You are on page 1of 11

ARTICLE

pubs.acs.org/EF

Economic Optimization of a CO2-Based EGS Power Plant


Aleks D. Atrens,* Hal Gurgenci, and Victor Rudolph
The University of Queensland, Queensland Geothermal Energy Centre of Excellence, St. Lucia, Queensland 4072, Australia

ABSTRACT: CO2-based enhanced geothermal systems (EGSs) have been examined from a reservoir-oriented perspective, and as a
result thermodynamic performance is well explored. Economics of the system are still not well understood, however. In this study,
the economics of the CO2-based EGS technology is explored for an optimized power plant design and best-available cost estimation
data. We demonstrate that near-optimum turbine exhaust pressure can be estimated from surface temperature. We identify that
achievable cooling temperature is an important economic site consideration alongside resource temperature. The impact of time
required to sufficiently dry the reservoir prior to power generation is also addressed. The role of sequestration as part of CO2-based
EGS is also examined, and we conclude that if fluid losses occur, the economic viability of the concept depends strongly on the price
associated with CO2.

1. INTRODUCTION The results from the economic estimation provide a baseline


Examination of the CO2-based enhanced geothermal system process cost for a reference case, and a comparative analysis with
(EGS) concept has focused to-date on technical aspects, pre- other cases is used to examine the trade-offs and guide economic
dominantly reservoir considerations,18 and process modeling optimization of CO2-based EGS, in particular, the following:
and design.9,10 Ultimately, the viability of the concept depends • thermodynamic and economic optimization of compressor
on economics, and the ultimately preferred design will be one usage, in particular selection of a turbine exhaust pressure;
that is economically optimized. A preliminary analysis of eco- • the trade-off during site selection between achievable cool-
nomics for the system has previously been conducted.11 That ing temperatures and resource temperatures;
analysis found the following: • the trade-off between resource temperature and resource
• There were economic arguments to proceed with further depth;
analysis. • the implications of CO2 reaction or dispersion in the
• Economics were highly tentative due to uncertainties of well reservoir in conjunction with CO2 costs;
cost estimates. • and the impact of dry-out times on economic performance.
• Larger wellbore diameters were economically favored. The examination of these aspects provides conclusions with
• Inclusion of compressors was economically favored. regard to the economic viability of the CO2-based EGS con-
• CO2-based EGSs should be focused on climates with lower cept, and implications for site selection and process design. The
ambient temperatures. numerical dollar values for costs are shown for comparative
This work extends that economic analysis using new data on purposes only: ‘real’ plant costs require close attention to what is
the cost of increasing well diameter for well cost estimation and included inside the estimate battery limits, among other con-
considering a number of additional issues. Analysis is accom- siderations. As discussed further in section 3.1.4, modification of
plished by finding the expected cost per kilowatt of electrical the costing basis can lead to a change in economic performance
generation capacity for a nominal plant size of 50 MWe net, of (50%, but conclusions based on comparison between differ-
operating within typical site constraints. A conceptual sketch of ent operating points or designs should be preserved.
the system is shown in Figure 1. The system consists of a number
of injection/production well doublets—each including one 2. METHODOLOGY
injection well (points 12), a reservoir (point 23), and a
The economic calculations for this work are based on the thermo-
production well (point 34)—and a surface plant—consisting
dynamic model of the CO2-based EGS system, which has been
of a turbine (point 45), a compressor (point 67), and air-
comprehensively described elsewhere.10 That model has been modified
cooled heat exchange equipment (points 56 and 71). The to incorporate a compression system and additional cooling unit in the
number of production and injection well doublets is increased surface plant. The calculation approach is as follows:
linearly to allow sufficient total fluid flow to achieve the 50 MWe (1) Site conditions are defined, typically reservoir impedance, re-
net target. Reservoir arrangements of greater complexity (for servoir pressure (P3), flow pathway/geometry, depth, tempera-
example, multiple coupled wells) are not considered in this work ture, and injection temperature (T1). Injection temperature is the
because the precise details regarding reservoir design and man- temperature achievable as a result of cooling and depends on the
agement depend decisively on the specific details and geology of process arrangements (e.g. wet or dry cooling system, optimized
a reservoir. The surface equipment is considered as a single set of heat exchanger design) and the local ambient conditions.
operating units. Note that this analysis does not include binary
power generation systems (those using a separate power con- Received: April 8, 2011
version loop), as a result of direct power conversion being a Revised: June 16, 2011
major incentive for using CO2.9 Published: June 21, 2011

r 2011 American Chemical Society 3765 dx.doi.org/10.1021/ef200537n | Energy Fuels 2011, 25, 3765–3775
Energy & Fuels ARTICLE

Table 1. Reference Parameters for Site


depth (m) 5000
reservoir length (m) 1000
reservoir temp (°C) 225
injection temp (°C) 25
min. reservoir width (m) 0.73
max. reservoir width (m) 250.73
reservoir impedance (MPa 3 s/L) 0.2
corresponding k.H (m3) 8.603  1011
reservoir pressure (MPa) 49.05
wellbore roughness (m) 0.0004
wellbore diameter (m) 0.231
isentropic efficiency, ηisen 0.85

