Applied Energy 228 (2018) 1837–1852

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Hybridizing a geothermal power plant with concentrating solar power and T

thermal storage to increase power generation and dispatchability
Joshua D. McTiguea, Jose Castrob, Greg Mungasc, Nick Kramerc, John Kingc, Craig Turchia,
⁎
Guangdong Zhua,
a
National Renewable Energy Laboratory, 15013 Denver West Parkway, Golden, CO 80401, USA
b
Coso Operating Company, CA, USA
c
Hyperlight Energy, Ltd., San Diego, CA, USA

H I GH L IG H T S

• The power output of a geothermal plant is increased by 2 MW (9%) by using solar heat.
e

• The hybrid plant’s solar thermal energy to electricity conversion efficiency is 24%.
• The hybrid plant cost (LCOE) is 25–50% lower than a PV array with battery storage.
• The cost is up to 15% lower than a concentrating solar plant with thermal storage.
• Over-sizing the solar field and using storage improve flexibility and reduce cost.

A R T I C LE I N FO A B S T R A C T

Keywords: Geothermal power plants often deploy less than their full power capacity due to declining geothermal resources.
Concentrating solar power Integrating heat from a concentrating solar power (CSP) collector field increases the power output at low cost.
Thermal energy storage This article considers five methods of solar heat addition in a double-flash geothermal plant. The most promising
Geothermal power solution converts solar heat into electrical work with an efficiency of 24.3%. The economic feasibility and
Retrofit
optimal sizing of the solar field and thermal stores are evaluated. A hybrid plant that increases power generation
Levelized cost of energy
from 22 to 24 MWe has a Levelized Cost of Electricity (LCOE) of 0.07 ± 0.01 $/kWhe. Adding three hours of
Hybrid power generation
storage increases the LCOE to 0.08 ± 0.01 $/kWhe. Photovoltaic systems are considered to be a low-cost re-
newable technology, but an equivalent photovoltaic system with battery storage costs 0.15 ± 0.07 $/kWhe due
to the high cost and replacement rate of batteries compared to thermal storage. The hybrid plant also has a lower
LCOE than a conventional CSP plant. If the dispatchability that thermal storage provides is rewarded with higher
electricity prices, calculations indicate storage becomes an attractive investment when discharged power re-
ceives 1.75 times the typical price of electricity.

1. Introduction and motivation
A geothermal resource typically experiences a reduction in the
temperature, pressure, or mass flow rate of its production fluids over
time, leading to decreased power generation and underutilized equip-
ment. If a Power Purchase Agreement exists plant operators may be
subject to additional fees for not producing agreed quantities of power
[1]. The extent of any resource decline is unique to that geothermal
field. A survey of geothermal power plants in California and Nevada
found that flash plants typically experienced a temperature decrease of
0.8% per year, while the value was 0.5% for binary plants [2].
Underperforming plants may be brought back to full capacity by the
addition of thermal heat. The unused capacity of the geothermal plant
provides an opportunity to install a concentrating solar plant at reduced
cost since investment in a power unit and condenser are not required.
Integrating thermal storage provides the plant with flexibility and dis-
patchability. Several novel contributions are made in this article. For
instance, the technical performance of several plant configurations are
compared. The solar field can be over-sized thereby allowing the hybrid
plant to operate at its design point for a greater portion of the year, and
the optimal solar field size is found. Excess heat can be sent to the
geothermal plant or stored, thus reducing curtailment. The impact of

⁎
Corresponding author.
E-mail address: Guangdong.Zhu@nrel.gov (G. Zhu).

https://doi.org/10.1016/j.apenergy.2018.07.064
Received 11 April 2018; Received in revised form 1 June 2018; Accepted 14 July 2018
0306-2619/ © 2018 Published by Elsevier Ltd.
J.D. McTigue et al. Applied Energy 228 (2018) 1837–1852

Nomenclature ηsol solar efficiency, %, see Eq. (13)

θ⊥, θ transversal and longitudinal incidence angles
Roman symbols ρ reflectance
ρf,s fluid or solid density, kg/m3
cp specific heat capacity, kJ/kg·K σ solar multiple
Ccap capital cost, $ τ transmittance
E energy, kJ τd discharging time of stores, s
h enthalpy, kJ/kg ϕ heat loss, W/m
ṁ mass flow rate, kg/s
M annual operations cost, $ Abbreviations
p pressure, bar
P power, W BES battery energy storage
Q̇ heat, W CSP concentrating solar power
s entropy, kJ/kg·K DPO eutectic mixture of diphenyl oxide and biphenyl
t time, s FCR fixed charge rate
T temperature, °C HTF heat transfer fluid
V volume, m3 IAM incidence angle modifier
Ẇ work, W IRR internal rate of return
LCOE levelized cost of electricity, see Eq. (14)
Greek symbols PV photovoltaics

