operating well below
their optimum operating value. Finally, it
highlights the unique role of gas and steam
turbines, which in 2009 had an average
capacity factor of 10% (see Table 4.1). This
low-capacity factor illustrates the peaking
function of these units, particularly the gas
turbines, that are routinely used only to meet
peak demand levels and which, absent breakthroughs in storage, are essential for following
time-varying electricity demand and accommodating the intermittency associated with
wind and solar power.
Historically, because of its higher fuel price
compared with nuclear, coal and renewables,
natural gas has typically had the highest
marginal cost and has been dispatched after
other generation sources. Consequently, natural
gas has set the clearing price for electricity in
much of the country. Lower natural gas prices,
the opportunities created by abundant
relatively low-cost supplies of unconventional
shale gas, increased coal costs and impending
environmental regulations that will add to the
cost of coal generation are, however, changing
the role of gas in power generation.
The Emissions Prediction and Policy Analysis
(EPPA) model employed in Chapter 3 is
designed to study multi-sector, multi-region
effects of alternative policy and technology
assumptions, and as a result it only approximates the complexities of electric system
dispatch. In this chapter, we analyze in greater
depth two of the cases studied there, employing
a more detailed model of the electric sector
MARKAL (see Box 4.1). This model is also
used to further explore the implications of
uncertainty in fuel and technology choices as
they influence natural gas demand in this
sector, extending the uncertainty analysis in
Chapter 3 which considers only the uncertainty in gas resources.

Table 4.1 2009 Average Capacity Factors by Select Energy Source, U.S. (numbers rounded)
Coal

Petroleum

Natural
Gas CC

Natural
Gas Other

Nuclear

Hydroelectric
Conventional

64

42

10

90

40

Other
All Energy
Renewables Sources
34

45

Source: EIA, Table 5.2, Average Capacity Factors by Energy Source

This chapter then considers two questions

about gas use in U.S. power generation:
(1) What is the potential for reducing CO2 by
changing the current generation dispatch order2
to favor NGCC over coal generation? (2) What
will be the effect of increased penetration of
wind and solar generation on natural gas
power generation?
To answer the first question it is important to
understand NGCC utilization patterns. NGGC
units are designed to be operated at capacity
factors of up to around 85% rather than the
current national average of 42%. This suggests
possible opportunities for displacing some coal
with gas generation, thereby lowering CO2
emissions from the sector. We examine how
much of this capacity could actually be applied
to this purpose without diminishing system
reliability. An important by-product of such
a change, also analyzed, would be associated
reductions in criteria pollutant emissions.
To explore the second question, the interaction
between intermittent renewables and natural
gas use is analyzed from two viewpoints: one
in the short term when additional intermittent
capacity is introduced into a system with other
sources fixed; and the other in the longer term
when the overall supply structure has time to
adjust to growth in intermittent capacity. In
this regard, we note that, at a more granular
level than is presented in Table 4.1, wind
turbines have an average capacity factor of
27%, solar thermal, 19%, and solar PV, 14%,
and gas combustion turbines and steam
turbines (used to balance load) have average
capacity factors of 5% and 14%, respectively.3

Study of these two questions is approached

with the use of two additional electric sector
models, each designed to simulate the power
system and its operations in detail over a range
of conditions and timescales (see Box 4.1),
enabling the following analyses:
s !N EXAMINATION OF RELIABILITY AND TRANSMISsion constraints, which helps to isolate and
understand the total generation required at
points in time to meet demand for electricity
and maintain operating reserve capacity and
adequate installed capacity margins. We
employ ReEDS for this analysis, which uses
multiple time periods for any given year and
reports results by geographic regions.
s !N EXPLORATION OF ANNUAL SCENARIOS AT THE
hourly level, which takes into consideration
details of real-time problems, such as uncertainty and variability in demand and in
generation patterns for intermittent technologies, and start-up and shut-down
characteristics for plant cycling. Here we use
the Memphis model.

Chapter 4: Electricity

75

ELECTRICITY SYSTEM OVERVIEW

The electricity system is complex; this overview

of how the system works, including the regimes
under which power plants operate and the
hierarchy of decision-making that influences
the capacity and generation mixes, is intended
to enhance the understanding of the implications of the modeling and analysis discussed
later in the chapter.
Electricity is produced from diverse energy
sources, varied technologies and at all scales.
Sources for electric generation include a mix of
renewables (sun, wind, hydro resources, among
others), fossil fuels (oil, natural gas, coal) and
uranium. As such, the generation of electricity
comprises a variety of technologies with the
type of fuel being used, and characterized by a
wide range of investment and operating costs.
Conventional power plants are operated under
different regimes, mainly depending on their
variable operating costs and operating
flexibility.4
s Baseload plants are characterized by expensive
capital costs and low variable costs, and they
are operated most of the time during the year.
They tend to be inflexible plants as they
cannot easily change their operational level
over a wide usage.
s Peaking plants are characterized by low
capital costs and higher variable costs, and
they are operated a few hours per year when
the electric load is the highest. They can be
characterized as flexible plants because of
their quick operating response.
s Intermediate plants have variable costs that
fall in between those of peaking and baseload
technologies, and they are operated accordingly. They can be characterized as cycling
plants, i.e., plants that operate at varying levels
during the course of the day and perhaps
shut down during nights and weekends.

76

MIT STUDY ON THE FUTURE OF NATURAL GAS

The expansion planning and operation of

electric power systems involve several decisions
at different timescales, generally based on
economic efficiency and system reliability
criteria. This process has a hierarchical structure,
where the solutions adopted at higher levels are
passed on to the lower levels incorporating
technical or operational restrictions at that level:5
s Long-term decisions are part of a multi-year
process (3 years up to 10 or more years) that
involves investments in generation and
network required to expand the system.
s Medium-term decisions are taken once the
expansion decisions have been made. They
are part of an annual process (up to 3 years)
that determines the generation unit and grid
maintenance schedule, fuel procurement
and long-term hydro resource scheduling.
s Short-term decisions are a taken on a weekly
time frame. They determine the hourly
production of thermal and hydroelectric
plants for each day of the week (or month),
subject to availability of the plants and to
hydro production quotas determined at the
upper decision level, and considering not
only variable operating costs, but also the
technologys own technical characteristics
such as start-up and shut-down cost and
conditions, a plants technical minima and
ramping times. In addition, these short-term
decisions are subject to generating reserve
capacity needed to immediately respond to
unexpected events.
s Real-time decisions involve the actual operation of the system (seconds to minutes). They
involve the economic dispatch of generation
units, the control of frequency so that
production and demand are kept in balance
at all times, while maintaining the system
components within prescribed safe tolerances
of voltages and power flows, accounting also
for possible contingencies.