6 The turbine exhaust pressure (P5) is varied until the minimum cost
per kilowatt of net electricity generation capacity is found; this
provides the optimal cost for the selected mass flow rate.
Figure 1. Conceptual diagram of the system. 7 This process is then repeated for different mass flow rates until an
overall economic profile across a range of flow rates (or their
corresponding injection pressures) can be found for the set of
(2) A CO2 mass flow rate (m) is selected. constraints defined in step 1.
(3) Injection wellhead pressure (P1) is calculated to be sufficient for Similar calculations are completed for different site conditions or
the pressure at the reservoir-production well interface (P3) to be economic assumptions. The list of site conditions considered, and the
equal to the specified reservoir pressure. numbers used for the reference case can be found in Table 1; we use the
O Injection well flow is adiabatic: change in enthalpy is equal to same reference case as in previous works10 for ease of comparison.
the change in gravitational potential energy and kinetic energy, Symbol nomenclature is provided in Table 2. Reservoir length is the
and change in pressure is equal to the static pressure change distance from the edge of the injection to the edge of the production
adjusted for pressure decline due to friction. well. Reservoir thickness is accounted for in the reservoir impedance
O Reservoir flow follows Darcy’s law, with the flow path width term. Reservoir pressure is assumed to be equal to the hydrostatic head.
linearly increasing from a minimum width at the wellbore to We use a constant well diameter for simplicity, although past EGS well
a maximum width at the midpoint of the reservoir, and with completions to date have typically included 7 and 95/8 in. sections.
CO2 temperature increasing linearly from the bottom of the 2.1. Thermodynamic Calculations. Thermodynamic perfor-
injection well to the bottom of the production well. mance is calculated from the model based directly on the thermody-
(4) The pressure and temperature at the bottom of the production namic states of the system:
well (P3, T3) are set at the reservoir conditions. Production
wellhead pressure (P4) and production wellhead temperature WT ¼ ðnW =2Þmη
_ isen ðh4  h5 Þ ð4Þ
(T4) are calculated on the basis of mass flow rate.
O Production well flow is adiabatic: change in enthalpy is equal to WC ¼ ðnW =2Þm=η
_ isen ðh7  h6 Þ ð5Þ
the change in gravitational potential energy and kinetic energy,
and change in pressure is equal to the static pressure change
and pressure decline due to friction. QHX1 ¼ ðnW =2Þmðh
_ 6  h5 Þ þ ð1  ηisen ÞWT ð6Þ
(5) A turbine exhaust pressure (P5) is selected to calculate heat
and power flows in surface equipment to use as a basis for QHX2 ¼ ðnW =2Þmðh
_ 1  h7 Þ þ ð1=ηisen  1ÞWC ð7Þ
cost estimation. Turbine exhaust gases are cooled prior to the
All thermodynamic properties are calculated using Helmholtz free
compressor so that T6 is equal to the injection temperature
energy equations of state for CO2, as are transport properties.12,13 All
(T1). The number of injection/production well pairs is set
work flows are in kilowatts. The number of wells, nW, is divided by 2 to
sufficient that net power production WNET is greater than
account for doublet pairs.
50 MWe.
The parasitic power consumption of the heat exchangers is calculated
O Net work is determined from a summation of surface plant
as a linear function of heat flow. Air-cooled heat exchangers are typically
power flows:
fan-forced, and while there is interest in utilizing natural-draft dry
WNET ¼ WT  WC  WHX1  WHX2 ð1Þ cooling as a means of reducing parasitic electricity losses, we examine
only the typical implementation of fan-forced systems.
O Total capital cost of the system is determined from a summa- The power consumption of fan-forced dry cooling can be estimated
tion of individual components: through a variety of different measures. We use values reported by the
Electric Power Research Institute14 for an inlet air temperature differ-
CTOT ¼ CT þ CC þ CHX1 þ CHX2 þ nW CW ð2Þ ence of 22 °C, where a parasitic load of approximately 18.9 kWe per
1 MWth was noted. Therefore, for calculation of the parasitic power
O Capital cost per unit of net power (Γ), the measure of merit for consumption by the heat exchanger fans, we use eq 8:
a given design, is calculated based on these values:
WHX ¼ εQHX ð8Þ
Γ ¼ CTOT =WNET ð3Þ with ε taking a value of 0.0189.