α average absorptance Subscripts and superscripts

β oversizing factor for storage tanks
γ intercept factor h, c hot and cold
ε packed bed void fraction in, out inlet and outlet flows
η1 first law efficiency, %, see Eq. (12)
ηopt optical efficiency, %

storage on system cost is investigated, and the conditions for storage to
be economical are explored. Furthermore, the hybrid plant economics
are compared to conventional systems, and are found to be competitive.
This article concentrates on the integration of solar heat into a
double-flash geothermal power plant: production fluids are flashed and
the resulting steam powers a steam turbine. Hybrid plants require good
solar and geothermal resources, and Fig. 1 indicates that the most
suitable locations in the United States are in the West, and are primarily
dependent on the geothermal resource location.
The installed geothermal capacity in the USA is 3134 MWe, most of
which is concentrated in California and Nevada, see Fig. 1. Flash plants
comprise 29% of US geothermal capacity, and notably 90% of US flash
plant capacity is over 25 years old, indicating that many existing plants
may be experiencing the effects of resource decline. The opportunity to
explore resource decline mitigation currently exists in the US, and the
results may have benefits internationally as flash plants make up 65%
of the world’s total installed capacity of 11 929 MWe.
The first geothermal/solar hybrid plant was developed at the
Stillwater Power Plant, Nevada, and began operation in 2015. A solar
field of parabolic trough collectors supplies 17 MWth of thermal power
to the 33 MWe geothermal power plant and augments production by 2
MWe [4]. The plant also includes a 26 MWe photovoltaic solar array,

Fig. 1. Potential locations for geothermal-solar hybrid plants and design requirements. Map of the USA showing locations of high solar irradiance and geothermal
temperatures at a depth of 3000 m. Data from National Solar Radiation Database [3].