Finally, meeting reliably the consumption of

electric power at all times requires having both
adequate installed capacity and secure operation procedures. A reliable operation involves
using ancillary services at different levels,
maintaining sufficient capacity in reserve
(quick-start units, spinning reserves) and with
enough flexibility to respond to deviations in
the forecast of demand or intermittent generation, and to unexpected events, such as the
sudden loss of lines or generation plants.
THE ROLE OF GAS GENERATION
UNDER A CO2 LIMIT

The EPPA model simulations in Chapter 3

provide insights into both the economy-wide
use of natural gas and its market share in
electric power under various assumptions
about greenhouse gas (GHG) mitigation.
Application of the MARKAL model, with its
greater electric sector detail, provides a check
on the adequacy of the EPPA approximations
for the power sector. MARKAL considers a
more complete listing of the generation alternatives, and it addresses the variation in the
level of electricity demand, as a result of the
diurnal, weekly and seasonal cycles (which
EPPA only roughly approximates). This variation is important because different technologies
are needed to run different numbers of hours
per year a pattern that changes over years
with demand growth and new investment. Also,
the MARKAL model allows for a more complete exploration of uncertainty in gas use in
the power sector.

For consistency with the analysis in Chapter 3,

certain MARKAL inputs are taken from the
EPPA model results, including electricity
demand, supply curves for natural gas and coal
and the reference costs of generation technologies. Also, two of the same policy cases are
considered: Scenario A, which assumes no new
GHG policy; and Scenario B, which imposes
a Price-Based mitigation measure. For the
Price-Based case, a cap on CO2 emissions for
the electric sector in MARKAL is set based on
the results for that scenario in Chapter 3.
The underlying technology mix computed
by the more-detailed electric sector model can
be illustrated by annual load duration curves,
which show the mix of generation dispatched
at different times to meet changes in the level
of electricity demand in the contiguous U.S.
electric system over the course of a year. These
curves for the year 2030, with and without
a policy of carbon constraints, are shown in
Figure 4.2. In the absence of a carbon policy
(Panel a), generation from hydro, coal and
nuclear occur at all times of the year while
generation from wind and hydro are supplied
whenever they are available.6
Without a carbon policy (Panel a), natural gas
generation from combined cycle and steam
turbines occurs for less than half of the time
over the course of the year during periods of
higher demand; and natural gas combustion
turbines are used for only a few hours per year
at the peak demand hours.
Under the carbon price policy (Panel b), NGCC
technology largely substitutes for coal to
provide baseload generation along with nuclear
generation.

Chapter 4: Electricity

77

Figure 4.2 Time blocks approximation to the Load Duration Curve for the (a) No Policy
and (b) 50% Carbon Reduction Policy Scenarios in 2030. Three seasons have been considered: summer, winter and spring/autumn. Within each season, there are four blocks:
peak time, daytime PM, daytime AM, and nighttime, as shown in the graphs. The peak
time block is very narrow.
4.2a
1000

1200

Solar

950

Oil

900

1000

Landfill

850

Geothermal

800

Diesel

750

800

Biomass

GW

700

Municipal Waste
Gas Steam

600

Gas Combustion Turbine

Gas Combined Cycle

400

Wind
Coal Steam
Hydro

200

Nuclear
0

2000

Summer

4000

Hours

6000

Winter

8000

Spring & Autumn

4.2b
1000

800
750

800

700
650

600

GW

600

400

200

2000

Summer
Source: MIT analysis

78

MIT STUDY ON THE FUTURE OF NATURAL GAS

4000

Hours

Winter

6000

8000

Spring & Autumn

The change over time in the energy mix in the

electric sector is shown in Figure 4.3 for both
the No Policy and the Price-Based cases. In the
No Policy case, under reference assumptions for
fuel prices, electricity demand and technology
costs and mean gas resources these results
show the same pattern of increasing gas use as
the simulation studies in Chapter 3. The gas use
in this sector in 2025 is essentially the same in
the two studies. Toward the end of the simulation period, MARKAL projects one-quarter to
one-third more gas-based generation than
EPPA, though gas generation is still small
relative to coal.
Under the Price-Based policy the overall
pattern of change remains the same as in EPPA:
coal is forced out and replaced by gas. In the
period to 2025 MARKAL projects a more rapid
phase-out of coal than does EPPA, in part

because MARKAL is a forward-looking model

and sees higher prices in the future whereas the
recursive dynamic (myopic) EPPA model does
not. Farther out in time coal is no longer in the
mix, and under a continuously tightening CO2
constraint conventional gas generation begins
to be replaced by non-carbon generation sources
such as nuclear, renewables and/or coal or gas
with carbon capture and sequestration (CCS).
The EPPA model expands nuclear generation
whereas MARKAL introduces natural gas with
CCS, yielding about a one-quarter greater level
of gas use. The outlook for gas in this sector is
consistently positive across the two studies, and
the difference in details of load dispatch is to be
expected for models of such different mathematical structure, and well below the level of
uncertainty in either (see Figure 4.3).