3766 dx.doi.org/10.1021/ef200537n |Energy Fuels 2011, 25, 3765–3775


Energy & Fuels ARTICLE

Table 2. Nomenclature to current pricing structures. This equates to a rise in costs of 67.7% from
2003 to 2009. The function we use for well cost estimation is therefore eq 9:
A area for heat transfer
"  2 #
CTOT total plant capital cost D
CT total turbine cost
CW ¼ 1:09 0:6662
n
þ 0:3338 Kebz ð9Þ
D0
CC total compressor cost
CHX total heat exchanger cost where n is the number of years from 2003 to 2009, per-well cost CW is
C0 base equipment cost calculated in millions of 2003 USD, D0 is 0.23125 m, z is the well depth, and
K and b are parameters given in Table 3. Effect of cost uncertainty and
D well diameter
variability is shown in section 3.1.4.
D0 reference case well diameter
2.2.2. Turbine. Our earlier assessments incorporated a cost estimation
F heat exchange orientation factor method for CO2 turbines that accounted for the higher density of CO2 as
FM equipment material factor the turbine fluid compared to typical power conversion fluids.9,11 The
FP equipment pressure factor equation to calculate the cost for the turbine is given in eq 10:
FS additional cost factor
C0T ¼ 3:5  RWTβ Fγ5 ð10Þ
h enthalpy
k.H Reservoir permeability multiplied by reservoir height where C0T is the base cost of the turbine, WT is the turbine work output,
m mass flow rate R and β and γ are constants, and F5 is the turbine exhaust density (see
nW number of wells Figure 1). This equation was fitted to the costs of steam turbines and
n number of years
CO2 turbines estimated in the previous work.9 Cost estimation for CO2
turbines is complex and a number of other approaches could be used,
Pi pressure at different points in the system (as in Figure 1)
including rigorous turbine design or estimation using different equip-
QHX heat exchanger heat transfer rate ment comparisons (e.g., other gas expanders, axial compressors). We use
Ti temp at different points in the system (as in Figure 1) eq 10 as it relatively simply links fluid density to turbine size, as has been
U overall heat transfer coefficient encountered with other organic fluids.18
WNET net work The values for the parameters used in this fit are given in Table 3.
WT turbine work Total cost in 2009 USD for the turbine is given in eq 11:
WC compressor work CT ¼ 525:7=575:4  FS  C0T ð11Þ
WHX heat exchanger fan parasitic power losses
CT is the cost of the turbine updated to represent 2009 USD, and FS is
z well depth
an additional factor to account for material, additional piping, control,
ΔTM average heat transfer difference
freight, labor, and other overheads.
ε heat exchanger parasitic power factor Turbine costs are given in the previous work in 2008 USD;11 they
Γ specific capital cost are updated to 2009 USD by the ratio of Chemical Engineering Plant
ηisen isentropic efficiency Cost Indices (CEPCI) for the relevant years: 2008 = 575.4; 2009 =
Remaining algebraic terms are cost correlation 525.7.19
constants defined in Table 3. 2.2.3. Compressor. The centrifugal compressor cost is based on
standard compressor costing methods.20 Equations 12 and 13 were
used to estimate compressor cost:
2.2. Cost Estimation. Cost estimation of the different process
units is discussed below. CC ¼ 525:7=397  2:8FS C0C ð12Þ
2.2.1. Wells. Well costs are estimated on the basis of typical cost 8 9
estimation methods, using an exponential increase in cost as a function < 10ðR1 þ R2 log WC þ R3 ½log WC 2 Þ
, WC < 3000 =
of depth. This has been modified with a parabolic scalar to account for C0C ¼
cases with increased well diameter. The relation for the cost increase due : WC =3000  10ðR1 þ R2 log 3000 þ R3 ½log 30002 Þ
, WC g 3000 ;
to diameter change is based on larger diameter wells drilled in New
ð13Þ
Zealand,15 where a reported change in cemented casing diameter from
95/8 to 133/8 in. increased costs by 17%. A parabolic scaling of costs Note that the factor of 2.8 in eq 12 is to account for material (carbon
in response to diameter is utilized as per our previous work,11 based steel), compressor type (centrifugal), as well as additional piping, freight,
on another work,16 which noted the proportionality of drilling times labor, etc. Carbon steel is appropriate, as water condensation does not
(and therefore expected cost) to the square of the diameter. occur under the pressure and temperature conditions, for suitably low
Considerable uncertainty remains regarding estimates of geothermal water content in CO2 from the production well. Uncondensed water
well costs, both in terms of the cost relationship to depth, and in dissolved in CO2 will not instigate corrosion.21 Additionally, C0C is
response to change in diameter. The uncertainty of the cost relationship reported in 2001 USD, so it is updated to 2009 USD based on the ratio of
with depth is mainly due to uncertainty in cost reduction due to CEPCI for the relevant years: 2001 = 397; 2009 = 525.7.19
maturation of geothermal drilling technologies. We adopt as our basis 2.2.4. Heat Exchangers. The base costs of the air-cooled heat
for well cost estimation the mathematical average of the following: exchangers are estimated from standard costing methods.20 Costing is
• Joint Association Survey (JAS) Oil & Gas well cost average based on air-cooled heat exchangers; in some cases, water cooling will be
(as derived elsewhere17); available. In these cases, the cost of cooling systems will be significantly
• To-date cost average for EGS geothermal wells, scaled for increases reduced. The cost of heat exchangers is estimated from eq 14:
in diameter.17
CHX ¼ 525:7=397ðB1 þ B2 FM FP ÞFS C0HX ð14Þ
As the well costs are reported in 2003 USD, we scale them to 2009
USD by assuming a constant increase per annum of well costs of 9%. where CHX is the heat exchanger cost, B1 and B2 are constants for an
2009 USD are used for all cost calculations to make results more relevant equipment type, FM is the material factor (for stainless steel), FP is the

3767 dx.doi.org/10.1021/ef200537n |Energy Fuels 2011, 25, 3765–3775


Energy & Fuels ARTICLE

Table 3. Calculation Parameters


constant value equation justification units

ηisen 0.85 4, 5, 6, 7 see text


ε 0.00825 8 14
K 0.2865 9 see text 2009 USD
b 6.657  104 9 see text m1
D0 0.23125 9 see text m
R 1.066 10 9 USD kW0.5439
m0.4416 kg0.1472
β 0.5439 10 9
γ 0.1472 10 9
FS 1.7 11, 12, 14 20
R1 2.2897 13 20
R2 1.3604 13 20
R3 0.1027 13 20 Figure 2. Pressure factor for air-cooled heat exchanger versus pressure.
B1 0.96 14 20
B2 1.21 14 20 extrapolated. The resulting extrapolation is eq 17:
FM 2.9 14 20
FP ¼ 0:939P0:04759 ð17Þ
K1 4.0336 15 20
K2 0.2341 15 20 The result of this is shown in Figure 2.
K3 0.0497 15 20
The base area for heat exchange, used as a basis for costing, is given by
eq 18:
C1 0.1250 16 20
C2 0.15361 16 20 A ¼ QHX =UFΔTM ð18Þ
C3 0.02861 16. 20 Design of a CO2-based power plant would include economic
F 0.91 18 see text optimization of the heat exchange/cooling system. This would involve
U 300 18 see text W m2 K1 an iterative optimization of heat exchange tube materials, geometry and
ΔTlm 11 18 see text K fins, as well as air flow and fan design, based on site constraints and the
trade-offs between capital cost, ongoing parasitic power consumption,
and CO2 pressure drop within the cooling system. Because heat
pressure factor, and C0HX is the cost for the heat exchanger made from exchanger design is well-understood but also involves a complex set
carbon steel operating at ambient pressure. Stainless steel is selected of trade-offs, we do not include it in this analysis. Instead, we cal-
as the construction material, as it is possible that some trace water and culate economic performance based on heat exchanger parameters
carbonic acid will condense during cooling. The constants are given representative of a near-optimal design. Precise optimization of the heat
in Table 3. exchangers would provide performance benefits, but they are of second
The base cost for carbon steel equipment is given by eq 15: order importance to this study (see section 3.1.4 for a discussion of the
8 9 sensitivity of economic performance to heat exchanger costs). Relevant
< 10ðK1 þ K2 log A þ K3 ½log A2 Þ , A < 10000 = heat exchanger design parameters are discussed below.
CHX ¼
0
Heat flow, Q, is defined for both cooling systems by the equations
: A=10000  10ðK1 þ K2 log 10000 þ K3 ½log 10000 Þ , A g 10000 ;
2