1838
J.D. McTigue et al.
which offsets the decrease in power generation that occurs at high
ambient temperatures [5]. Operators of the 13.2 MWe Gümüşköy geo-
thermal binary plant in Turkey aimed to reduce the detrimental impact
of high ambient temperatures by using solar heat to augment genera-
tion [6]. Tests were conducted with a 200 kWth array of parabolic
trough collectors in 2014. Other previous investigations have con-
centrated on either hybrid systems using binary plants [7–9] and
single-/double-flash plants [10–13].
Heat generated by the solar field can be added at numerous points
in the geothermal system. The optimal hybrid configuration depends on
the location of components, the properties of the geothermal fluids, and
the performance of the plant. One of the first hybrid plants was pro-
posed in 1979 and used geothermal heat for feedwater heating, and
solar heat to generate steam for a steam turbine [14]. At the time, these
hybrid plants demonstrated no economic advantages over stand-alone
geothermal plants. However, CSP prices have dropped significantly and
it is appropriate to reconsider this technology.
Lentz and Almanza [10] investigated using solar collectors to di-
rectly heat the geothermal fluids, thereby increasing its enthalpy and
the steam generation. The fluids could either be heated before or after
the first flash tank. However, the authors noted that this method was
limited by silica deposition in the collectors, and is a more significant
issue if the collectors are placed after the flash tank as the brine has a
higher salt concentration.
Four methods for adding solar heat to the Ahuachapán geothermal
field in El Salvador to increase the power output by 2 MWe to 97 MWe
have been considered [11,12]. Heating the brine directly from the
production well was determined to require too large a heat exchanger.
A method that was successfully tested involved heating brine from the
first separator and directing the generated steam into the high-pressure
turbine. The authors calculated that increasing power output by 2 MWe
would require a solar field of 30 acres (121,405 m2) to heat the heat
transfer fluid (HTF) to 225 °C. Cardemil et al. [13] compared two lo-
cations of solar-heat addition in single- and double-flash plants in the
Atacama Desert, Chile. After the flash tank, the solar field could either
superheat the steam, or evaporate the brine. Using 2nd Law analysis, it
was concluded that superheating the steam was slightly more effective
because the turbine operated more efficiently with dry steam. Double-
flash systems were also found to be preferable to single-flash plants.
Hybridization of solar heat with binary geothermal plants typically
involves different integration methods to flash plants, although pre-
heating the geothermal fluids has also been investigated in this context
[15]. Other strategies involve superheating the binary cycle working
fluid [16] or using the solar heat to drive a topping cycle – either as part
of the binary plant [7] or in a steam turbine [17], thereby making better
use of relatively high-exergy solar heat. These studies find the LCOE to
be relatively high, obtaining values such as 0.163 $/kWhe [17] and
0.22–0.43 $/kWhe [16]. However, estimated solar field costs in the
range of 250–300 $/m2 are used, which are high compared to recent
estimates which are closer to 150–200 $/m2. This is particularly re-
levant given that the solar field cost dominates the total capital cost.
1.1. Scope and novelty of the present work
In this article, several methods of integrating solar heat into a
geothermal power plant are investigated, and the cost of the novel
system is compared to conventional renewable technologies. A ther-
modynamic model of the system is first described in Section 2. The
geothermal power plant is based on the Coso geothermal plant, and the
model has been validated against operational data. Several methods of
integrating the solar heat with the geothermal plant are considered in
Section 3 and annual calculations are undertaken in Section 4. Previous
studies of hybrid systems typically do not include thermal storage or
consider over-sizing of the solar field. ‘Over-sizing’ the solar field is
commonplace in concentrating solar applications, otherwise the system
operates below its design point for significant portions of the year.
1839
Applied Energy 228 (2018) 1837–1852
However, over-sizing can lead to the curtailment of excess solar heat if
sufficient storage is not available. Retrofitting a geothermal plant with
solar means that curtailment can be reduced as excess solar heat is sent
to the geothermal plant. Storage can increase the quantity of energy
dispatched, and the flexibility of power generation, and a two-tank li-