Figure 4.3 Future Energy Mix in Electricity Sector

Figure 4.3b With Price-Based Climate Policy

5
4

Others
Renewable
Hydro
Nuclear

Natural Gas

Coal

Electricity generation (TkWh)

Electricity generation (TkWh)

Figure 4.3a With No Climate Policy

5
4
3
2
1

2010 2015 2020 2025 2030 2035 2040 2045 2050

2010 2015 2020 2025 2030 2035 2040 2045 2050

Source: MIT Analysis

Chapter 4: Electricity

79

The systems studies in Chapter 3 consider only

uncertainty in the estimates of gas resources
(Figures 3.2, 3.3 and 3.0). Applying the
MARKAL model and the reference assumptions
discussed above, a study was carried out of the
effect on gas use of uncertainties not only in
resources but in other prices, electricity
demand and technology costs. The same two
cases were considered: No Policy; and the
Price-Based policy. Here we describe results for
a 50% confidence interval: i.e., a 25% chance
of gas use above the high level as shown, and a
25% chance of use below the low level. Details
of the analysis are provided in Appendix 4B.
By 2030, with no additional mitigation policy,
the gas demand by the electric sector runs
17% above and 19% below the mean value of
6.3 trillion square feet (Tcf) (50% confidence
interval). The main factors leading to this range
are the demand for electricity, the prices of
natural gas and coal and the costs of new
technologies, in particular the cost of new coal
steam and IGCC technologies.
Under the Price-Based policy the uncertainty is
substantially greater, ranging from 47% above
to 42% below the mean value of 12.8 Tcf (50%
confidence interval). The main influence
behind this greater uncertainty is in the costs of
technologies that might substitute at large scale
for fossil-based generation, such as wind, solar
and advanced nuclear generation technologies.
The share of natural gas in the generation mix
is a result of the interplay between technologies
that both compete with and complement each
other at the same time as they supply different
segments of demand over the year.
The uncertainty ranges given here are intended
to caution the reader against giving too much
weight to the actual numbers in future projections in this chapter and elsewhere in the
report. Rather, the critical insights are about the
trends and relationships, which are more robust
across a wide range of possible futures.

80

MIT STUDY ON THE FUTURE OF NATURAL GAS

NEAR-TERM OPPORTUNITIES FOR

REDUCING CO2 EMISSIONS BY
ENVIRONMENTAL DISPATCH

Near-term opportunities for CO2 emission

reductions in the power sector are limited by
the current generation mix and transmission
infrastructure, the cost of renewables and other
low-emission sources and technologies, as well as
the lag times associated with siting and building
any new generation capacity. The re-ordering of
generation between coal and gas units (modeled
here as a form of environmental dispatch forced
by a CO2 constraint7) may be the only option for
large-scale CO2 emissions reduction from the
power sector which is both currently available
and relatively inexpensive.
As noted, the current fleet of NGCC units has
an average capacity factor of 41%, relative to
a design performance of approximately 85%.
An electric system requires capacity to meet
peak demands occurring only a few hours per
year, plus an operating reserve, so the system
always includes some generation units that run
at capacity factors below their design value.
However, the U.S. has enough spare capacity
in other technologies to allow dispatching more
NGCC generation, displacing coal and reducing CO2 emissions, without major capital
investment. An additional benefit of this
approach would be to substantially reduce
emissions of air pollutants such as sulfur
dioxide (SO2), nitrogen oxide (NOx), mercury
(Hg) and particulates.
NGCC Potential if Fully Dispatched
Figure 4.4 suggests the scale and location of the
potential for shifting among generation units.
Plotted there is the geographic distribution of
fully-dispatched NGCC potential (FDNP),
defined as the difference between the electricity
that would be produced by NGCC plants at
an 85% capacity factor and their actual 2008
generation. Figure 4.4 also shows the geographic
distribution of coal generation, divided into

Figure 4.4 Scale and Location of Fully Dispatched NGCC Potential (FDNP)
and Coal Generation (MWh, 2008)

Source: USREP, MIT

less and more efficient units where a less

efficient unit is defined as one with a heat rate
over 10,000 Btu/kWh.
In many regions FDNP generation matches
well with less efficient coal capacity, suggesting
opportunities for displacing emissions-intensive
units, while other locations show few such
opportunities. For example, Southeastern states
such as Texas, Louisiana, Mississippi, Alabama
and Florida appear to have relatively larger
opportunities, while those in Midwestern states
such as Illinois, Indiana and Ohio are relatively
smaller.

Possible Contribution of NGCC Capacity

to a CO2 Reduction Goal
Figure 4.4 represents only the average potential
available over the course of the year, aggregated
by state, therefore providing an upper limit of
the substitution potential; it does not equate
to surplus generation capacity. For this
discussion, surplus is defined as the amount
of NGCC generation that can be used over the
course of one year to replace coal while respecting
transmission limits, operation constraints and
demand levels at any given time.

Chapter 4: Electricity

81

To account for a number of system characteristics that may better identify the range of
opportunities for fuel substitution, we apply
the ReEDS model (see Box 4.1). This model is
well suited for examination of reliability and
transmission constraints, demand fluctuations
and reserve capacity margins that will limit
these opportunities. Also, as noted, ReEDS
reports results by geographic regions.8

s -)3/ AND 0*- ARE HEAVILY INTERCONNECTED

they import and export electricity from each
other, but have a relatively small amount of
NGCC surplus;
s )3/ .% AND &2## HAVE SURPLUS .'##
but New England has relatively little coal
generation, whereas Florida has a significant
percentage of inefficient coal capacity that
might be a candidate for displacement.

This enables us to identify opportunities to

change the fuel dispatch order nationwide,
and provides insights into five regions of the
country: the Electric Reliability Council of
Texas (ERCOT), Midwest Independent Transmission Operator (MISO), PennsylvaniaNew Jersey-Maryland (PJM), New England
(ISO-NE) and Florida Reliability Coordinating
Council (FRCC). Each region has different
generation costs, fuel mixes and ability
to trade electricity:

We analyze the potential for a version of

environmental dispatch by running the ReEDS
model for the year 2012 in three scenarios: CO2
unconstrained, a 10% reduction in U.S. electric
sector CO2 emissions, and a 20% reduction. Runs
for the year 2012 are used because the model does
not invest in new capacity in this time period; as
such, CO2 reductions are attributable to shift of
generation among existing units.

s %2#/4 IS ESSENTIALLY ELECTRICALLY ISOLATED

from the rest of the country;

Figure 4.5 illustrates the changes in generation

by technology under the three scenarios. In the
20% CO2 reduction scenario, the NGCC fleet
has an average capacity factor of 87%, displaces

Figure 4.5 Generation by Technology under Various CO2 Constraints, U.S.9, 2012
2,500

Generation (TWh)

2,000

1,500

1,000

500
0
Base Case

10% CO2 Reduction

Coal

Gas-CC

Wind

Nuclear

Hydroelectric

Other*

20% CO2 Reduction

Source: MIT Analysis

82

MIT STUDY ON THE FUTURE OF NATURAL GAS

20.0

Figure 4.6 NGCC and Coal Generation in Select Regions under a 20% CO2 Constraint,
U.S., 2012
600

Generation (TWh)

