Energy 29 (2004) 2125–2144

www.elsevier.com/locate/energy

Incorporating externalities into a full cost approach to

electric power generation life-cycle costing
Ian F. Roth a,
, Lawrence L. Ambs b
a
Noresco, 129 South Street, Boston, MA 02111, USA
b
Mechanical and Industrial Engineering Department, Center for Energy Efficiency and Renewable Energy,
University of Massachusetts Amherst, Amherst, MA 01003, USA

Abstract

This study presents a full cost approach to determine the levelized cost of energy (LCOE) of 14 elec-
tricity generation technologies. It encompasses costs incurred at all stages of the fuel cycle, including
those that are traditionally omitted from economic evaluations of generation technologies. Incorporating
these ‘‘externalities’’ increases the likelihood of developing the most economical and sustainable power
resource from a societal perspective. The following externalities are included in this analysis: damage
from air pollution, energy security, transmission and distribution costs, and other environmental impacts.
Incorporating externalities has a large impact on the LCOE and the relative attractiveness of electricity
generation options. Results indicate that clean and efficient generation technologies are the most attract-
ive when all options are examined using a full cost, levelized approach.
# 2004 Elsevier Ltd. All rights reserved.

1. Introduction

The goal of this study was to develop a model of electric power generation pricing that
reflects long-term economic feasibility and sustainability and that encourages optimal resource
selection. The model was used to compare the affordability of electricity produced from different
electricity generation technologies.
Electric utility companies use factors such as fuel costs, capital costs, and plant lifetimes to
determine the most affordable method of power generation. This process gives limited consider-
ation to more indirect factors that are treated as external to the energy economy, including
environmental damage and other social costs. The failure of power rates to reflect external

Corresponding author. Tel.: +1-617-542-8567; fax: +1-617-338-2563.
E-mail address: iroth@noresco.com (I.F. Roth).

0360-5442/$ - see front matter # 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.energy.2004.03.016
2126 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144

Nomenclature

C1 conversion (1 ton/2000 lbs)

C2 conversion (106 MMBtu/Btu)
C3 conversion (2.205 lbs/kg)
C4 conversion (106 GW h/kW h)
CF capacity factor (%)
DC damage cost ($/tonpollutant)
EF emission factor (lbpollutant/MMBtu)
EF0 emission factor (kgpollutant/GW h)
e escalation rate (%)
FC fuel cost ($/MMBtu)
FCR fixed charge rate (%)
FOM fixed O&M cost ($/kW year)
HR heat rate (Btu/kW h)
HY hours per year
LCOE levelized cost of energy ($/kW h)
LF levelization factor (%)
n percentage of NOx removed by scrubbing
p percentage of PM removed by scrubbing
PC total plant cost ($/kW)
PL plant life (years)
r discount rate (%)
s percentage of SO2 removed by scrubbing
VOM total variable O&M cost ($/kW h)
XC external cost ($/kW h)

effects and the fact that ratepayers do not directly pay for them are viewed by many economists
as inefficiencies of the market. On the other hand, the benefits of clean and efficient power gen-
eration are sometimes recognized through tax breaks and environmental regulations. However,
the ability of these political mechanisms to reflect the real benefits to society of clean and
efficient power generation and to appropriately influence resource-planning decisions is ques-
tionable. The life-cycle costing (LCC) method presented here offers a more direct approach to
choosing optimal electricity supply options based on monetary values. This method accounts
for traditional costs, environmental costs, and other social costs and facilitates objective com-
parison of generation technologies using these criteria.
As in previous studies, externalities are defined here as ‘‘benefits or costs, generated as a
byproduct of an economic activity, that do not accrue to the parties involved in the activity’’ [1].
In the context of electricity pricing, the term ‘‘externalities’’ refers to costs associated with the
fuel cycle that are not incorporated into the electric utility cost structure. The fuel cycle is
defined by the US Energy Information Administration (EIA) as ‘‘the series of physical and
 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144 2127

chemical processes and activities required to generate electricity from a specific resource, includ-
ing primary resource extraction and preparation, transport and storage of resources and materi-
als, processing and conversion, and disposal’’ [2]. A comprehensive examination and inclusion
of externalities serves as a means of obtaining the most efficient and ‘‘best’’ generation source
from a standpoint that considers ‘‘environmental preservation, human health, and long-term
stability as well as traditional monetary costs and benefits’’ [3]. This approach, known as full
cost pricing, is the method of analysis used in this study.
The 14 different generation technologies examined are: Conventional Coal Boilers, Advanced
Fluidized Bed Combustion (AFBC), Integrated Gasification Combined Cycle (IGCC), Conven-
tional Oil Boilers, Simple Cycle Gas Turbines, Advanced Gas Turbines, Advanced Combined
Cycle, Mass Burn Municipal Solid Waste (MSW), Landfill Gas (LFG) Recovery, Solid Oxide
Fuel Cells (SOFCs), Utility Scale Wind Turbines, Utility Scale Flat Plate Photovoltaics (PV),
Hybrid Solar Thermal Parabolic Troughs, and Biomass Combustion.
The analysis uses the existing literature to find quantitative costs associated with emissions
and other social impacts of electric power generation at all stages of the fuel cycle and includes
them in calculations of the levelized cost of energy (LCOE) for each generation option.
The LCOE is a LCC metric that allows clear and direct comparison of generation technologies
based on their long-term economic feasibility. It can be used for many different evaluative
purposes, including utility resource selection, dispatch decisions, electricity pricing, energy
conservation programs, R&D incentives, subsidy determination, and environmental policy
planning.