specified in section 2.2. The three parameters U, F, and ΔTM are


ð15Þ therefore critical to the economic efficiency of the heat exchange system.
For this analysis, we use a value for F, the temperature difference
where K1, K2, and K3 are constants for the heat exchanger type, and A is correction factor, of 0.91, which is an estimate for a single-tube-pass air-
the area over which heat exchange occurs in the heat exchanger. The cooled heat exchanger.22 This value would increase toward unity with
constants are given in Table 3. the addition of additional tube rows, but for the purpose of this
For heat exchange areas greater than 10,000 m2, multiple parallel assessment, we assume a conservatively simple design.
units are used (rather than scaling up the size), so there is no cost For the average temperature difference ΔTM, we assume a value of
reduction from economies of scale. 11 K. This is comparable to typical minimum approach temperature
Pressure factors are given by eq 16: differences for air-cooled heat exchangers of 814 K.22 This value is
therefore taken as a conservative but representative value for the average
log P þ ½C3 log P2 Þ
FP ¼ 10ðC1 þ C2 ð16Þ temperature difference for an optimized air-cooled heat exchanger.
Values for the overall heat exchange coefficient, U, are typically
where C1, C2, and C3 are constants for the heat exchanger type, and P is determined experimentally. As values for the optimized heat exchange
the design pressure (in bar) of the equipment. The values of these between CO2 and air are not readily available, we use values for U for
constants are given in Table 3. heat exchange between air and fluids with similar convective heat
The range of pressure factor estimation given in the literature20 is transfer coefficients to that for CO2. CO2 has reported convective
limited to below 100 bar for air-cooled heat exchangers. As some design heat exchange coefficients typically in the range of 24 kW 3 m2 3 K1,
pressures for the CO2 thermosiphon may be slightly above this range, a although this increases substantially for conditions near the critical point,
small extrapolation of these pressure factors is used. The extrapolation is reaching values greater than 18 kW 3 m2 3 K1.23,24 The typical range is
derived from the fit of a power law to the higher-pressure region similar to that of the convective heat transfer coefficients reported for
(i.e., 50100 bar) of the pressure-factor calculation, which is then cooling and condensation of light hydrocarbons: 26 kW 3 m2 3 K1.25

3768 dx.doi.org/10.1021/ef200537n |Energy Fuels 2011, 25, 3765–3775


Energy & Fuels ARTICLE

Figure 4. Power flows in the surface plant versus exhaust pressure, for
Figure 3. Net work, turbine power, compressor power, and specific the optimum injection pressure for one well doublet in the reference
capital cost as a function of injection pressure for the reference case. scenario.

Typical overall heat transfer coefficient, U, values reported for light


hydrocarbon condensation are 250 to 350 W 3 m2 3 K1.26 Therefore,
we adopt a value for U for CO2 of 300 W 3 m2 3 K1.
Variation in these values is incorporated into the discussion of cost
sensitivity in section 3.1.4.

3. RESULTS
3.1. Reference Case. For the initial consideration, we examine
the reference case with parameters as specified in Table 3.
3.1.1. Single Doublet Operation. First we examine results for
the case of a single well doublet. For a single doublet, the
thermodynamic and economic performance can be seen in
Figure 3, as a function of the injection pressure (P1).
Note here that as we examine only a single doublet, the WNET
is not 50 MWe but a much lower value; for the cases considering
50 MWe, sufficient identical parallel doublets are calculated to Figure 5. Cost per kilowatt versus exhaust pressure at 15 °C, 35 °C, and
produce sufficient power output, which is developed in section 55 °C.
3.1.3 and onward.
Figure 3 shows that there are both thermodynamic and of compression work required; at lower turbine exhaust pres-
economic optimums for the system. These optimums are not very sures, compressor work increases rapidly, reducing net power
sensitive to the injection pressure (i.e., (1 MPa from the optimum produced, while, at higher turbine exhaust pressures, the com-
will result in only relatively small variations in performance). pressor work is low, but power produced through the turbine
The economic and thermodynamic optimums are for similar continues to decline. Heat exchanger load is relatively constant,
injection pressures but are not identical due to an economic leading to constant parasitic power consumption.
incentive toward minimizing equipment size. The economic optimum corresponds very closely to the
3.1.2. Turbine Exhaust Pressure Optimization. Exhaust pres- thermodynamic optimum, with some bias toward minimizing
sure can be selected to optimize thermodynamic, or more equipment sizes (the economic performance optimum is at
importantly, economic performance. The results discussed from 8.63 MPa). The similarity between economic optimum and
section 3.1.1 are given for optimized exhaust pressures. If we thermodynamic optimum was found to extend across all the
examine the range of performance at different exhaust pressures surface plant cases we examined.
for a single injection pressure, we can see the manner in which the The optimum exhaust pressure to select for the process was
system behaves in response to changing the turbine exhaust found to be strongly dependent on achievable cooling temperature
pressure parameter. Figure 4 shows the response to turbine (which depends on ambient temperature). The economic perfor-
exhaust pressure of power loads for the turbine, compressor, heat mance at the optimum injection pressure versus exhaust pressure
exchangers, and overall net work produced for the optimum is shown in Figure 5 for three different cooling temperatures.
injection pressure from Figure 3 (13.8 MPa). As seen in Figure 5, the optimum exhaust pressure to maximize
Figure 4 shows a clear thermodynamic performance optimum, economic performance changes as surface temperature changes.
corresponding to an exhaust pressure of 8.4 MPa. In the region This is predominantly due to the thermodynamic behavior of the
near this pressure, at the injection temperature (35 °C), there is a system: at lower temperatures, the region of rapid increase in
rapid change in density. The peak in performance at these fluid density is also at a lower pressure. At lower temperatures,
thermodynamic conditions is due to a change in the gradient the optimum for performance is also more sharply defined, as the
3769 dx.doi.org/10.1021/ef200537n |Energy Fuels 2011, 25, 3765–3775
Energy & Fuels ARTICLE

Figure 6. Optimum exhaust pressure versus temperature for different


reservoir temperatures and depths. Figure 7. Cost per kilowatt versus nominal plant capacity.