500
400
300
200
100
0
16: MISO
Coal (base case)

21: ERCOT

23: PJM

29: FRCC

32: ISO-NE

Gas Surplus

Source: MIT Analysis

about one-third of 2012 coal generation

(700 terawatt-hours (TWh)) and increases gas
consumption by 4 Tcf.9
In Figure 4.5, as the carbon constraint
increases, most of the electricity generation by
technology does not change. Coal and natural
gas are the exceptions: as the carbon constraint
increases, coal generation significantly declines,
and NGCC proportionally increases.
Although NGCC displacement of coal generation is nearly one-for-one at the national level,
the change in generation and emissions is not
uniform across regions. Figure 4.6 shows
regional results, comparing coal generation in
the absence of a CO2 target to surplus NGCC
generation in a 20% reduction scenario.
In Figure 4.6, the left bars represent the amount
of regional coal generation absent carbon
constraints, using ReEDS 2012 forecasts. This is
the business as usual scenario. The right bars

represent the amount of additional NGCC

generation that is available for dispatch in the
current system after satisfying all system requirements. This additional amount of generation is
calculated as the difference between the NGCC
generation dispatched in the base case and in the
20% CO2 reduction scenario. The largest
potential for substitution of NGCC for coal
generation is in PJM, although in both PJM
and MISO coal continues to dominate.

20.0

A closer look at how the imposition of a CO2

limit would shift generation among units can
be seen in the revised unit dispatch at different
demand levels. For this analysis, we look at
ERCOT, a system that is isolated from the rest
of the U.S. and, in our re-dispatch scenarios,
has regional percentage of CO2 reductions that
tracks national reductions. Because of these
similarities to the country, and because of the
greater availability of operations information
from ERCOT, an analysis of ERCOT, using

Chapter 4: Electricity

83

Figure 4.7 Changes in Dispatch Order to Meet ERCOTs 2012 Demand Profile, with and without
a 20% CO2 Constraint

MW

Oil-Gas-Steam Turbine

90,000

Natural Gas Combustion Turbine

Natural Gas Combined Cycle

80,000

Coal
Wind

70,000

Hydro
Nuclear

60,000

Annual load duration curve

50,000

40,000

30,000

20,000

10,000

Nameplate
capacity

Base
0

Fuel
Switch
1000

Base
2000

Super peak demand 40 hrs total

Late summer afternoons

3000

4000

Fuel
Switch
5000

Average annual dispatch

profile over 8,760 hrs

Base
6000

7000

8000

Fuel
Switch
9000

Low demand 736 hrs total

Typical spring nights

Source: MIT Analysis

ReEDS provides additional details about fuel

switching on a more granular timescale.
Figure 4.7 illustrates how existing capacity
would be dispatched to meet 2012 projected
demand for the highest peak, average and low
demand situations, with and without the CO2
target to force a change in unit dispatch.10 The
figure shows an unconstrained base case and
a case with a 20% CO2 reduction. The average
profile shows the generation dispatch for all
technologies across an entire year (8,760 hours),
not a single time slice.

84

MIT STUDY ON THE FUTURE OF NATURAL GAS

In Figure 4.7, the red line represents 17 time

periods of demand for the year, sorted from
greatest to least demand. The bar graphs to the
right of the nameplate capacity bar show the
dispatch profile in those time periods under
two carbon scenarios: no reduction and 20%
reduction.
Not surprisingly, the results indicate that the
greatest opportunities for displacement of coal
generation exist during average and low
demand periods. Figure 4.7 also shows that coal
generation is dispatched in every time period,
indicating that not enough NGCC surplus
exists in ERCOT to completely displace coal;

Bar graphs
represent
dispatch pro

Table 4.2 National Emissions for CO2-Reduction Scenarios

Base Case

Case 1
10% CO2
Reduction

Case 2
20% CO2
Reduction

CO2 (million metric tons)

2,100

1,890

1,680

SO2 (million tons)

5.66

5.66

5.46

4%

NOx (million tons)

4.66

3.92

3.16

16%

32%

48

40

32

17%

33%

Hg (tons)

% Reduction % Reduction
from
from
Base Case
Base Case
for Case 1
for Case 2

Source: MIT analysis

conversely, surplus NGCC capacity exists and

can displace some coal capacity in all demand
periods examined, even during the super peak,
although the amount is small.
Effect of System Re-Dispatch
on Criteria Pollutants
The Clean Air Act (CAA) requires power plant
controls on SO2, NOx, particulates and Hg.
According to the EPA, 60% of the uncontrolled power plant units are 31 years or older,
[some] lack advanced controls for SO2 and
NOx, and approximately 100 gigawatts (GW)
out of total of [more than 300] GW of coal are
without SO2 scrubbers. 11
Table 4.2 contains results from ReEDS under
the three scenarios that indicate the potential
effects of the CO2 constraint (also shown) on
emissions of SO2, NOx and Hg. (The model
does not project particulate emissions, which
also would be reduced.) While ReEDS does not
fully model the trading markets for SO2 and
NOx, it makes a reasonable approximation by
capping national emissions levels and making
economically efficient dispatch decisions under
these constraints. In all three simulations the
cap for SO2 emissions is based on the 2005
Clean Air Interstate Rule (CAIR) interpolated
for 2012.12

Changes to the dispatch order of generation,

from coal to gas, would lower prices in the SO2
market, and might even yield a reduction in
national emissions below the CAIR limit, as
shown with a 4% change in Case 2. Importantly, the reductions in NOx and Hg emissions
could be substantial, by as much as one-third
under the more stringent CO2 limit.
Table 4.3 shows the corresponding emissions
profiles by region for CO2 and Hg. (ReEDS does
not provide adequate regional detail for SO2
and NOx). Each region acts in its own best
economic interests under the given constraints.
And, because of variation in generation costs,
installed capacity and transmission differences
between regions, some regions have comparative advantage dispatching less CO2 intensive
generation. Depending on the regulatory
structure, regions with these advantages may
produce more electricity, export it and/or sell
credits (assuming a cap-and-trade approach);
and regions which typically deploy technologies
that are more CO2 intensive take opposite
actions. This leads to uneven emissions effects
on individual regions.
A 20% emissions reduction in electric sector
CO2 emissions through coal-to-gas displacement would represent mitigation of 8% of the
U.S. total. The ReEDS model does not provide