2. Economic model

LCOE can be interpreted as ‘‘a constant level of revenue necessary each year to recover all
expenses over the life of a power plant’’ [4] and is expressed as a cost per kW h. Externalities
are included by adding the external cost term ‘‘XC’’ to the LCOE equation used by the Cal-
ifornia Energy Commissions (CEC) [4]. The result is the following equation, where abbrevia-
tions may be found in the Nomenclature above:


 
FCR  PC FOM
LCOE ¼ þ LF  þ VOM þ ½LF  FC  HR þ XC (1)
HY  CF HY  CF

The levelization factor, LF, accounts for the time value of money and is found as follows:
"   #
rð1 þ rÞPL ð1 þ eÞ 1 þ e PL
LF ¼   1 (2)
ð1 þ rÞPL  1 ðr  eÞ 1þr

The degree to which the LCOE of each generation technology will be affected by externality
valuation uncertainty depends upon the relative importance of each externality in that technol-
ogy’s LCOE. For example, the damage cost attributed to SO2 emissions will have a considerable
influence on the LCOE of coal plants, but will not affect the LCOE of wind turbines. Results
2128 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144

Table 1
Economic parameters
Parameter Best estimate
Discount rate 5.50%
Fixed charge rate
Plant life 20 years 13.5%
Plant life 25 years 12.6%
Plant life 30 years 12.0%
Plant life 35 years 11.6%
O&M escalation rate 0.73%
Fuel escalation rates
Coal 1.0%
Oil 2.5%
Natural gas 1.6%
Municipal solid waste 1.8%
Biomass 0.0%
Fuel costs ($/MMBtu)
Coal 1.21
Oil 2.42
Natural gas 2.55
Municipal solid waste 3.06
Biomass 1.92

are therefore presented as ranges and discussed bearing this uncertainty in mind. Other plant
and economic parameters also introduce variance in the LCOE [5]; however, in order to focus
on the importance of XC valuation, best estimate values for these other parameters are used in
all calculations. Monetary values are expressed in constant (real) US dollars with 1999 as the
base year, and the real discount rate is used. The percentage change in consumer price index
(CPI) is used for inflation rates.
Since the LCOE is a measure of the cost to produce energy over the lifetime of the power
plant, all capital, operations and maintenance (O&M), and fuel costs must be accounted for.
Calculating these components of the LCOE requires information on plant operating parameters,
on the costs of building and running the plants, on the energy market, and on the general econ-
omy. Values for the economic parameters used in this study are presented in Table 1. Fuel costs
and escalation rates for 1999, as reported by the Energy Information Administration (EIA) [6]
in 1999 US dollars, are used for coal, oil, and natural gas. For MSW and biomass, averages
from CEC [4] are used. The fuel cost of MSW facilities is negative because the tipping charge
for receiving waste generates revenue. The discount rate and O&M escalation rate are also
based on the CEC study [4].
Values used for plant parameters are derived from five different sources [4,7–10] and are
shown in Table 2. Some parameters are dependent upon location, so when necessary, generators
are assumed to be sited in the northeastern US. For a different location or specific generators,
readers may wish to use different values in the model to determine full cost LCOEs.
Table 2
Plant parameters
Plant Conven- Advanced Integrated Conventional Simple Advanced Advanced Mass Burn Landfill Solid Utility PV: Util- Hybrid Biomass
parameter tional Boiler Fluidized Gasifi- Boiler (Oil)— Cycle Gas Combined MSW Gas Oxide Scale ity Scale Solar
(Coal)— Bed Com- cation Rankine Gas Turbine Cycle Recovery Fuel Cells Wind Flat Thermal
Rankine bustion Comb Cycle Turbine Turbine Plate Parabolic
Cycle Cycle Trough

Capital cost $ 1800 $ 2200 $ 2100 $ 1300 $ 700 $ 400 $ 600 $ 5700 $ 1500 $ 1600 $ 1000 $ 4700 $ 3700 $ 2400
($/kW)
Capacity 85% 83% 85% 80% 10% 70% 90% 85% 70% 95% 25% 13% 25% 90%
factor (%)
Fixed O&M $ 56.70 $ 57.10 $ 49.10 $ 15.10 $ 0.50 $ 0.10 $ 16.10 $ 164.10 $ 56.70 $ 245.80 $ 15.60 $ 10.40 $ 48.00 $ 107.00
($/kW year)
Variable O&M $ 1.50 $ 7.90 $ 1.90 $1.50 $ 10.90 $ 3.90 $ 0.80 $ 17.10 $ 0.00 $ 31.20 $ 8.30 $ 0.00 $ 5.60 $ 10.00
($/MW h)
HR (Btu- 9950 9750 8893 9430 11500 10900 6828 16870 12150 7584 0 0 3456 14310
HHV/kW h)
Plant life 35 35 35 35 25 25 30 25 20 25 25 30 30 35
(years)
I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144
2129
2130 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144

3. Air pollution externalities

To determine the XCs associated with emissions of CO2, SO2, NOx, and other air pollutants
emitted from electric power generation plants, it is necessary to calculate the quantities released
in the production of a unit of electricity. This is accomplished by using a damage cost, DC (US
dollars per ton of pollutant), ascribed to the emissions and an emission factor, EF (tons of pol-
lutant emitted per unit of fuel energy consumed). The cost per unit of electricity produced is cal-
culated using these factors and incorporated directly into the LCOE equation. The generalized
equation used for calculating the XC (¢/kW h), of common air pollutants is as follows:
XC ¼ DC  EF  C1  HR  C2 (3)