change in fluid density is more rapid. The very sharp minimum


visible for 15 °C corresponds to moving the compressor inlet
from the gas side of the liquid vapor envelope to the fluid side
(and a step change in compressor fluid density), drastically
reducing compressor costs.
The optimum conditions for plant optimization—that is,
exhaust pressure as a function of surface temperature (effecti-
vely compressor inlet conditions)—can be plotted for a range of
surface temperature conditions, for a variety of reservoir tem-
peratures and reservoir depths (i.e., different site conditions).
This plot is shown in Figure 6, with an additional line denoting
the liquidvapor saturation curve.
Figure 6 allows determination of the optimum exhaust pres-
sure for a variety of different site conditions. As can be seen in
Figure 6, the optimum exhaust pressure tracks very closely to the
liquidvapor saturation curve until the critical point. At tem-
peratures above the critical point (where the liquidvapor Figure 8. Cost components of the plant for the (A) reference case, (B)
saturation curve ends), the optimum exhaust pressure tends to favorable performance case, and (C) unfavorable performance case.
follow the trend seen below the critical point.
For temperatures below and near the critical point, there
is only a small deviation from this line: the sharp change in Figure 7 shows that there is a large initial reduction in unit costs
thermodynamic performance from operating a condensing cycle for moving from a doublet to a larger, multiwell power plant, but
leads to an economic optimum. As temperature rises above the the effect becomes less pronounced at larger plant sizes. This is
critical point, the optimum in thermodynamic performance be- because large geothermal power plant costs are relatively in-
comes less defined, and so too does the economic optimum (this sensitive to economies of scale: the wells and air-cooled heat
can also be seen in Figure 6 for 55 °C). This leads to a wider spread exchanger fan bays and to a degree the compressors, which
of optimum exhaust pressure points at higher temperature for the represent a large component of the facility costs, increase in
range of different reservoir conditions, as the economic optimum is number as energy output size goes up, and so they scale linearly
less dependent on closely matching the thermodynamic optimum. with cost. Turbine costs do provide some benefit from econo-
Typically, a near-optimum performance point for turbine mies of scale.
exhaust pressure for a CO2-based EGS power plant can be 3.1.4. Cost Breakdown and Uncertainty. For the 50 MWe
selected on the basis of Figure 6. The exact optimum can then plant used as the reference case, the overall cost components are
be refined on the basis of the reservoir conditions for the site, as given in Figure 8. Also shown in Figure 8 are breakdowns of the
well as the economic conditions for the project. Operating at the costs of different equipment components for the same plant
critical point may require specially designed equipment as a result design basis under a set of more favorable costing assumptions
of rapid thermodynamic property change. and under a set of less favorable assumptions. This is not
3.1.3. Scale-up to 50 MW. To examine the effect of scaling the intended to be an extended sensitivity study on the effects of
plant size up to larger facility sizes, we examine the change in component costs, as they are mostly self-evident, but it may
performance as we move from a single well doublet to a 50 MWe provide an indication of the overall variability due to costing
plant and then to a 300 MWe plant. The minimum capital assumptions. The changes used for these different cases are
expenditure per kilowatt (from Figure 3) for a single doublet and detailed in Table 4.
for increases in plant size is plotted versus net capacity in As shown in Figure 8, the total cost of the plant is very sensitive
Figure 7. to the economic calculation basis. The substantive part of the
3770 dx.doi.org/10.1021/ef200537n |Energy Fuels 2011, 25, 3765–3775
Energy & Fuels ARTICLE

Table 4. Modified Performance Characteristics


higher lower
parameter performance performance units

ΔTlm 14 8 K
U 400 200 W m2 K1
K 0.1099 0.4786 $
b 7.804  104 6.132  104 m1
R 1.066 1.3189 USD kW0.5439
m0.4416 kg0.1472
β 0.5439 0.5353
γ 0.1472 0

differences is due to the well cost, where if typical oil and gas well
costs are assumed, the total plant capital expenditure is reduced Figure 9. Capital cost per kWe of net power produced versus injection
by 25%. Changes to the heat exchanger costing assumptions will pressure for well diameters of 0.23125 m, 0.3 m, and 0.4 m.
also impact the overall price of the plant, but by a relatively
modest percentage, which justifies the approach of not under-
taking optimization calculations for those systems.
3.2. Design Optimization: Well Diameter. Here we briefly
revisit the trade-off of increasing well diameter on increased costs
and increased performance using better cost estimation data,
as discussed in section 2.2.1. This balance has been examined
previously,11 favoring larger well sizes than currently contem-
plated. Figure 9 shows the cost per kilowatt of a 50 MWe plant for
different well diameters.
Figure 9 suggests that given the cost data we have used for this
analysis, there is continued justification for increasing well
diameters to improve the performance of CO2-based EGSs.
Performance continues to be more limited by wellbore consid-
erations than by flow in the fluid in-reservoir. It also indicates
that, at least for the reference case, there is unlikely to be a net
economic benefit from increasing the well diameter beyond
0.3 m (1200 ). Larger diameters are likely to be more favorable
for hotter resources, for more permeable resources, or for lower
overall drilling costs. Figure 10. Contours of capital cost in 2009 USD per kilowatt of capacity
3.3. Site Considerations/Optimizations. Sites for CO2- as a function of injection temperature and reservoir temperature.
based EGS should be selected to maximize economic perfor-
mance. For H2O-based EGS, the primary site considerations are 3.3.1. Ambient Temperature versus Reservoir Temperature.
reservoir temperature, depth, and permeability. For CO2-based A contour plot of cost per kilowatt as a function of injection
EGS, reservoir temperature and depth are important, but per- temperature and reservoir temperature for a reservoir depth of
formance is altered less by permeability,10 assuming a perme- 5000 m is shown in Figure 10.
ability representative of past EGS projects can be achieved. Figure 10 shows how the economic performance of CO2-
For CO2-based EGSs, economic performance is substantially based EGSs increases for lower injection temperatures and
benefitted by lower injection temperatures, due to an increase higher reservoir temperatures. The capital cost remains con-
in density in the injection well leading to a higher static pres- stant when injection pressure and reservoir temperature change
sure increase in the wellbore, which leads to greater flow together at a ratio of approximately 1:3. The interpretation is that
throughput.10 Additionally, both compressor inlet tempera- every degree lower injection temperature (i.e., lower ambient
ture and the injection temperature are assumed to be set by a temperature, since these are coupled by the approach tempera-
minimum approach temperature to ambient (defined by detailed ture of the heat exchangers) has an equivalent value of 3 degrees
heat exchanger optimization, but explored in this work over the of higher reservoir temperature. Comparatively, on a simple
range 814 K). Therefore, the economic performance of CO2- economic basis, a site in an area with a high ambient temperature,
based EGSs is sensitive to injection temperature, reservoir say 30 °C, compared to a temperate site, say 20 °C ambient,
temperature, and depth. Site selection should take into account would need to have reservoir temperatures about 30 °C hotter to
the relative effects of these key characteristics on performance. compensate.
We assess the relative importance of these characteristics by 3.3.2. Reservoir Temperature versus Depth. Deeper wells
examining the economic performance of CO2-based EGSs for generally access higher resource temperatures, leading to higher
different reservoir temperatures and injection temperatures for a power generation per well, but they also incur greater well costs.
constant depth of 5000 m in section 3.3.1 and for different To properly optimize power plant design, it is necessary to select
reservoir temperatures and depths for a constant injection the optimum depth for economic performance. For a defined
temperature of 35 °C in section 3.3.2. site, this depends on the local temperaturedepth relationship.
3771 dx.doi.org/10.1021/ef200537n |Energy Fuels 2011, 25, 3765–3775
Energy & Fuels ARTICLE