quid thermal energy storage is used here. In this article, the optimal
sizing of the thermal storage and solar field is investigated, and the
influence of these parameters on one another is discussed. In Section 6,
the levelized cost of electricity of the hybrid plant is compared to an
equivalent photovoltaic array with batteries, and a stand-alone CSP
plant, and therefore assesses the relative economic feasibility of the
hybrid system. In Section 7, the effect of the electricity price on optimal
storage size is investigated. Alternative storage systems that are more
compact and cost effective than two-tank liquid storage are discussed in
Section 8.
2. Modeling of the hybrid system
The geothermal plant is a double-flash power plant: steam produced
from geothermal fluids in a flash tank is diverted to the high-pressure
stage of a steam turbine. The brine enters a second flash tank, and the
produced steam is injected into the low-pressure steam turbine stage.
The power plant is based on the units at the Coso geothermal field, and
was modelled in the IPSEpro heat balance software, developed by
SimTech [18]. Design and off-design performance were validated
against data provided by the Coso Operating Company who operate the
double-flash geothermal plant in China Lake, California.
The hybrid plant consists of a double-flash geothermal power plant,
and array of mirrors to concentrate sunlight, and two liquid tanks that
contain the heat transfer fluid (HTF) and are used as thermal storage.
2.1. Geothermal production fluids
The geothermal well produces a fluid with a characteristic curve.
The fluid pressure is inversely related to the mass flow rate, and is as-
sumed to follow a choked flow correlation [19]. Data points are tabu-
lated in Table 1. Intermediate points are evaluated with cubic spline
interpolation.
The values of pmax and ṁ max were set at pmax = 30 bar,
ṁ max = 125 kg/s in order to obtain the design power output (30 MWe)
of the geothermal plant, and to match the fluid properties around the
cycle. The current operating point is 75% of the design output (22.5
MWe) and was obtained by reducing the maximum mass flow rate to
ṁ max = 112 kg/s. The geothermal fluid was assumed to have a specific
enthalpy of 1500 kJ/kg which is in line with typical enthalpies ob-
served at Coso. The salt content of geothermal production fluids is low,
and the fluid can therefore accurately be modelled using water prop-
erties.
Table 1
Pressure-mass flow characteristic for a
geothermal well.
ṁ / ṁ max p/ pmax
0.000 1.00
0.135 0.99
0.339 0.98
0.508 0.96
0.678 0.90
0.763 0.80
0.950 0.55
0.980 0.38
0.990 0.19
1.000 0.00
J.D. McTigue et al.
2.2. Steam turbine
Off-design behavior of the steam turbine is modelled using Stodola’s
ellipse which relates the mass flow rate, and the inlet and outlet pres-
sures through [20]:
ṁ ( T01 )/ p01 = k [1−(p0e / p01 )2]1/2
where ṁ is the mass flow rate, T01 is the inlet temperature, p01 is the
inlet pressure, p0e is the outlet pressure, and k is a constant of pro-
portionality. Assuming that k is constant for all cases, off-design per-
formance may be related to design performance with the following
relation [21]:
ṁ ⎛ T01 ⎞ p01, ref [1−(p0e / p01 )2]1/2
=
ṁ ref ⎜ T01, ref ⎟ p01 [1−(p0e, ref / p01, ref )2]1/2
⎝ ⎠
A variety of physical mechanisms lead to entropy generation in the
steam turbine, as described in Ref. [22]. Rather than undertaking de-
tailed stage-by-stage calculations, stages are grouped into sections, and
correlations of each section efficiency are used to predict turbine per-
formance. These correlations are based on experimental results from
turbine units, and from stage-by-stage calculations. The correlations
were originally developed by Spencer-Cotton-Cannon [23,24], and
computer code was developed in [25] and implemented by IPSEpro.
The steam turbine is primarily operated with ‘sliding pressure’
control where the inlet pressure floats in response to the varying mass
flow rate in accordance with Stodola’s ellipse to maintain a constant
volumetric flow rate. It is assumed that the outlet pressure is fixed by
the condenser pressure and is constant. The condenser is water cooled
and its pressure therefore does not vary significantly with ambient
temperature. This was found to be consistent with operational data
from Coso power plant.
The turbine is constrained to a maximum inlet pressure 10% greater
than the design value. If more mass is available, a valve before the
turbine is throttled to maintain the inlet pressure at its maximum value
(which leads to larger losses). The excess steam is flashed in the second
flash tank, and more steam enters the low-pressure turbine stage.
2.3. Flash tanks