3.1. Emission factors

Emission levels are functions of the chemical composition of the fuel burned, the power cycle
and generation technology, and what pollution control equipment is used. The EF values used
in this study are presented in Table 3. They were obtained primarily from US Environmental
Protection Agency data [11,12] with supplementary values derived from other sources [13–17].
For MSW generators the upstream EF, which represents emissions associated with extracting,
processing, and transporting fuels, is set equal to zero since the fuel burned is material manufac-
tured for other purposes and would otherwise be transported to a landfill. For LFG Recovery
generation, it is assumed that if the available gas were not burned to produce electricity, it
would be burned off with a flare. Consequently, EFs associated with combustion of LFG are all
equal to zero. Fuel cell emissions depend upon the technology employed and the fuel used. The
EFs used in this study are based on emissions from SOFCs, and the fuel used is assumed to be
natural gas.
Biomass EFs are based on emissions from the combustion of wood residue [11]. Similar to
the treatment of upstream emissions for MSW generators, the upstream EF associated with
using wood residue in biomass generators is considered negligible. This would not be the case if
dedicated crops requiring energy, water, and chemicals were used as the biomass fuel source;
however, in the US, ‘‘residues from the wood products industry, logging residues, and crop resi-
dues constitute the largest source of potential biomass generation’’ [15]. The CO2 EF for bio-
mass power plants is zero given that all carbon emitted during combustion was recently
sequestered. Similarly, the CO2 EF for MSW is low compared to fossil fuels because it is
assumed that only 11% of the waste combusted (by mass) is non-biogenic carbon [12]. To
accommodate cases where biomass is not grown sustainably, upper range CO2 EFs used in the
uncertainty analysis for biomass and MSW are 195 and 218 (ton/MMBtu), respectively [11].
To account for the fact that some generation technologies and control equipment limit the
emissions of certain air pollutants, the scrubbing terms s, n, and p are incorporated into Eq. 3
as follows:
XCSO2 ¼ ½DC  EFð1  sÞSO2  C1  HR  C2 (4a)
XCNOx ¼ ½DC  EFð1  nÞNOx  C1  HR  C2 (4b)
XCPM ¼ ½DC  EFð1  pÞPM C1  HR  C2 (4c)
Table 3
Emission factors (lb/MMBtu)
Air pol- Conven- Advanced Integrated Conventional Simple Advanced Advanced Mass Burn Landfill Solid Oxide Utility PV: Utility Hybrid Biomass
lutant tional Fluidized Gasifi- Boiler (Oil)— Cycle Gas Gas Combined MSW Gas Fuel Cells Scale Scale Flat Solar
Boiler Bed Com- cation Rankine Turbine Turbine Cycle Recovery Wind Plate Thermal
(Coal)— bustion Comb Cycle Turbine Parabolic
Rankine Cycle Trough
Cycle

CO2 205 205 205 174 117 117 117 89 0 117 0 0 117 0
CH4 0.0015 0.0015 0.0015 0.0020 0.013 0.013 0.013 0 0 0 0 0 0.013 0.021
N2O 0.0035 0.0035 0.0035 0.00066 0.00018 0.00018 0.00018 0.022 0 0 0 0 0.00018 0.013
Upstream 20.0 20.0 20.0 34.1 24.5 24.5 24.5 0 0 24.5 0 0 24.5 0
CO 0.019 0.019 0.019 0.033 0.082 0.082 0.082 0.051 0 0 0 0 0.082 0.60
SO2 1.84 1.84 1.84 1.48 0.00058 0.00058 0.00058 0.38 0 0.00058 0 0 0.00058 0.025
NOx 0.84 0.84 0.84 0.44 0.40 0.40 0.40 0.40 0 0.001 0 0 0.40 0.220
PM 0.10 0.10 0.10 0.11 0.0066 0.0066 0.0066 0.023 0 0.0009 0 0 0.0066 0.50
VOC 0.0013 0.0013 0.0013 0.0023 0.0021 0.0021 0.0021 0.0089 0 0.005 0 0 0.0021 0.013
I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144
2131
2132 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144

Table 4
Scrubbing terms
Technology s n p
Conventional Boiler (Coal)—Rankine Cycle 0.0 0.0 0.0
Advanced Fluidized Bed Combustion 0.9 0.6 0.9
Integrated Gasification Comb Cycle 0.9 0.9 0.9
Conventional Boiler (Oil)—Rankine Cycle 0.0 0.0 0.0
Gas Generators 0.0 0.6 0.0
Biomass 0.0 0.6 0.9

Hence, s, n, and p represent the fraction of SO2, NOx, and particulate matter (PM) reduction,
respectively. Table 4 shows the values used in calculations.

3.2. Damage costs

The XCs of air pollution are difficult to quantify, and there is much discrepancy among exter-
nality studies. Some base their estimates on the actual economic costs of the damage done. This
method is often referred to as ‘‘damage costing’’ and is highly complex, as it demands difficult
judgments in the valuation of external effects such as damage to ecosystems, health impacts, and
loss of human life.
Control costing, on the other hand, is more straightforward. XCs are based on the cost to
control or clean up emissions with the assumption that these are reasonable approximations of
the economic impact of the damage done. The term ‘‘damage costs’’ as it is used in this study
represents the cost of the external effects of an air pollutant (in dollars per ton), and includes
proxy damage costs (DCs) derived from control costs.
Damage costs used in this study were determined by examining and synthesizing information
from numerous externality studies [2,15,18–20]. Due to inherent uncertainty in externality valu-
ation, each DC is presented as a range of values. Best estimates are akin to median values of the
damage costs found in the literature and typically represent costs to install equipment to reduce
air pollution emissions. Lower and upper range values are chosen to represent a consistent
range found in the literature, but are not typically the most extreme values reported, as those
are considered outliers. The DCs used in this study are summarized in Table 5. More details on

Table 5
Damage costs ($/ton)
Air pollutant Lower range ($) Best estimate ($) Upper range ($)
CO2 9.90 26.40 41.60
CO 506.23 1055.87 2494.26
SO2 1635.98 1869.77 4933.99
NOx 1049.27 7919.03 10,030.77
PM 3128.55 4839.41 13,616.33
VOC 1113.06 5265.79 6489.20
 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144 2133

valuation of air pollutant externalities and their environmental impacts may be found in
Roth [5].
The DCs for greenhouse gas emissions other than carbon dioxide (CO2) are determined by
assigning each a Global Warming Potential (GWP). The GWP represents the estimated climate
impact expressed as an equivalent release of carbon dioxide. For CH4 and N2O the GWP is
24.5 and 320, respectively [13]. Since upstream emissions are expressed as CO2 equivalent emis-
sions, their GWP is 1. The formula used to calculate XCs of greenhouse gas emissions during
combustion is based on Eq. 3 and is as follows:
XC ¼ ðDCCO2  GWPÞ  EF  C1  HR  C2 (5)
Results of the air pollution XC calculations using best estimates are presented in Table 6.