Table 5. Levelized Cost of Electricity Parameters


parameter value

capacity factor 90%


inflation rate 3%
operating costs 1¢ 3 kWh1
interest rate 8%
equity rate 15%
equity proportion 50%
debt proportion 50%
debt payback period 10 yrs
plant lifetime 30 yrs
depreciation schedule MAR ACH 15 year accelerated
operating cost escalation rate 1%
tax rate 38%

Figure 11. Contours of cost in 2009 USD per kilowatt of capacity as a


function of reservoir temperature and reservoir depth.

For a generic overview of the optimization between temperature


and depth, it is necessary to examine a broad range of values for
temperature and depth. Figure 11 shows contours of cost per
kilowatt of capacity for a 50 MWe CO2-based EGS power plant
as a function of reservoir temperature and depth. All reservoir
pressures are equivalent to hydrostatic pressure at the specified
depth; permeability does not vary with depth, and injection
temperature is 35 °C.
As shown in Figure 11, the capital cost decreases monotoni-
cally as reservoir temperature increases at constant depth. At a
constant reservoir temperature, variation of capital cost with
depth has a minimum at depths between 2000 and 3000 m. This
is distinctly different from the behavior of H2O-based EGSs,
where deeper reservoirs of similar temperature and geological Figure 12. Levelized cost of electricity as a function of dry-out time.
conditions are always less economically favorable.
The nonlinear behavior of CO2-EGS costs in response to
depth is not intuitive. It occurs because the CO2-based EGS relies operating life (i.e., does not include those initial two years of
on high densities in the reservoir for low in-reservoir pressure capital expense/construction).
drops, and large well lengths to provide a large net buoyant 3.4.1. Dry-out Time. Dry-out time is unique to CO2-based
force from the difference in densities between injection and EGSs and represents the time required to displace initial
production wells.10 If CO2 is used to extract heat from a very reservoir fluids such that produced CO2 is sufficiently water-free
shallow reservoir (of reference impedance), much more of the to directly drive a turbine.6 It can be represented in an economic
energy extracted is used to push the low density CO2 through the model as a delay after construction before the power plant starts
reservoir instead of being converted to net electrical power generating electricity and, therefore, revenue. During the dry-out
generation, which leads to higher costs per-unit of power. period, water would be removed from produced fluids either by
3.4. Time-Dependent Economic Performance. There are physical separation after cooling or using desiccants (e.g., zeolite,
two CO2-based EGS features that do not affect capital costs but alumina). Power could potentially be generated during this
will affect economic performance because of their impacts over period with a binary power cycle. We assume the reference case
the lifetime of the power plant. These features are dry-out time system and examine the effect on LCoE of delaying power
and sustained fluid losses in the reservoir. To encompass the generation for up to five years after construction of the power
time-dependent behavior of these features, their impact on plant. The results are shown in Figure 12.
levelized cost of energy (LCoE) is assessed. Changes in resource It is apparent from Figure 12 that, for the reference case, the
temperature are neglected for transparency and simplicity. levelized cost of electricity under the assumptions in Table 5 will
The costing methodology discussed in section 2 provides an be approximately 24¢ 3 kWh1. LCoE increases as dry-out time
estimate of the capital cost of the power plant. This is translated increases in a slightly nonlinear manner whereby larger dry-out
into a levelized cost of electricity by the same approach used by times are increasingly more detrimental.
MIT in a report on nuclear power.27 A detailed description of the 3.4.2. Fluid Loss and CO2 Sequestration. Fluid loss is present
method can be found in that source. The assumptions used for for water-based systems but is more complex for the CO2-based
that model are given in Table 5. EGS. Short-term H2O-based EGS trials have reported steady
In the scenarios we examine, we assume that the capital state fluid losses sustained over months of operation as high as
expenditure occurs in equal portions over two years and that 30%.28 These losses are particularly pertinent to CO2-based EGS
no power is produced during that time. The plant life refers to the because of their potential to contribute to sequestration of CO2.
3772 dx.doi.org/10.1021/ef200537n |Energy Fuels 2011, 25, 3765–3775
Energy & Fuels ARTICLE