Chapter 4: Electricity

85

Table 4.3 Emissions of Select Regions Before and After Re-Dispatch, 2012
Base Case

MISO

ERCOT

PJM

FRCC

ISO-NE

CO2 (million metric tons)

543

153

446

67.2

19

Hg (tons)

13.4

2.77

11

1.32

0.138

CO2 (million metric tons)

394

121

351

78.9

25.4

Hg (tons)

9.30

1.43

7.58

1.13

0.10

% Hg reduction

31%

48%

31%

14%

27%

Case 2 20% CO2 Reduction

Source: MIT Analysis

There is sufficient surplus NGCC capacity to

displace roughly one-third of U.S. coal generation,
reducing CO2 emissions from the power sector
by 20% and yielding a major contribution to
control of criteria pollutants. This would require
an incremental 4 Tcf per year of natural gas, which
corresponds to a cost of $16 per ton of CO2.
an accurate estimate of the national economic
cost of this option, but an approximation can
be made by comparing the break-even CO2
price at which the cost of NGCC generation
equals the cost of coal generation, given their
different variable operations and maintenance
costs, heat rates and CO2 emissions rates.13
The result is an implicit cost of about $16 per
ton CO2.
More analysis is required to determine whether,
because of the geographic differences between
NGCC and coal units, some new transmission
infrastructure may be necessary. Nonetheless,
a more complete analysis is very likely to prove
the cost of this option to be low compared to
most other mitigation options. For example,
one estimate of the per-ton CO2 emissions
avoidance cost estimate to retrofit a typical
sub-critical coal plant with post-combustion
CSS is $74 per ton.14
It should also be noted that coal-to-gas fuel
switching is already occurring. According to the
Energy Information Agency (EIA), The increase
86

MIT STUDY ON THE FUTURE OF NATURAL GAS

in delivered coal prices and the decrease in

delivered natural gas prices, combined with
surplus capacity at highly efficient gas-fired
combined-cycle plants resulted in coal-to-gas
fuel switching. Nationwide, coal-fired electric
power generation declined 11.6 percent from
2008 to 2009, bringing coals share of the
electricity power output to 44.5 percent, the
lowest level since 1978. 15
In sum, there is sufficient surplus NGCC
capacity to displace roughly one-third of U.S.
coal generation, reducing CO2 emissions from
the power sector by 20% and yielding a major
contribution to control of criteria pollutants.
This would require an incremental 4 Tcf per
year of natural gas, which corresponds to a cost
of $16 per ton of CO2. Currently there is no
national price on CO2, but there are both
regional programs and federal regulatory
activities underway.
R E CO M M E N D AT I O N

The displacement of coal generation with

NGCC generation should be pursued as
the most practical near-term option for
significantly reducing CO2 emissions from
power generation.

INTERMITTENT RENEWABLE ELECTRICITY

SOURCES AND NATURAL GAS DEMAND

Effects in the Short Term

To elucidate the short-term effects, we use:

In this section, we explore the impacts of the

introduction of significant amounts of intermittent wind and solar electricity generation
on natural gas generation and overall natural
gas demand.
This analysis first explores the short-term effects
of intermittent wind and solar generation on
gas generation and demand, a scenario which
assumes that the capacity from technologies
other than wind or solar is fixed. Some
European countries already approximate this
situation, where substantial volumes of wind
or solar generation have been installed during
the last few years. Also in some U.S. states, the
proportion of intermittent generation exceeds
10% and the dispatch of existing conventional
generation units has had to adjust accordingly.
We then turn to longer-term impacts, where
the deployment of intermittent generation
is assumed to take place gradually, possibly
in response to government policies that, for
example, set a mandatory target for renewable
generation. Over time, capacity additions and
retirements of other technologies are made as
the system adjusts to intermittent generation.

s A  PROJECTED GENERATION PORTFOLIO AS THE

base case, obtained from the ReEDS CO2
Price-Based policy scenario (see Box 4.1); and
s THE -EMPHIS MODEL SEE "OX  APPLIED TO
daily dispatch patterns for ERCOT which, as
noted earlier, is an isolated system that can be
studied without the complicating influence of
inter-regional transmission.
With this 2030 generation portfolio as our
reference point, we examine the daily dispatch
patterns of all generation technologies, including natural gas, when greater or lesser levels of
wind or solar electricity generation are made
available to be dispatched and the capacities
of the other technologies are held constant.
Wind generation. The results for varying levels
of wind generation are seen in:
s &IGURE A THE BASE CASE WHICH IS A representative day for ERCOT;
s &IGURE B WHEN WIND PRODUCES half the
amount of generation as in the base case; and
s &IGURE C WHERE WIND PRODUCES twice the
amount of generation as in the base case.

Chapter 4: Electricity

87

Figure 4.8 Impact of Wind on a One-Day Dispatch Pattern for ERCOT

4.8a Wind Base Case
60,000

Hydro
Solar

Production (MW)

50,000

Natural Gas
Gas Turbine
Natural Gas
Combined Cycle

40,000

Integrated
Gasification
Combined Cycle
with CCS

30,000
20,000

Biomass
Coal

10,000

Wind
Nuclear
h24

h23

h22

h21

h20

h19

h18

h17

h16

h15

h14

h13

h12

h11

h10

h09

h08

h07

h06

h05

h04

h03

h02

h01

4.8b Wind 0.5

60,000

Production (MW)

50,000
40,000
30,000
20,000
Hydro
Solar

10,000

Gas GT
Gas CCGT

h24

h23

h22

h21

h20

h19

h18

h17

h16

h15

h14

h13

h12

h11

h10

h09

h08

h07

h06

h05

h04

h03

h02

h01

Wind

60,000

Nuclear

50,000

Production (MW)