4. Other environmental externalities

For other environmental impacts, the XCs are expressed directly in ¢/kW h. The values used
are included in Table 6. For more detailed descriptions of the external effects and valuations, see
Roth [5].
Land use externalities reflect the occupation of land for plant sites, fuel storage, transmission
lines, and waste disposal. The XC for fossil fuel plants is based on the cost to control or miti-
gate these impacts [15]. MSW and LFG Recovery plants are assumed to already occupy land
that would be used to store wastes, and are therefore assigned land use XCs of 0. Biomass fuel
cultivated from dedicated cropland would occupy significant areas but since wood residue
is obtained from land intended for other purposes, biomass land use XCs are considered
negligible. For all lower and upper range values, the best estimate is multiplied by 0.5 and 3,
respectively.
Fossil fuel plant solid waste can have adverse effects on both surface water and groundwater.
This damage is included under water-related impacts and is assigned a best estimate XC that
represents the cost to control or mitigate residual water discharges [15]. Lower and upper range
values are 0.5 and 3 times the best estimate value, respectively.
When condenser water is removed from and returned to lakes and rivers, aquatic life is
adversely affected. Using a closed loop cooling system can prevent this damage. The cost of
installing such a system [15] is used as an estimate of the impacts on wildlife XC. The lower and
upper range values used are 0.022 and 0.029 ¢/kW h, respectively.
For MSW facility toxic emissions (organics and heavy metals), the best estimate is set equal
to the damages from conventional pollutants. The lower and upper range values used are 5%
and 200% of the XCs of conventional emissions. Because MSW generators also serve to elimin-
ate garbage, some of the XCs are allocated to this function. As suggested in the Pace University
study, MSW XCs are offset by the ‘‘cost that would otherwise be incurred in landfilling solid
waste’’ [15]. The best estimate of this cost is 7.72 ¢/kW h and the lower and upper ranges are
5.10 and 10.24 ¢/kW h.
The total XC associated with wind turbines (land use, visual impacts, noise pollution) is
reported to be between 0 and 0.13 ¢/kW h [15]. This range is used here with the upper limit also
used as the best estimate.
 2134

Table 6
Best Estimate External Costs (¢/kW h)
External Conven- Advanced Integrated Conven- Simple Advanced Advance- Mass Burn Landfill Gas Solid Utility PV: Utility Hybrid Biomass
cost tional Fluidized Gasifi- tional Cycle Gas Gas Tur- d Com- MSW Recovery Oxide Fuel Scale Wind Scale Flat Solar
Boiler Bed Com- cation Boiler Turbine bine bined Cells Turbine Plate Thermal
(Coal)— bustion Comb (Oil)— Cycle Parabolic
Rankine Cycle Rankine Trough
Cycle Cycle

CO2 2.692 2.638 2.406 2.166 1.776 1.683 1.054 1.979 0.000 1.171 0.000 0.000 0.534 0.000
CH4 0.000 0.000 0.000 0.001 0.005 0.004 0.003 0.000 0.000 0.000 0.000 0.000 0.001 0.010
N2O 0.015 0.015 0.013 0.003 0.001 0.001 0.001 0.158 0.000 0.000 0.000 0.000 0.000 0.079
Upstream 0.262 0.257 0.234 0.425 0.372 0.352 0.221 0.000 0.000 0.245 0.000 0.000 0.112 0.000
CO 0.010 0.010 0.009 0.016 0.050 0.047 0.030 0.046 0.000 0.000 0.000 0.000 0.015 0.453
SO2 1.715 0.168 0.153 1.307 0.001 0.001 0.000 0.606 0.000 0.000 0.000 0.000 0.000 0.033
NOX 3.306 1.296 0.295 1.650 0.725 0.687 0.430 2.642 0.000 0.002 0.000 0.000 0.218 0.499
PM 0.241 0.024 0.022 0.243 0.018 0.017 0.011 0.095 0.000 0.002 0.000 0.000 0.006 0.173
VOC 0.003 0.003 0.003 0.006 0.006 0.006 0.004 0.039 0.000 0.010 0.000 0.000 0.002 0.049
Land use 0.526 0.526 0.526 0.526 0.263 0.263 0.526 0.000 0.000 0.263 0.000 0.000 0.263 0.000
Water- 0.131 0.131 0.131 0.131 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
related
Wildlife 0.025 0.000 0.000 0.025 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Energy 0.000 0.000 0.000 3.220 3.102 2.941 1.892 0.000 0.000 2.046 0.000 0.000 0.958 0.000
security
supply
Energy 1.143 1.143 1.143 1.143 1.143 1.143 1.143 0.000 0.000 1.143 0.000 0.000 1.143 0.000
security
I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144

depletion
Distri- 2.000 2.000 2.000 2.000 2.000 2.000 2.000 2.000 2.000 0.000 2.000 0.000 2.000 0.000
bution
Other 0.000 0.000 0.000 0.000 0.000 0.000 0.000 2.156 0.000 0.000 0.131 0.990 0.263 0.000
Total XCs 12.07 8.21 6.94 12.86 9.46 9.15 7.31 5.41 2.00 4.88 2.13 0.99 5.51 1.30
 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144 2135