The flash tanks are assumed to be isenthalpic and to separate the
vapor and liquid with 100% efficiency. In practice, typical separation
efficiencies are of the order of 99.9% [26].
2.4. Heat exchanger
Solar heat is transferred from a relatively hot heat transfer fluid
(HTF) to the cooler geothermal fluid in a heat exchanger. The heat
exchangers are modelled using a simple energy balance, such that
ṁ c (hc, out −hc, in ) = ṁ h (hh,in−hh,out )
where h is the enthalpy and is a function of temperature and pressure.
The geothermal fluid inlet temperature Tc, in is known (from the geo-
thermal model), and its outlet temperature Tc, out and pressure are
specified. The hot outlet temperature is given by assuming Th, out = Tc,
in + ΔT, where ΔT = 25 °C. The hot inlet temperature is specified; two
cases are examined in this paper, Th, in = 300 °C and Th, in = 400 °C. It is
assumed that the solar field heats the HTF up to a constant temperature,
Th, in which is achieved by varying the mass flow rate of fluid through
the solar collectors.
2.5. Solar field model
The available solar resource was given by TMY3 (Typical
Meteorological Year) data for China Lake, California, and was obtained
from the System Advisor Model (SAM) [27]. TMY3 presents solar data
Applied Energy 228 (2018) 1837–1852
at hourly intervals that is representative of long-term trends and the
expected range of weather conditions for that location. Using this data
therefore accounts for short-term and seasonal variations in solar irra-
diance.
The solar field uses linear Fresnel reflector (LFR) technology to
concentrate sunlight. The LFR arrays are based on those designed by
(1) Hyperlight Energy Ltd., and the optical and thermal efficiencies are
based on Hyperlight Energy’s calculations and experiments. The optical
efficiency ηopt of an LFR array may be written as
ηopt = γρτα·IAM(θ⊥, θ∥) (4)
where γ is the intercept factor, ρ is the reflectance, τ is the transmit-
tance, α is the average absorptance of the receiver surface, IAM is the
incident angle modifier, and θ⊥, θ are the transversal and longitudinal
(2) incidence angles respectively. The incidence angles are calculated using
the FirstOPTIC method developed in Ref. [28]. The IAM is a function of
these angles and further details are provided in Ref. [29].
Heat transfer losses from the solar receiver are calculated using the
empirical correlations similar to those in Ref. [30] and has the form:
4
ϕ = 0.141Tabs + 6.48 × 10−9Tabs (5)
where ϕ is the heat loss in W/m, and Tabs is the temperature in °C.
However, the correlations have been updated to be more applicable to
the LFRs developed by Hyperlight Ltd.
Heat transfer fluid (HTF) fills the absorber pipe and surrounding
pipework in the solar field. The volume of HTF per m2 of mirror area is
calculated as 3.34 × 10−3 m3/m2 using dimensions of Hyperlight’s LFR
array. There is a single absorber tube with a diameter of 90 mm for a
pair of basins. The basin supports the mirrors. A basin has a length of
15.2 m and a mirror area of 72.5 m2. The HTF volume per m2 of mirror
is therefore 6.67 × 10−4 m3/m2. This value is then multiplied by a
factor of 5 to represent the additional HTF that is required to fill the
pipelines.
The solar field inlet and outlet temperatures are assumed to be kept
constant by varying the HTF mass flow rate through the collectors.
Therefore, as the solar irradiance increases, so does the mass flow rate.
The size of the solar field is quantified using the ‘solar multiple’, σ. A
CSP field with a solar multiple of one can provide the rated thermal
capacity when the Direct Normal Irradiance (DNI) is at its maximum
expected value (∼950 W/m2). Such a field will operate below its design
point for much of the year, and it is conventional to oversize the solar
field. Increasing the solar multiple proportionally increases the solar
field size and its power output. A CSP field with a solar multiple of two
will have double the area, for example. Typical values of solar irra-
diance and temperature for China Lake, California are shown in Table 2
and demonstrate a good solar resource.
2.6. Thermal storage model
(3) The storage system comprises two tanks that contain the liquid HTF
at different temperatures. In this analysis the cold store is at 190 °C and
two different hot stores are compared. Depending on the maximum
operating temperature of the HTF, hot stores at 300 °C and 400 °C are
considered.
Table 2
Typical solar properties at China Lake, California.
Annual average China Lake, CA
2
Direct normal irradiance kWh/m /day 7.75
Global horizontal irradiance kWh/m2/day 5.75
Temperature °C 18.2
Wind speed m/s 3.5
Elevation m 677
Latitude °N 35.68
Longitude °E −117.68
1840
J.D. McTigue et al. Applied Energy 228 (2018) 1837–1852