Biomass
Coal

4.8c Wind 2.0

40,000
30,000
20,000
10,000

Source: MIT Analysis

MIT STUDY ON THE FUTURE OF NATURAL GAS

h24

h23

h22

h21

h20

h19

h18

h17

h16

h15

h14

h13

h12

h11

h10

h09

h08

h07

h06

h05

h04

h03

h02

h01

88

Gas CCGT CCS

In Figure 4.8a, the base case depicts the estimated existing contribution from wind in
ERCOT in 2030. The nighttime load (roughly
hours 01 through 04) is met by nuclear and
coal baseload plus wind generation. There is
no appreciable output from gas between hours
01 and 04 because it has higher variable costs
than nuclear and coal and it gets dispatched
last. Natural gas also has the flexibility to cycle.
In hours 05 through 23, when overall demand
increases during the early morning and
decreases in the late evening, NGCC generation
adjusts to match the differences in demand.
As depicted in Panel 4.8b, when less wind
is dispatched, the NGCC capacity is more
fully employed to meet the demand, and the
cycling of these plants is significantly reduced.
The baseload plants continue to generate at
full capacity.
In Panel 4.8c with twice as much wind as the
base case, natural gas generation is reduced
significantly; the gas capacity that is actually
used is forced to cycle completely. Baseload
coal plants are also forced to cycle because
of the relatively low nighttime demand; coal
plant cycling can increase CO2, SO2 and NOx
emissions.16
Solar Generation. Like wind, for solar there
are figures depicting: a base case in ERCOT
(Figure 4.9a); a case where solar provides half
the amount of generation as the base case
(Figure 4.9b); and a case where solar provides
twice the generation seen in the base case
(Figure 4.9c).

The pattern with solar is somewhat different

than for wind. The solar generation output
basically coincides with the period of high
demand, roughly between hours 06 and 22.
As seen in the base case Figure 4.9a, this is also
when NGCC capacity gets dispatched. The
natural gas plants are used more when solar
output is less (see Figure 4.9b). Conversely,
when solar is used more, less gas is dispatched
(see Figure 4.9c).
The baseload plants are largely unaffected and
cycling is not a problem for them, since there is
no intermittent solar-based generation during
the low-demand night hours.
In sum, our short-term analysis shows that the
most significant impacts of a quick deployment
of additional wind or solar at any given future
year will most likely be both a reduction in
production from, and an increase in cycling of,
gas-fueled NGCC plants; there is a less significant
fall in production for the much-less-employed,
single-cycle gas turbines and steam gas units.

[In the short term].the most significant impacts

of a quick deployment of additional wind or solar
will most likely be both a reduction in production
from, and an increase in cycling of, gas-fueled
NGCC plants.
The displacement of gas is greater for solar than
for wind, since solar production has a stronger
correlation with demand than does wind
generation.
Large wind penetrations may also displace
some coal production and result in some
cycling of these plants. No impact on nuclear
production is expected with the average U.S.
technology mix.

Chapter 4: Electricity

89

Figure 4.9 Impact of Solar CSP (no storage) on One-day Dispatch Pattern for ERCOT
4.9a Solar Base Case
60,000

Solar
Hydro

50,000

Biomass

Production (MW)

Gas GT

40,000

Gas CCGT CCS

Gas CCGT

30,000

Coal IGCC CCS

Coal Old Biomass

20,000

Coal Old NoScrub

Wind

10,000

Nuclear

h24

h23

h22

h21

h20

h19

h18

h17

h16

h15

h14

h13

h12

h11

h10

h09

h08

h07

h06

h05

h04

h03

h02

h01

4.9b Solar Base Case x 0.5

60,000

Solar
Hydro

50,000

Biomass

Production (MW)

Gas GT

40,000

Gas CCGT CCS

Gas CCGT

30,000

Coal IGCC CCS

Coal Old Biomass

20,000

Coal Old NoScrub

Hydro
Wind

10,000

Solar
Nuclear
Gas GT
Gas CCGT
h24

h23

h22

h21

h20

h19

h18

h17

h16

h15

h14

h13

h12

h11

h10

h09

h08

h07

h06

h05

h04

h03

h02

h01

Gas CCGT CCS

Biomass
Coal

4.9c Solar Base Case x 2.0

Wind

60,000

Nuclear
Solar
Hydro

50,000

Biomass

Production (MW)

Gas GT

40,000

Gas CCGT CCS

Gas CCGT

30,000

Coal IGCC CCS

Coal Old Biomass

20,000

Coal Old NoScrub

Hydro
Wind

10,000

Solar
Nuclear
Gas GT
Gas CCGT
h24

h23

h22

h21

h20

h19

h18

h17

h16

h15

h14

h13

h12

h11

h10

h09

h08

h07

h06

h05

h04

h03

h02

h01

Gas CCGT CCS

Biomass

Source: MIT Analysis

Coal
Wind
Nuclear

90

MIT STUDY ON THE FUTURE OF NATURAL GAS

Effects in the Long Term

To explore the effects of the penetration of
intermittent generation over the long term,
we examine two policy scenarios, both with a
system expansion to 2050 and a target leading
to a 70% reduction of CO2 emissions in the
U.S. power sector.
We look at two different versions of the 70%
reduction case because the means by which
the target is implemented through different
mitigation policy instruments has an effect
on how the system responds to more or less
expensive renewable generation. The two policy
instruments we examine are:
s THE IMPOSITION OF A CO2 price to achieve
the CO2 emissions reduction target; and
s THE IMPOSITION OF AN emissions constraint
to achieve the same target.
We then analyze how the electric system, and
gas use over time, would differ if the capital
costs of solar or wind generation capacity were
higher or lower than the reference levels for the
two base cases. Again the ReEDS model is
employed.17
In the ReEDs simulations of both policy
scenarios, the generation mix evolves over time,
similar to that shown in Chapter 3, Figures 3.4a
and 3.4b. During the early-to-middle decades
of the simulation period the dominant event is
the substitution of coal generation by NGCC
units. At the same time, wind generators, with
gas turbine back-up, begin to be deployed as
a baseload technology.18

This combination of wind production and

flexible generation capacity competes with
potential new nuclear capacity and also erodes
NGCC production. Wind impacts the preferred
new baseload generation technology, the one
that is most economic but for which expansion
is not subject to environmental or other limits.
Late in the period, conventional coal production
has been replaced, economically-competitive
wind resources start becoming exhausted and
nuclear plus some solar penetration begins.
CO2 Price-Based Case. In the CO2 Price-Based
case, the nature of the system adjustments in
these simulations can be illustrated using an
example of the changes that would be brought
about by lower-cost wind capacity. First, the
increased intermittent renewable generation
needs to be accompanied by flexible back-up
capacity, albeit with low utilization levels. In
the U.S., spare capacity of gas-fueled plants is
enough to meet this requirement initially, but
eventually additional investment is needed
(gas turbines in these scenarios).
As this combination of new intermittent renewable and flexible electricity plants grows, it
starts to replace the expansion and utilization
of baseload generation technologies, nuclear
or fossil generation with CCS (coal without
CCS has already been forced out of the system
by its CO2 emissions). However, these classic
baseload technologies are not increasing;
therefore, the low-cost renewable capacity plus
flexible generation increases in baseload and
even in mid-merit service, at the expense of
gas generation.