Table 7
PV manufacturing EFs and XCs
Air pollutant EF0 (kg/GW h) XC (¢/kW h)
Lower Best estimate Upper
CO2 164,208 0.179 0.478 0.753
SO2 226 0.041 0.047 0.123
NOx 233 0.027 0.203 0.257
Total 0.247 0.728 1.133

The Pace University study [15] suggests that XCs for solar technologies are between 0 and
0.53 ¢/kW h. This range is adopted here with the average used as the best estimate. The XC
represents the lost opportunity cost of the land as derived from real estate prices and is on the
order of the land use XCs of fossil fuel power plants. Manufacturing energy input is an
additional source of externalities in the PV electricity generation fuel cycle. For other generation
technologies, the air pollution damages associated with plant construction were found to be
small or negligible compared to other externalities [21,22]; however, the processing of materials
for PV power plants is relatively energy intensive. Emission factors used in this study for PV
power plant manufacturing are presented in Table 7. They are based on the average German
energy mix and technology efficiency [21] and the plant life and capacity factors (CFs) presented
in Table 2. Using the DCs presented in Section 3.2, the XCs presented in Table 7 were calcu-
lated as follows:
XC ¼ DC  EF0  C1  C3  C4 (6)

5. Non-environmental externalities

Neither government expenditures in acquiring and ensuring supplies of fossil fuels nor the
risks to national economies of relying on foreign markets are fully accounted for in the current
electricity pricing process. Based on their study results, Coiante and Barra [23] suggest that the
US government spends more than 100% of the fuel cost paid by utilities and consumers to
secure oil resources. Therefore, the energy security supply XC for oil is made equal to the fuel
cost. For natural gas, these authors suggest that ‘‘diplomatic and military external assistance to
producer countries,’’ along with some upstream emissions, can easily lead to a 100% cost
increase with respect to the market fuel price [23]. Moreover, natural gas is a resource for which
the US is becoming increasingly dependent on imported supplies. In order to prevent overlap
with upstream emissions quantified as air pollution, the energy security supply XC for natural
gas technologies is made equal to the fuel cost minus the upstream emissions XC. Coal supplies
benefit from internal assistance, which is discussed qualitatively below, but since the US pro-
duces all of the coal it consumes no energy security supply XC is assigned to this fuel resource.
Other energy security externalities include depletion of the world’s finite resources and the
resulting limitation of future energy supply options. A study by Vollebergh [24] used an XC of
approximately 1.14 ¢/kW h for depletion of fossil fuel resources. This value is applied to all fos-
2136 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144

sil fuel power plants examined in this study, and a range of 0.57–2.29 ¢/kW h is used. Table 6
displays these and other non-environmental XCs.
Modular, distributed electric power plants save transmission and distribution (T&D) costs
and can obviate the need for extensions to the power grid, thereby avoiding visual and electro-
magnetic impacts. In addition, by greatly reducing line losses, which effectively lower the
efficiency of electricity generation and consumption, distributed power plants create fewer
‘‘downstream emissions.’’ On the other hand, electric power is typically transmitted over large
distances, and since T&D and downstream emissions are part of the fuel cycle, they are treated
as externalities. Based on values reported by DPCA [25], a best estimate of 2.0 ¢/kW h is
applied to all non-distributed generators, with a range of 0.7–7.0 ¢/kW h.
MSW and LFG Recovery generators are typically restricted as to their location and capacity,
and although suitable distributed sites may be available for small wind turbine capacity, wind
farms are often located in remote locations with small electrical loads. Hence, distribution XCs
are assigned to these three technologies.

6. Unquantified externalities

Krupnick and Burtraw [22] examined two factors that are significant in full cost comparisons
of electricity generation technologies: impacts on employment patterns and fiscal effects. The
latter pertains to the various fiscal effects, such as differences in tax payments that can favor cer-
tain generation technologies. A very complex model would be required in order to incorporate
these and other energy subsidies and there is a considerable range in their valuation [26]. How-
ever, neglecting employment and fiscal externalities effectively assigns them XCs of zero. Studies
indicate that employment impacts are potentially large compared to other XCs presented here
[22] and also that subsidies are significant enough to influence electricity resource selection [26];
therefore, their importance is acknowledged in order to encourage their consideration on a
qualitative basis.

7. External cost results

The total XCs associated with each of the 14 generation technologies are shown in Fig. 1.
The results presented use best estimate values for DCs and XCs. In Fig. 1, the category ‘‘other
externality costs’’ encompasses CH4, N2O, upstream emissions, CO, PM, and VOC’s, as well as
water-related impacts, impacts on wildlife, and externalities not otherwise categorized. XCs vary
greatly among the different generation technologies, with newer, cleaner technologies having the
lowest total XC values.
Conventional Oil Boilers have the highest total XCs at 12.86 ¢/kW h. Their EFs are generally
not as high as Coal Boilers, but their lower air pollution XCs are offset by high energy security
costs.
The XCs associated with Conventional Coal Boilers are also very high (12.07 ¢/kW h).
Although AFBC and IGCC plants have slightly higher fuel conversion efficiencies than Conven-
 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144 2137

Fig. 1. External costs for 14 generation technologies.