Each store is governed by a mass and energy balance, given by: where ṁ is the mass flow rate, h is the enthalpy, t is time, and in and out
represent the flows into and out of the store, respectively.
dṁ
= ṁ in−ṁ out The volume of the containers required to store a quantity of energy
dt (6)
E between two tanks at different temperatures is given by
d(mḣ )
= ṁ in h in−ṁ out hout E = (βρf Vcp T )h−(βρf Vcp T )c
dt (7) (8)

COND

a
T
PRE
P
FT1
CND
To cooling
tower
HS
Heat to geothermal plant

CW

FT1
SF HX
FT2

Water
CS Steam
HTF
INJ

b c
18.0
INJ 25
1 st law efficiency, %

INJ2
Solar efficiency, %

COND
17.5 PRE
FT1
20
17.0

16.5 15
0 5 10 15 20 25 0 5 10 15 20 25
Solar heat addition, MWth Solar heat addition, MWth

d e
28 25
Net power, MW e

Net power, MW e

26
24
24

22 23
0 5 10 15 20 25 200 250 300 350
Solar heat addition, MWth Temperature of heat addition,°C

Fig. 2. Solar heat can be added at different points in the geothermal power plant. (a) Schematic of hybrid power plant with solar heat added in five ways: INJ –
injection brine is heated to saturation temperature; INJ2 – same as INJ, except fluid is heated until the vapor fraction is 50%; COND – The condensed water is heated;
PRE – geothermal production fluids are preheated; FT1 – brine at the first flash tank exit is heated. (b) Variation of first-law thermal efficiency with the quantity of
solar heat addition, Qsol. (c) Variation of solar-conversion efficiency with Qsol. (d) Variation of net power with Qsol. (e) Variation of net power with the temperature of
solar heat addition. Key: SF – solar field; HS – hot storage; CS – cold storage; HX – heat exchanger; P – production well; I – injection well; FT1 – flash tank 1; FT2 –
flash tank 2; T – turbine; CND – condenser; CW – cooling water. Valves, pumps etc. are not shown for clarity. In (b, c, and d) the solar field HTF is heated to 190 °C.
The HTF cold temperature is set to be 25 °C greater than the geothermal inlet fluid.