Chapter 4: Electricity

91

This interaction can be illustrated with a

summary of what happens in the base case
system for the ERCOT region when different
renewable costs are simulated, therefore
changing the intermittent generation penetration levels.
The results in Figure 4.10 are plotted to highlight the way cumulative gas generation
changes with different assumptions about
wind-generation costs and the corresponding
wind-generation levels. The figure shows the
total generation in TWh by type of generation
technology over the simulation period from
2005 to 2050 and assumes the underlying
emissions target is imposed by a CO2 price.
It illustrates that the displacement of gas by
wind takes place through changed patterns of
investment and generation over many years.

As Figure 4.10 shows, increased cumulative

wind generation, as a consequence of lower
wind investment costs, or an aggressive renewable portfolio standard, has a direct impact on
the new investment and associated production
by natural gas, equal to almost one TWh of
reduced natural gas generation for one TWh
of wind output. This happens because NGCC
is the technology that is most vulnerable to
wind competition, both before and after coal
has been driven out of the market. It should
also be noted that, while the cumulative
generation of gas turbines (Gas-CT in Figure
4.10) does not change enough to show in the
graph, gas turbine capacity actually increases
substantially to support the additional wind
contribution.

Figure 4.10 Cumulative Generation in ERCOT in the Period 20052050 for All Technologies Given
Alternative Levels of Wind Penetration (TWh)
5,000

Wind (Gen)

4,500

Nuclear (Gen)
Coal Old Scrubbed

4,000

Coal Old Unscrubbed

TWh

3,500

Oil-Gas Steam

3,000

Gas-CC

2,500

Gas-CT

2,000

Hydro
Utility PV

1,500
1,000
500
0
250

450

650

Source: MIT Analysis

92

MIT STUDY ON THE FUTURE OF NATURAL GAS

850

1,050

1,250

1,450

1,650

In Figure 4.10, the horizontal axis is cumulative

wind output, the vertical axis is the cumulative
output for all technologies, including wind
(if the two axes were plotted to the same scale
the function for wind would be a 45 line). The
base-case level of wind generation is indicated
with a vertical line, so that output to the right
of that point results from lower capital costs
and the output to the left results from higher
capital costs.
Figure 4.10 also shows that the difference in
cumulative generation by the other technologies is not much affected by changes in the
contribution of wind generation. It should be
repeated that this is a result for ERCOT. The
differences in generation mix in other regions
will vary, though viewed at the national level
the pattern is very similar to that shown here.19

generation does not require back-up from

flexible gas plants as much as wind does. In fact,
solar can partially fulfill a peaking plant role.
In summary, our analysis of gradual and
sustained long term penetration of wind and
solar shows that large-scale penetration of wind
generation, when associated to flexible natural
gas plants, will assume a mostly baseload role,
and will reduce the need for other competing
technologies such as nuclear, coal or even
gas-fueled combined cycles, if expansion with
coal and nuclear technologies does not take place

Our analysis shows that a gradual and sustained

long term substantial penetration of wind, when
associated with flexible natural gas plants, will
assume a mostly baseload role, and will reduce the need
for other competing technologies such as nuclear, coal
or even gas-fueled combined cycles. This effect is less
pronounced in the case of solar.

CO2 Cap Case. The result differs somewhat if

emissions mitigation is accomplished by a CO2
cap instead of a price. The fixed CO2 constraint
implies that an increment in wind output that
displaces NGCC production and investment
also reduces the need for other low-CO2
baseload capacity to reduce the emissions.

because of economic, environmental or any

other reasons. This effect is less pronounced in
the case of the solar technology, because of its
characteristic daily production pattern.

Cheaper wind creates slack under the emissions

constraint, which may be filled by whatever is
the cheapest generation source. In some
simulations, this cheap generation comes from
otherwise almost-idle coal-fired plants. Thus,
as a minor perverse effect, under the CO2
constraint more wind can imply a small
increment of additional coal production
a condition that does not occur when coal
is burdened by a CO2 price.

Although our analysis has been limited to a few

alternative scenarios, we can observe a consistent pattern for the impact of intermittent
renewable generation: We see that an increase
of wind or solar output systematically results
in a proportionally significant reduction of
natural gas fueled production, while, at the
same time, the total installed capacity of flexible
generation (typically also natural gas fueled
plants) is maintained or increased.

The case of solar generation without storage

is similar to wind in many respects. However,
since the production profile of solar has a high
level of coincidence with the daily demand and
has a more stable pattern, an increment in solar

Precise numerical estimations and any second

order impacts are heavily dependent on the
specific energy policy instruments and the
assumptions on the future costs of fuels and
technologies.

Chapter 4: Electricity

93

The detailed operational analysis of plausible

future scenarios with large presence of wind
and solar generation reveals the increased need
for natural gas capacity (notable for its cycling
capability and lower capital cost) to provide
reserve capacity margins. This does not however necessarily translate into a sizeable utilization of these gas plants.

R E CO M M E N D AT I O N

In the event of a significant penetration of

intermittent renewable production in the
generation technology mix, policy and
regulatory measures should be developed
to facilitate adequate levels of investment
in natural gas generation capacity to ensure
system reliability and efficiency.

Additional Implications
In deregulated wholesale markets with substantial penetration of renewables, the volatility of
marginal prices can be expected to increase.
Also, mid-range technologies, of which NGCC
is the most likely candidate, will see their
output reduced. The uncertainty regarding the
adequate technology mix, and the economics of
such a mix under the anticipated future prices
and operating conditions, raises concern about
attracting sufficient investment in gas-fueled
plants under a competitive market regime.
This issue is presently being addressed by
several European countries with significant
penetration of wind generation, where the
patterns of production of NGCC and single
cycle gas turbines and also of some baseload
technologies, have already had major impacts.
Similar situations are developing in some parts
of the U.S. Presently there is no consensus on
a suitable regulatory response to this situation,
which could include enhancements of any
capacity mechanisms such as those already
in place in most U.S. wholesale markets, new
categories of remunerated ancillary services
or other instruments.