tional Coal Boilers, their significantly lower XCs (in the range of 7–8 ¢/kW h) result from their
much lower emissions of SO2 and NOx.
Due to the clean nature of natural gas combustion, Gas Turbines and Combined Cycle plants
have lower environmental XCs than coal and oil technologies. However, their considerable
energy security supply XCs keep their total XCs in the range of clean coal technologies. Simple
and Advanced Gas Turbine generation technologies have total XCs between 9 and 9.5 ¢/kW h,
while Advanced Combined Cycle power plant XCs are almost 2 ¢/kW h lower. The lower XC
results for Advanced Combined Cycle result from this technology’s high fuel efficiency.
The large quantities of conventional and toxic pollutants emitted from MSW plants and their
high heat rate lead to high environmental XCs. However, the landfilling cost offset keeps total
XCs for this technology at a relatively low 5.41 ¢/kW h.
Hybrid Solar Thermal Parabolic Troughs have XCs in the 5.5 ¢/kW h range, with a break-
down similar to those of natural gas-fired plants and a small fraction of XCs associated with
land use and other impacts of solar generators.
For SOFCs facilities, the XCs are lower than they are for any of the conventional fossil fuel
technologies. Energy security externalities account for over 50% of total XCs, but these are fuel-
dependent and would be close to zero if renewable fuels were used. XCs associated with SOFCs
using natural gas are 4.88 ¢/kW h.
LFG Recovery has only the standard 2 ¢/kW h distribution XC associated with it, while
wind turbines have both this cost and a small XC from the ‘‘other’’ category. PV and biomass
plants have very low total XCs (i.e., below 1.50 ¢/kW h). Air pollution externalities associated
with PV power plant manufacturing are 0.73 ¢/kW h. In contrast, the air pollution XCs of fos-
2138 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144

sil fuel combustion technologies are 2.5–11 times as high. For Biomass combustion, damages
from emissions of CO, NOx, and PM represent over 85% of its total XC of 1.30 ¢/kW h.
XCs associated with CO2 emissions are significant (over 1 ¢/kW h) for all technologies that
rely on fossil fuels. While SO2 emissions are very small for natural gas plants and zero for
SOFCs, they make notable contributions to the XCs of coal, oil, and MSW plants (i.e., between
0.15 and 1.72 ¢/kW h). Emissions of NOx are significant for all combustion plants (i.e., between
0.30 ¢/kW h for IGCC and 3.3 ¢/kW h for a Coal Boiler without emissions controls), but small
for SOFCs (<0.01 ¢/kW h). In the ‘‘other environmental externalities’’ category, land use XCs
are the most consequential, with values between 0.26 and 0.53 ¢/kW h.
While often omitted from externality studies, XCs associated with energy security and distri-
bution make important contributions to the total XC of the technologies examined. In the
absence of distribution costs, total XCs for non-distributed generation technologies would be
between 15% and 100% lower, while the omission of energy security XCs would lower the costs
of all fossil fuel technologies by 1.1–4.3 ¢/kW h.
In summary, results of XC calculations produce the following ranking of generation tech-
nologies, in ascending order of total XC: PV, Biomass Combustion, LFG Recovery, Utility
Scale Wind Turbines, SOFC, MSW, Hybrid Solar Thermal Parabolic Troughs, IGCC,
Advanced Combined Cycle, AFBC, Advanced Gas Turbines, Simple Gas Turbines, Conven-
tional Coal Boilers, and Conventional Oil Boilers. Using multicriteria analysis, Mirasgedis and
Diakoulaki [27] compared the environmental impacts of a smaller number of energy sources
and produced findings similar to the present study. In ascending order of their environmental
impact, Mirasgedis and Diakoulaki ranked wind, natural gas, oil, and then coal. If non-environ-
mental externalities are excluded from the present study, the same ranking is achieved.

8. Levelized cost of energy

Using a full cost LCC approach to compare generation technologies, the focus of afford-
ability is the overall cost over the lifetime of the plants. Results from the LCOE calculations are
listed in Table 8 and presented in Fig. 2 for each of the 14 generation technologies. Graphical
results use best estimate values for XCs.
Levelized capital costs make up a large fraction of the LCOE for renewable technologies,
while for fossil fuel plants they represent a lesser fraction of total costs. For all plants, capital
costs are highly dependent upon capacity factor. Levelized O&M costs are a small fraction of
the LCOE for most technologies. They are especially low for fossil fuel plants and highest for
SOFCs.
Plants with low heat rates have relatively low fuel costs, while some renewables and LFG
Recovery have none. Levelized fuel costs are driven by energy conversion efficiency and the cost
per unit of fuel consumed. A comparison of Conventional Coal Boilers and Advanced Gas Tur-
bines demonstrates the impact of the price paid for the fuel consumed. Despite similar heat
rates, Advanced Gas Turbines have a much higher levelized cost of fuel because of the high
price of natural gas relative to coal. Low heat rates lead to both low fuel consumption and low
emissions, and consequently to low levelized fuel and XCs. Hence, the LCOE approach rewards
energy efficiency, especially when externalities are incorporated.
Table 8
Levelized costs with ranges (¢/kW h)
Levelized Conven- Adv Fluid- Integrated Conven- Simple Advanced Advanced Mass Burn Landfill Solid Utility PV: Utility Hybrid Biomass
cost tional ized Bed Gasifi- tional Cycle Gas Gas Tur- Combined MSW Gas Oxide Fuel Scale Wind Scale Flat Solar
Boiler Combus- cation Boiler Turbine bine Cycle Recovery Cells Turbine Plate Thermal
(Coal)— tion Comb (Oil)— Parabolic
Rankine Cycle Rankine Trough
Cycle Cycle