1841
J.D. McTigue et al. Applied Energy 228 (2018) 1837–1852

Table 3 30–40 injection wells. Each well has its own characteristics and fluid
Design and off-design inputs and performance data used to model a two-stage properties which can vary with time and the operation of other wells.
steam turbine unit. Data provided by Coso Operating Company and is re- There are nine 30-MWe turbine sets, and numerous pipelines, headers,
presentative of the Coso geothermal power plant. separators and flash tanks. The simplified model contains one produc-
Design conditions Current operating tion well, one injection well, and one turbine set. Differences between
conditions the model results and the operational data may be attributed to these
simplifications. Other factors that may cause errors are the age-related
High Low High Low
pressure pressure pressure pressure
performance changes of the steam turbines, future changes in the be-
havior of the geothermal wells, changes in climate that affect the solar
Mass flow kg s−1 48.0 25.0 43.8 7.1 resource, and the impact of integrating solar heat into the geothermal
Temperature °C 169.2 132.5 159.8 98.3 plant. While these issues may cause problems with modelling the be-
Inlet pressure bar 6.3 1.4 5.7 0.95
havior specific to Coso, the results should still be representative of a
Gross power MWe 29.6 22.5
Net power MWe 29.4 22.2 typical double-flash geothermal plant.
Efficiency % 19.7 16.9
3. Comparison of hybrid plant configurations

where ρf is the fluid density, V is the volume, cp is the specific heat Concentrating solar and thermal storage are added to a double-flash
capacity, T is the temperature, and β is an additional factor which ac- geothermal power plant which is assumed to be running at 75% of its
counts for the fact that the tanks should be slightly oversized. The design capacity. Fig. 2a indicates the layout of a double-flash geo-
subscripts h and c represent the hot and cold store, respectively. The thermal power plant with a two-stage steam turbine, and the design
storage should deliver a power P for τd hours in order to have stored E. point and current operating conditions are given in Table 3. The CSP
The energy is also given by field concentrates sunlight and heats a thermal oil which transfers its
E = Pτ = ṁ (cp, h Th−cp, c Tc ) τd (9) heat to the geothermal system. The thermal storage uses two liquid-
tanks which contain the same heat transfer fluid as used in the solar
The volume of each store is then given by: collectors. This is the most readily deployable storage solution available
mτ̇ d and has been used in several CSP installations previously [33].
βV = Five methods of adding solar heat to the geothermal plant are
ρf (10)
considered, and a schematic diagram is presented in Fig. 2. In summary,
HTF thermal properties were collected from manufacturers and ty- these are:
pically included specific heat capacity, density, vapor pressure, and
viscosity. Properties vary smoothly with temperature, so intermediate 1. INJ: Add solar heat to brine that was going to be re-injected into the
values are evaluated with cubic spline interpolation. The enthalpy h geothermal reservoir.
and entropy s were evaluated using 2. INJ2: Add solar heat to the re-injection brine until it is 50% vapor.
T2 T2 cp 3. COND: Add solar heat to condenser water.
h= ∫T1
cp dT ; s = ∫T
1 T
dT
(11) 4. PRE: Add solar heat directly to the geothermal production fluids.
5. FT1: Add solar heat to brine that leaves the first flash tank.
A mineral oil and a synthetic oil are considered in this article, and
thermal properties are provided in Appendix A. Water properties are The impact of adding solar heat Qsol on the power output and effi-
calculated using CoolProp software [31]. ciency for these five configurations is presented in Fig. 2b–e. These
figures indicate that Qsol increases power output and first-law efficiency
2.7. Hybrid plant efficiencies η1, while the solar-conversion efficiency ηsol is fairly constant. In-
creasing Qsol increases the steam flow rate that is delivered to the high-
The first-law thermodynamic efficiency of the hybrid system is pressure turbine, thereby increasing the inlet pressure. However, the
given by: characteristic curve of the geothermal well has mass flow rate inversely
̇
Wnet related to pressure [19]. Increasing the turbine inlet pressure (and
η1 =
̇ ̇ therefore the well pressure) requires a lower mass flow rate from the
Q + Qsol (12)
geothermal field, thereby potentially increasing the lifetime of the
̇ is the net work output, Q̇ is the heat addition across the
where Wnet geothermal resource. The additional mass required by the turbine is
power block when no solar heat is added, and Qsol ̇ is the solar heat supplied either by recirculating fluid or increasing steam generation.
addition. Another useful metric is the solar thermal conversion effi- The turbine is operated with ‘sliding-pressure’ control, whereby the
ciency which is defined as the net increase in electrical work divided by inlet pressure and mass vary in response to changes in the flow from the
the solar heat input to the cycle: wells and the solar field. Above the maximum turbine inlet pressure, the
̇ −Wȯ
Wnet turbine inlet is throttled, and a noticeable kink is apparent in the
ηsol = trendlines of Fig. 2b–e. The power continues to increase albeit at a
̇
Qsol (13)
slower rate, but ηsol drops significantly, as does η1. This is a unique
where Wȯ is the work output when no solar heat is added. advantage to retrofitting geothermal plants with solar. In a conven-
tional CSP plant, once the turbine reaches its design point any excess
2.8. Simplifying assumptions and sources of uncertainty solar heat has to be curtailed or stored. However, the hybrid plant has
additional spare capacity in the turbine so that excess heat can be
The off-design performance of the geothermal plant was validated utilized – albeit at a lower efficiency.
against operational data provided by Coso Operating Company for units The five configurations are:
at the Naval Air Weapons Base in China Lake, California [32], with INJ: In normal operation of the Coso geothermal plant, unflashed
model results agreeing with power plant data to within ± 5%. How- brine is pressurized to 15 bar and re-injected into the ground at
ever, uncertainties exist for several reasons. The most substantial of ∼104 °C. In INJ, a fraction of this brine is heated by the solar field until
these is the simplifications made to model the geothermal plant. For becoming saturated liquid, and is then mixed with the production
instance, Coso operates with between 80–90 production wells and fluids. This configuration shows the poorest performance with the