94

MIT STUDY ON THE FUTURE OF NATURAL GAS

Although limited in scope, our analysis shows

the diversity and complexity of the impacts that
a significant penetration of intermittent
generation (mostly wind and solar, in practice)
have on the technology mix and the operation
of any considered power system. The possible
future emergence of electricity storage options,
as well as enhanced demand responsiveness,
will also affect the need for flexible generation
capacity, which is presently fueled by natural
gas. The level and volatility of future energy
prices will determine the volume and nature of
investment in future generation under market
conditions. Other regulatory frameworks
should also be considered.
These complicated implications and trade-offs
cannot be spelled out without the help of
suitable computer models. The accuracy of
the estimates of future fuel utilization and the
adequate technology mix critically depends
on the performance of these models. Unfortunately, the state-of-the-art computer models
that simulate and optimize the capacity expansion and the operation of power systems and
electricity markets such as ReEDS or Memphis are still in a development phase and fall
short of the requirements to incorporate
intermittent generation, storage and demand
response realistically, under a variety of energy
policies and regulatory environments.

R E CO M M E N D AT I O N

A comprehensive appraisal of the economic, environmental and reliability

implications of different levels of significant penetration of renewable
generation should be performed for power systems with different generation
technology portfolios and under different energy policy scenarios.
The information obtained from this appraisal should inform a central piece
in the design of energy policies that contemplate mandating large amounts
of solar or wind generation.
Additional efforts should be made to expand or develop the sophisticated
computation electric system models that are needed for this task.

Chapter 4: Electricity

95

NOTES
1

Nameplate capacity is the nominal, maximum

instantaneous output of a power plant.

Absent other considerations, generation units are

normally dispatched in economic merit order, i.e.,
those with lower variable operating costs first.

10

Although the trend for NGCC displacement of

coal generation remains the same for this updated
scenario, these results are numerically different
than the results presented in the interim report.
The interim report showed opportunities for coal
displacement in all time periods. The difference
stems from assumptions about how much NGCC
capacity exists in ERCOT. The NGCC capacity
numbers used for this 2012 simulation are more
conservative, and projected forward from 2006
EIA capacity and generation data (2006 is the start
year for ReEDS).

11

Presentation, Reducing Pollution from Power

Plants, Gina McCarthy, Assistant Administrator,
U.S. EPA Office of Air and Radiation, October 29,
2010.

12

For a variety of reasons, deployment of required

controls has been delayed, largely by court findings
of legal flaws in various rulemakings. The New
Transport Rule, which will replace Clean Air
Interstate Rule (CAIR) in place today, is expected
to be finalized in mid-2011 and will be implemented over time, with most coverage finalized
by 2014. The Transport Rule will cover SO2 and
NOx. EPA released a proposed rule for mercury
emissions from coal and oil-fired power plants in
March, 2011 and plans to finalize the rule by the
end of the year. A final rule on CO2 for power
plants is expected sometime in 2012.

13

This break-even price assumes a NGCC

variable O&M cost of $3.20/MWh, fuel price
of $5.38/mmBtu, heat rate of 6.04 mmBtu/MWh,
and CO2 emissions of 0.053 tons/mmBtu. For coal,
the calculation assumes a variable O&M cost of
$4.30/MWh, fuel price of $2.09/mmBtu, heat
rate of 10 mmBtu/MWh, and CO2 emissions of
0.098 tons/mmBtu. The cost of NGCC and coal
generation break-even when the sum of the variable
O&M cost and price per ton CO2 multiplied by the
amount of CO2 emitted are equal to each other, for
the respective fuels. Start-up and shut-down costs,
ramp rates, associated changes in emissions, and
other costs that have not been fully modeled are
not included in this calculation.

14

MIT Energy Initiatives report on Retrofitting

of Coal-Fired Power Plants for CO2 Emissions
Reductions, Cambridge, MA, 2009 (http://web.mit.
edu/mitei/docs/reports/meeting-report.pdf).

15

EIA AEO 2010.

Channele Wirmin, EIA, private communication.

3
4

Steinhurst, W., The Electric Industry at a Glance,

Nuclear Regulatory Research Institute, 2009.

Electric Power Research Institute, A Primer on

Electric Power Flow for Economists and Utility
Planners, 1995; Prez-Arriaga, I., Rudnick H.,
Rivier, M., Chapter One: Electric Energy Systems,
An Overview.

Hydroelectric generation, shown in Figure 4.2

as constant over demand periods, will in fact tend
to be concentrated in particular seasons and peak
periods of the day. The MARKAL model does not
represent this detail, though its inclusion would
have only a small effect on the figure as it aggregates all the national hydroelectric facilities.

The same change in unit dispatch could be

approached using various forms of direct
regulation, options not studied here.

The ReEDS model captures key characteristics of

the electricity networks transmission constraints
and reliability requirements by splitting the country
into 134 geographic partitions. Each partition
balances demand and supply of electricity by
independently generating, importing, and
exporting electricity. Collectively, subsets of these
balancing areas constitute the independent system
operators (ISOs) and regional transmission
organizations (RTOs).

96

As noted in the introduction of this section, the

expected maximum capacity factor for an NGCC
plant is 85%. The EIA projects that this could
increase to 87% by 2016 (http://www.eia.doe.gov/
oiaf/aeo/pdf/2016levelized_costs_aeo2010.pdf).
The average fleet capacity factor of 87% from
ReEDS for the 20% CO2 reduction scenario
approaches the upper generation threshold of the
countrys current NGCC fleet.

MIT STUDY ON THE FUTURE OF NATURAL GAS

16

See Bentek study, How Less Became More: Wind,

Power and Unintended Consequences In the
Colorado Energy Market, April 2010.

17

See Impact of intermittent renewable electricity

generation on the technology mix and fuel
consumption in the U.S. power system. Yuan Yao,
Ignacio J. Prez-Arriaga. CEEPR (Center for
Energy and Environmental Policy Research), MIT,
May 2011.

18

The ReEDS simulations of this level of mitigation

show a greater penetration of renewable generation than do the results of the EPPA model shown
in Chapter 3, but the difference is not an important influence on the insights to be drawn from
these calculations.

19

Details of these cases are provided by Yao and

Prez-Arriaga, op cit.

Chapter 4: Electricity

97