parameters, particularly those affected by well costs (e.g., target


depth), will depend primarily on economic performance and
cannot be estimated from thermodynamics alone. This analysis
examined a limited range of parameters. In particular, the impact
of permeability, which is a significant characteristic for H2O-
based systems, was not assessed. This is much less important for
CO2-based EGS performance, but further research on this topic
is worthwhile in order to evaluate whether that assessment
remains legitimate over the full range of potential values for
other site parameters.
The range of the parameters assessed in this work was also quite
limited and may not be suitable for extrapolation. For example, for
lower temperature geothermal resources, it may become favorable
to target much shallower systems than the depths examined here.
The simultaneous interaction between multiple site and process
parameters was not comprehensively considered. Instead the effect
Figure 13. Levelized cost of electricity as a function of CO2 cost for of changes in individual or pairs of parameters from the reference
different steady state fluid loss rates.
case was assessed. This means that the behavior of the system
where multiple parameters are simultaneously changed from the
To examine the impact of fluid losses, we assess LCoE as a reference case is unquantified by this work, since there may be
function of the price of CO2 for two scenarios: fluid losses of 5% interactions. However, we expect that the characteristic behavior
and 10%, sustained over the lifetime of the power plant. Price for presented here is representative for parameter values within
CO2 represents the cost of supplying CO2 at the boundary of the bounds typically encountered for geothermal reservoirs.
power plant and could be estimated as market price for CO2 An isentropic efficiency of 85% was assumed for both the
emissions minus per-tonne costs of sequestration and transport turbine and compressor. That value is an estimate for well-
of the CO2 to the geothermal site. Negative values represent a developed CO2 turbomachinery technology. Currently, CO2
payment for the service of CO2 sequestration. Figure 13 shows compression is relatively well understood, but CO2 turbines
the LCoE versus CO2 cost for the two scenarios. are not currently commercial. Lower isentropic efficiencies will
Figure 13 shows that a value associated with CO2 can reduce thermodynamic and economic performance of the system
significantly impact the LCoE for the CO2-based EGS power compared to the results discussed here and bias the optimization
system. Expenses associated with procuring CO2 dramatically of the system toward reduced compressor usage and higher
increase the LCoE, but income from CO2 disposal makes the turbine exhaust pressures.
concept much more economically feasible. We estimate 5% fluid Details of reservoir development were not addressed in this
losses from a 50 MW power plant would be on the order of 23 work. We have assumed linear increases in the number of
million tonnes per annum of CO2 stored underground. injection and production well doublets to achieve sufficient
power output for a 50 MWe system. Favorable operation of a
real system should incorporate a reservoir development plan that
4. DISCUSSION would account for the interaction between wells and, con-
The results in this paper depend on the assumptions and sequently, change the ratio of injection to production wells as
parameters used. Uncertainty in these parameters and assump- the scale of the system increases. Injection is comparatively easier
tions leads to uncertainty in results and consequently limits the than fluid production,10 and so one might generally expect fewer
conclusions that may be drawn. The effects of various assump- injection wells compared to production wells. Exploiting these
tions and parameter choices on results and conclusions are considerations and particular geological features of an individual
discussed below. reservoir is likely to result in an overall improvement in perfor-
Well costs are highly variable and can change for individual mance relative to the results reported here (for the same overall
wells due to difficulties while drilling, for a reservoir system due resource specifications).
to geological characteristics, and for an industry depending on We have examined the economic impact of CO2 prices only
technology and drilling experience. For this work, a single well from underground fluid losses. A CO2 price will also have
costdepth relationship was used, nominally representing the immediate economic consequences because large quantities of
industry average well cost for EGSs after additional experience CO2 will be required to displace fluids (i.e., water) that are
accumulation and technology development. The impact on initially in the reservoir. There is a high degree of uncertainty
economic performance of underestimating or overestimating regarding the total volume that would be required. Preliminary
drilling costs was assessed in section 3.1.4. The impact of well calculations indicate that this may run to millions of tonnes of
costs on optimization of system parameters is unclear. It is CO2 per well doublet for fluid displacement,6 and CO2 supply,
unlikely to substantively affect turbine exhaust pressure optimi- price (or credit), and logistical considerations will dramatically
zation or the importance of low injection temperatures. Different affect economic performance for quantities of that magnitude.
well costs will substantially alter the optimal reservoir depth, A sequestration or disposal value for CO2 would provide an
however. immediate revenue boost for such a project, making this tech-
There is a close correspondence between thermodynamic and nology highly competitive with conventional power generation
economic optima. This implies that, for a preliminary design of a and other sequestration projects. Additional exploration of
CO2-based EGS, turbine exhaust pressure and injection pressure economics regarding the opportunities and issues for combined
can be estimated from thermodynamic performance. Other sequestration and power generation is warranted.
3773 dx.doi.org/10.1021/ef200537n |Energy Fuels 2011, 25, 3765–3775
Energy & Fuels ARTICLE