Capital 2.81 3.52 3.28 2.15 10.05 0.82 0.91 9.63 3.30 2.42 5.74 49.53 20.27 3.54
costs
O&M costs 1.00 1.73 0.93 0.40 1.24 0.42 0.31 4.22 0.99 6.55 1.66 1.00 3.00 2.59
Fuel costs 1.06 1.04 0.95 3.22 3.47 3.29 2.11 5.16 0.00 2.29 0.00 0.00 1.07 2.75
XCs: Best 12.07 8.21 6.94 12.86 9.46 9.15 7.31 5.41 2.00 4.88 2.13 0.99 5.51 1.30
estimate
Lower 4.45 2.86 2.64 6.03 4.62 4.45 3.46 7.70 0.70 2.75 0.70 0.25 2.38 0.41
range
Upper 27.08 18.83 16.91 27.66 19.33 18.85 16.47 27.93 7.00 8.39 7.13 1.66 13.44 9.28
range
LCOE: 16.94 14.50 12.10 18.64 24.22 13.68 10.65 14.10 6.29 16.15 9.54 51.51 29.86 10.17
Best
estimate
Lower 9.32 9.15 7.81 11.81 19.38 8.98 6.79 0.99 4.99 14.01 8.11 50.77 26.72 9.29
range
I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144

Upper 31.95 25.12 22.07 33.44 34.09 23.39 19.81 36.62 11.29 19.65 14.54 52.18 37.78 18.15
range
2139
2140 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144

Fig. 2. LCOE for 14 generation technologies.

With the exception of Simple Cycle Gas Turbines, fossil fuel combustion plants have total
levelized XCs that exceed all other levelized costs combined. The inclusion of externalities
approximately triples the LCOEs for these technologies. In their study, Coiante and Barra [23]
produced similar results. They used a LCC model that incorporated externalities to determine
the relative cost increase to build and operate a ‘‘new type of clean hydrocarbon-fueled power
station’’ that will reduce all emissions by 90%. They conclude that the ‘‘real cost of energy’’ for
such a power plant can be estimated at approximately three times that of ‘‘the present conven-
tional production cost.’’
In contrast, XCs account for a small fraction of the LCOE for wind, solar, and biomass
facilities. This suggests that by ignoring externalities in decision-making and policy-planning
processes, the current system inappropriately advantages power generated from fossil fuel
resources.
The generation technology with the lowest LCOE is LFG Recovery (at 6.29 ¢/kW h). For
this technology, fuel is considered free and XCs are low. At 9.54 ¢/kW h, the cost of electricity
generated from Utility Scale Wind Turbines is also low. XCs for this technology are small and
fuel costs are zero, but distribution costs are quite significant. Biomass facilities are very com-
petitive (10.17 ¢/kW h), with low XCs due to the sustainability of the fuel resource.
Advanced Combined Cycle, the most efficient fossil fuel technology examined, has a low
LCOE of 10.65 ¢/kW h. When not used as part of a Combined Cycle plant, Advanced Gas
Turbines have a higher LCOE (13.68 ¢/kW h).
AFBC and IGCC have LCOEs of 14.50 and 12.10 ¢/kW h, respectively. While capital costs
are similar for all coal technologies, the cleaner emissions and higher efficiencies of IGCC and
 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144 2141

AFBC result in lower XCs than Conventional Coal Boilers. Oil Boilers have a high LCOE
(18.64 ¢/kW h); as with Conventional Coal Boilers, this is due to very high XCs.
Because they are modeled as peaking plants and therefore have a low capacity factor, Simple
Cycle Gas Turbines have high capital costs and an LCOE over 24 ¢/kW h. MSW facilities have
high capital costs, O&M costs, and XCs. However, the negative fuel cost (5.16 ¢/kW h) offsets
these other costs, bringing the LCOE down from 19.26 to 14.10 ¢/kW h. This is competitive
with clean coal technologies and less expensive than conventional fossil fuel plants.
SOFCs have higher capital costs than Combined Cycle plants but lower XCs due to their very
low emissions. Their high levelized O&M costs are a result of their requiring a stack replace-
ment approximately every 5 years [28]. The LCOE for SOFCs is 16.15 ¢/kW h. As the tech-
nology matures, levelized capital and O&M costs are expected to drop, making these generators
more cost competitive.
The high levelized capital costs of solar technologies keep their LCOEs among the highest of
those examined in this study. However, with a rapidly growing market and advancing tech-
nology, their capital costs are expected to drop significantly over time.
The LCOE results indicate that in electricity generation portfolios that focus on long-term
economic feasibility and sustainability, renewable energy power plants are key technologies. If
the XC portion of the LCOE is omitted from the results presented here, it is clear that the
inclusion of externalities greatly impacts the relative attractiveness of the generation options.
With XCs excluded, Combined Cycle plants have the lowest LCOE, followed by Landfill Gas
Recovery and Advanced Gas Turbines. In this scenario, oil and coal technologies are also low-
cost options. Hence, a full cost approach incorporating externalities makes fossil fuel tech-
nologies less attractive in relation to other generation technologies.

9. Uncertainty analysis

In order to show the influence of the range of XCs on the relative attractiveness of the 14
generation technologies, Fig. 3 presents their LCOEs using no externalities, best estimate values,
and lower and upper range values. Fig. 3 shows that LCOEs for generation technologies with
low XCs change little across ranges applied. On the other hand, for fossil fuel and MSW plants,
which have high levelized XCs, uncertainty in the DCs of air pollutants can potentially double
or halve the LCOE in some cases. In the no-externality case, fossil fuel technologies are highly
attractive but as XCs increase, their fuel intensity and emissions raise their LCOEs well above
those of wind, LFG, biomass, and SOFCs. Advanced Gas Turbines and Combined Cycle plants
remain very competitive as a result of their fairly clean emissions and their high efficiencies.
Although there is considerable uncertainty in the valuation of externalities, even the lower
range scenario alters the traditional ranking of generation technologies, with some renewable
technologies becoming more affordable than many coal and oil technologies. This scenario also
makes renewable energy more attractive in relation to natural gas combustion and places LFG
Recovery ahead of Advanced Combined Cycle. In the upper range scenario, SOFCs and Utility
Scale Wind Turbines become more affordable than Advanced Combined Cycle generation, and
the cost gap between renewable and coal technologies increases. Regardless of the range
applied, LCOE results indicate that renewable energy power plants and fuel cells are more
2142 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144

Fig. 3. LCOE ranges for 14 generation technologies.

affordable in relation to fossil fuel technologies than conventional economic models predict.
Moreover, results suggest that renewable energy and clean fossil fuel technologies are crucial
resources in a sustainable electricity generation portfolio.

10. Conclusions

The following conclusions are drawn from the results of this study:

. Including externalities in a full cost, levelized approach to determine the affordability of elec-
tricity generation options has a major impact on their relative attractiveness.
. Newer, cleaner technologies have low XCs (some less than 1.5 ¢/kW h), while older fossil fuel
technologies have XCs over 12 ¢/kW h.
. When externalities are considered, renewable electricity generation is comparable in cost to
fossil fuel generation.
. Costs associated with externalities are generally higher and more subject to uncertainty for
fossil fuel technologies with high emission rates than for cleaner, more efficient technologies.
. Externalities should not be overlooked in efforts to develop optimal electricity resources.

References
[1] Carlin J. Environmental externalities in electric power markets: acid rain, urban ozone, and climate change.
Energy Information Administration; 1998 [Available from: http://www.eia.doe.gov/cneaf/pubs_html/rea/
feature1.html].
 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144 2143

[2] Energy Information Administration. Electricity generation and environmental externalities: case studies.
Washington (DC): U.S. Department of Energy; 1995.
[3] Kozloff K, Dower R. A new power base: renewable energy policies for the nineties and beyond. World Resour-
ces Institute; 1993.
[4] California Energy Commission. Energy technology status report, 1996. Sacramento (CA): CEC; 1998.
[5] Roth I. Life-cycle costing of electric power generation plants in New England: a complete fuel cycle approach
incorporating externalities. Amherst (MA): University of Massachusetts Amherst Engineering Department; 1999.
[6] Energy Information Administration. Annual energy outlook, 2001. Washington (DC): U.S. Department of
Energy; 2000.
[7] California Energy Commission. Electricity cost by generation technology, Sept 1981. Sacramento (CA): CEC;
1981.
[8] California Energy Commission. Comparative cost of California central station electricity generation technologies.
Sacramento (CA): CEC; 2003.
[9] U.S. Department of Energy. Market-based advanced coal power systems. Report no. DOE/FE-0400.
Washington (DC): U.S. DOE; 1999.
[10] McGowan J, Connors S. Review of renewable energy technologies for Massachusetts: status report. Amherst
(MA): University of Massachusetts Amherst Engineering Department; 1998.
[11] U.S. Environmental Protection Agency. 5th ed.. Compilation of air pollutant emission factors, AP-42 vol. 1.
U.S. EPA; 1995 [Supplements A, B, C, D, E, F, update 2001 and update 2002].
[12] U.S. Environmental Protection Agency. Estimating greenhouse gas emissions, EIIP document series vol. VIII.
Washington (DC): U.S. EPA; 1999 [Prepared by ICF Inc].
[13] Northeast States for Coordinated Air Use Management, Massachusetts Executive Office of Environmental
Affairs, Massachusetts Divisions of Energy Resources. Massachusetts greenhouse gas inventory. Boston (MA):
1996.
[14] Energy Information Administration. Electric power annual, 1992. Washington (DC): U.S. Department of
Energy; 1994.
[15] Pace University Center for Environmental Legal Studies. Environmental costs of electricity. New York: Oceana
Publications; 1990.
[16] National Resources Defense Council. Choosing clean power: bringing the promise of fuel cells to New York.
NRDC; 1997 [Available from: http://www.nrdc.org/nrdcpro/ccp/chapII.html].
[17] DOE-Fossil Energy techline. Solid oxide fuel cell reaches one year of operations. U.S. DOE; 2000 [Available
from: http://www.fe.doe.gov/techline/t1_sofcdemo.html].
[18] Bernow S, Marron D. Valuation of environmental externalities for energy planning and operations: May 1990
update. Boston (MA): Tellus Institute.
[19] Rowe R, Lang C, Chestnut L. Critical factors in computing externalities for electricity resources. Resource and
Energy Economics 1996;18(4):363–94.
[20] Palmer K, Krupnick A, Dowlatabadi H, Siegel S. Social costing of electricity in Maryland. The Energy Journal
1995;16(1):1–26.
[21] Kaltschmitt M, Stelzer T, Wiese A. Ganzheitliche bilanzierung am beispiel einer bereitstellung elektrischer ener-
gie aus regenerativen energien. Zeitschrift fur Energiewirtschaft 1996;20(2):177–97.
[22] Krupnick A, Burtraw D. The social costs of electricity: do the numbers add up?. Washington (DC): Resources
for the Future; 1996.
[23] Coiante D, Barra L. Practical method for evaluating the real cost of electrical energy. International Journal of
Energy Research 1995;19(2):159–68.
[24] Vollebergh H. Environmental externalities and social optimality in biomass markets: waste-to-energy in the
Netherlands and biofuels in France. Energy Policy 1997;25(6):605–21.
[25] Distributed Power Coalition of America (DPCA). Benefits of distributed power. 1999 [Available from: http://
www.dpc.org/utilities.html].
[26] Koplow D, Dernbach J. Federal fossil fuel subsidies and greenhouse gas emissions: a case study of increasing
transparency for fiscal policy. Annual Review of Energy and the Environment 2001;26:361–89.
2144 I.F. Roth, L.L. Ambs / Energy 29 (2004) 2125–2144

[27] Mirasgedis S, Diakoulaki D. Multicriteria analysis vs. externalities assessment for the comparative evaluation of
electricity generation systems. European Journal of Operational Research 1997;102(2):364–79.
[28] Ellis M. Fuel cells for building applications. Atlanta: American Society of Heating, Refrigerating, and Air-Con-
ditioning Engineers, Inc; 2002.