5. CONCLUSIONS (2) Pruess, K. On production behaviour of enhanced geothermal


systems with CO2 as working fluid. Energy Convers. Manage. 2008,
In this work, the economics and related thermodynamics of
49, 1446–1454.
CO2-based EGSs have been examined. Optimization of some (3) Pruess, K.; Azaroual, M. On the feasibility of using supercritical
system parameters and site considerations has been addressed. CO2 as heat transmission fluid in an engineered hot dry rock geothermal
A number of conclusions can be drawn from the results: system. In Thirty-first workshop on geothermal reservoir engineering;
• Significant reductions in capital expenditure per unit of Stanford University: Stanford, CA, 2006.
power capacity can be achieved from moderate increases in (4) Pritchett, J. W. On the relative effectiveness of H2O and CO2
the scale of the power plant. as reservoir working fluids for EGS heat mining. GRC Trans. 2010,
• Large diameter wellbores continue to have some thermo- 33, 235–239.
dynamic and economic justification, but on the basis of (5) Apps, J.; Pruess, K. Modeling geochemical processes in enhanced
current cost information, benefits are small. This may change geothermal systems with CO2 as heat transfer fluid. In Thirty-Sixth
Workshop on Geothermal Reservoir Engineering; Stanford University:
as the cost effects of large diameters are better understood.
Stanford, CA, 2011.
• Ambient temperature conditions, through their influence on (6) Atrens, A. D.; Gurgenci, H.; Rudolph, V. In Removal of water for
the injection temperature, remain a major factor in thermo- carbon dioxide-based EGS operation. In Thirty-Sixth Workshop on
dynamic and economic performance. Geothermal Reservoir Engineering, Stanford University, Stanford, CA,
• Including compression as part of the system design leads to 2011; Stanford University: Stanford, CA, 2011.
substantial improvement in thermodynamic performance, (7) Wan, Y.; Xu, T.; Pruess, K. Impact of fluid-rock interactions on
resulting in a benefit to economic performance. enhanced geothermal systems with CO2 as heat transmission fluid. In
• Near-optimal turbine exhaust pressure or compressor inlet Thirty-Sixth Workshop on Geothermal Reservoir Engineering; Stanford
pressure can be specified as a function of injection tem- University: Stanford, CA, 2011.
perature alone. (8) Magliocco, M.; Kneafsey, T. J.; Pruess, K.; Glaser, S. Laboratory
experimental study of heat extraction from porous media by means of
• There is an economic trade-off during site selection between
CO2. In Thirty-Sixth Workshop on Geothermal Reservoir Engineering;
surface ambient temperature (through its decisive influence Stanford University: Stanford, CA, 2011.
on injection temperature) and reservoir temperature, (9) Atrens, A. D.; Gurgenci, H.; Rudolph, V. CO2 thermosiphon for
whereby economic performance is approximately constant competitive power generation. Energy Fuels 2009, 23 (1), 553–557.
for a decrease in reservoir temperature of 3 K if injection (10) Atrens, A. D.; Gurgenci, H.; Rudolph, V. Electricity genera-
temperature is also reduced by 1 K. tion using a carbon-dioxide thermosiphon. Geothermics 2010, 39 (2),
• There is an optimum economic reservoir depth, around 161–169.
23 km, due to thermodynamic benefits of deeper wells (11) Atrens, A. D.; Gurgenci, H.; Rudolph, V. Economic analysis of a
offsetting the additional well cost up to a point. CO2 thermosiphon. In World Geothermal Congress; Bali, Indonesia,
• Capital costs of CO2-based EGSs are dominated by well costs. 2010.
(12) Vesovic, V.; Fenghour, A.; Wakeham, W. A. The Viscosity of
• If substantial fluid losses are present, system economics are
Carbon Dioxide. J. Phys. Chem. Ref. Data 1998, 27 (1), 31–44.
dominated by CO2 price. (13) Span, R.; Wagner, W. A new equation of state for carbon
• CO2 costs may be of significant importance even without dioxide covering the fluid region from the triple-point temperature to
fluid losses due to the need to displace pre-existing reservoir 1100 K at pressures up to 800 MPa. J. Phys. Chem. Ref. Data 1996, 25 (6),
fluids. 1509–1596.
A generalized conclusion can also be drawn about broad (14) EPRI. Air-cooled condenser design, specification, and operation
economic viability. Implementation of CO2-based EGS technol- guidelines; Electric Power Research Institute: Palo Alto, CA, 2005.
ogy relies on one of the following: (15) Bush, J.; Siega, C. Big bore well drilling in New Zealand—a case
study. In Proceedings World Geothermal Congress 2010; Bali, Indonesia,
• Supply of CO2 in abundance at low cost.
2010.
• Utilization of an existing CO2-filled reservoir suitable for (16) Kaiser, M. J. A survey of drilling cost and complexity estimation
geothermal power generation. models. Int. J. Pet. Sci. Technol. 2007, 1 (1), 1–22.
• Combination with a carbon-capture and sequestration project. (17) Augustine, C.; Tester, J. W.; Anderson, B.; Petty, S.; Livesay,
Given fulfillment of one of the above, this work presents a basis B. A comparison of geothermal with oil and gas well drilling costs. In
for comparison of potential sites for this technology, and initial Thirty-First Workshop on Geothermal Reservoir Engineering; Stanford
parameter estimation for plant design. University: Stanford, CA, 2006.
(18) Sauret, E.; Rowlands, A. S. Candidate radial-inflow turbines and
high-density working fluids for geothermal power systems. Energy
’ AUTHOR INFORMATION 2011, 36 (7), 4460–4467.
Corresponding Author (19) Lozowski, D. Economic Indicators. Chem. Eng. 2010, 117 (1), 63.
*E-mail: aleks.atrens@uq.edu.au. (20) Turton, R.; Bailie, R. C.; Whiting, W. B.; Shaeiwitz, J. A.
Analysis, synthesis, and design of chemical processes, 2nd ed.; Prentice Hall
PTR: Upper Saddle River, NJ, 2003.
(21) Seiersten, M. Material selection for separation, transportation
’ ACKNOWLEDGMENT and disposal of CO2. In CORROSION\2001; NACE International:
We thank the State Government of Queensland for providing Houston, TX, 2001.
the funding that made this research possible. (22) Shilling, R. L.; Bell, K. J.; Bernhagen, P. M.; Flynn, T. M.;
Goldschmidt, V. M.; Standiford, F. C.; Timmerhaus, K. D. Heat-Transfer
Equipment. In Perry’s Chemical Engineers’ Handbook; Perry, R. H., Green,
’ REFERENCES D. W., Maloney, J. O., Eds.; McGraw-Hill: New York, 1997.
(1) Pruess, K. Enhanced geothermal systems (EGS) using CO2 as (23) Son, C.-H.; Park, S.-J. An experimental study on heat transfer
working fluid—A novel approach for generating renewable energy with and pressure drop characteristics of carbon dioxide during gas cooling
simultaneous sequestration of carbon. Geothermics 2006, 35, 351–367. process in a horizontal tube. Int. J. Refrig. 2006, 29, 539–546.

3774 dx.doi.org/10.1021/ef200537n |Energy Fuels 2011, 25, 3765–3775


Energy & Fuels ARTICLE

(24) Oh, H.-K.; Son, C.-H. New correlation to predict the heat
transfer coefficient in-tube cooling of supercritical CO2 in horizontal
macro-tubes. Exp. Therm. Fluid Sci. 2010, 34 (8), 1230–1241.
(25) Park, K.-J.; Jung, D.; Seo, T. Flow condensation heat transfer
characteristics of hydrocarbon refrigerants and dimethyl ether inside a
horizontal plain tube. Int. J. Multiphase Flow 2008, 34, 628–635.
(26) Ulrich, G. D.; Vasudevan, P. T. Chemical engingeering process
design and economics: a practical guide, 2nd ed.; Process Publishing:
Durham, NH, 2004.
(27) MIT. The Future of Nuclear Power; 2003.
(28) Murphy, H.; Brown, D.; Jung, R.; Matsunaga, I.; Parker, R.
Hydraulics and well testing of engineered geothermal reservoirs.
Geothermics 1999, 28, 491–506.

3775 dx.doi.org/10.1021/ef200537n |Energy Fuels 2011, 25, 3765–3775

You might also like