1842
J.D. McTigue et al. Applied Energy 228 (2018) 1837–1852

power input increasing 0.16 MWe per MWth heat addition. The 1st law solar multiples of σ = 2 and 4, with the summer and winter solstices
efficiency η1 decreases as solar heat is added because the solar-con- illustrating the extreme cases. At σ = 2 , the plant cannot deliver the
version efficiency, ηsol is less than the 1st law efficiency when Qsol = 0 target thermal power (8 MWth) in the winter solstice and stores are
(proof in Appendix B). Increasing the re-heat brine temperature im- underutilized: over winter they discharge only 0.1% of the energy than
proves performance slightly, by increasing the specific enthalpy of the if they were fully charged and discharged each day, see Table 5. The
mixed fluid. storage utilization can be used to evaluate whether the storage is ap-
INJ2: Evaporating the injection brine to a vapor fraction of 50% propriately sized, and is defined as the ratio of the actual energy dis-
improves the performance compared to INJ. The solar-conversion effi- charged to the total quantity that could have been dispatched if the
ciency increases to 22.5% from around 17% and η1 increases with heat stores had been fully charged and discharged once per day.
addition. However, this approach may result in increased mineral de- Solar multiples of 4 are large compared to most current installa-
position in the pipes and heat exchanger. Using a shell-and-tube heat tions. However, the stores are still used to only a small extent
exchanger, with the brine inside the tubes will make it easier to clean throughout the winter, with utilizations of 25% for an 8-h store. On the
although phase change normally occurs on the shell-side. other hand, during the summer the stores are fully charged daily with
COND: Water leaves the condenser at ∼38 °C and 0.1 bar and can be utilizations over 99%. If the stores are full and the geothermal plant
evaporated without scaling issues. Saturated vapor is injected directly cannot accept all the available thermal power, then the power is cur-
into the high-pressure turbine stage, and its temperature is set by the tailed, as shown in Fig. 3b, which indicates that the solar field is not
turbine inlet pressure. Alternatively, the turbine exhaust steam could be cost efficient. There is an inherent trade-off in sizing the solar field and
heated, which would increase the recirculated mass for the same heat thermal stores: large solar fields are required to provide the required
addition since the exhaust is hotter (∼46 °C) than the condenser water. power in the winter. However, this leads to an oversupply of energy in
However, the exhaust is a low pressure (0.1 bar) two-phase mixture and the summer, requiring large stores to avoid curtailment. These large
compressing this fluid to the required turbine inlet pressure (6 bar) stores are then underutilized in the winter.
requires a considerable work input and a two-phase pump.
Furthermore, an isentropic compression increases the fluid temperature
5. Levelized cost of electricity (LCOE) of the hybrid plant
to 550 °C thereby rendering the solar field redundant.
PRE: The two-phase production fluids are heated directly, thereby
The levelized cost of electricity (LCOE) is calculated over a 30-year
increasing their vapor fraction. This method sees the highest increase in
lifetime, for various sizes of solar field, storage durations, and HTF
power output per unit heat addition, with ηsol = 25%. However, eva-
temperatures. The levelized cost of electricity is the cost that, if as-
poration can lead to an increased rate of scaling [34] although the
signed to every unit of electrical energy produced over the lifetime of
precipitation may be delayed by the introduction of H2SO4 [35]. A
the plant, equals the total lifecycle costs when discounted back to the
larger heat exchanger may also be required due to the larger flow rates
current year [36]. Here, the total lifecycle costs include capital costs
and the lower heat transfer coefficients of two-phase flows. For in-
and operational and maintenance costs, but other factors can also be
stance, Qsol = 10 MWth heats 110.6 kg s−1 of production fluids com-
included. For the hybrid plant where the power block and geothermal
pared to 24.6 kg s−1 for INJ and 62.0 kg s−1 for FT1.
wells already exist, the annual electrical energy is given by the marginal
FT1: Brine leaves the first flash tank (FT1) at ∼165 °C compared the
increase in electrical energy above the base rate provided by the geo-
FT2 outlet which is 104 °C. Heating FT1 brine therefore recirculates
thermal plant.
more mass per unit heat addition. This approach has a comparable
The LCOE is calculated using the fixed charge rate (FCR) method:
performance to PRE, with ηsol = 24.3%. Increasing the temperature that
the fluid is heated to reduces net power by a small quantity – the in- Ccap FCR + M
LCOE =
crease in enthalpy of the mixed fluid is outweighed by the increased E (14)
pumping power to keep the fluid saturated. This configuration has the
best performance with the least risk of scaling and is investigated fur- where Ccap is the capital cost, M is the annual operational and main-
ther below. tenance costs, E is the annual electricity generation, and FCR is the
fixed charge rate. FCR is defined as the revenue per unit of investment
4. Annual simulations that must be collected annually to pay for the carrying charges of the
investment. FCR calculation details may be found in Appendix C and
Variations in available solar resource affect the hybrid plant per- Ref. [36]. The cost of each component is evaluated using correlations
formance. The thermal power delivered by the solar collector is a developed from the literature and industry representatives and are
function of the sun angle, the direct normal irradiance (DNI), the mirror presented in Appendix C.
properties and the temperature of the HTF.
The design point of the hybrid plant has a net power generation of Table 4
Operating point of hybrid plant using FT1 configuration.
24.2 MWe which is a 2 MWe increase in the electrical work output of the
geothermal plant. This corresponds to a thermal power of 8 MWth for HP stage LP stage
the FT1 configuration. Generating this thermal power requires a solar
Temperature °C 162.6 100.0
field with a mirror area of 14,159 m2 at the peak DNI on the summer Pressure bar 6.10 1.01
solstice (973 W m−2). A solar field of this size has a solar multiple σ = 1. Mass flow rate kg/s 47.2 54.2
Increasing the solar multiple proportionally increases the solar field size Recirculated temperature °C 162.6
and its power output. Recirculated pressure bar 6.60
Recirculated mass flow rate kg/s 147.9
The operating point of the geothermal system with 8 MWth power
Cold store temperature °C 190.0
addition is shown in Table 4. If the solar collectors provide more than 8 Hot store temperature* °C 300.0/400.0
MWth, then the excess energy is stored. Once the thermal stores are full, Solar thermal energy MWth 8.0
the excess energy is dispatched to the geothermal plant, up to a max- Gross power MWe 24.5
imum thermal input of 16 MWth, which corresponds to the point where Net power MWe 24.2
1st-law efficiency % 17.3
the inlet pressure to the turbine can no longer be increased and the Solar thermal conversion efficiency % 24.3
turbine must be throttled. Beyond this heat input, thermal energy from
the solar field is curtailed. * The hot store temperature depends on the maximum operating tempera-
The hourly behavior of the hybrid system is illustrated in Fig. 3 for ture of the HTF. Two cases are investigated in this article.

1843
J.D. McTigue et al. Applied Energy 228 (2018) 1837–1852

a b
Collector thermal power
30 30
Delivered thermal power
Power delivered to/from stores
Power, MW th

Power, MW th
20 Dumped
20

10 10

0 0

-10 -10

20-Jun 12 AM 21-Jun 12 AM 22-Jun 12 AM 23-Jun 12 AM 20-Jun 12 AM 21-Jun 12 AM 22-Jun 12 AM 23-Jun 12 AM

Summer solstice Summer solstice
c d
15 15

10 10
Power, MW th

Power, MW th
5 5

0 0

-5 -5

20-Dec 12 AM 21-Dec 12 AM 22-Dec 12 AM 23-Dec 12 AM 20-Dec 12 AM 21-Dec 12 AM 22-Dec 12 AM 23-Dec 12 AM

Winter solstice Winter solstice
Fig. 3. Power flows in the hybrid power plant for the summer solstice and winter solstice. (a) Summer solstice, solar multiple = 2. (b), Summer solstice, solar
multiple = 4. (c) Winter solstice, solar multiple = 2. (d), Winter solstice, solar multiple = 4.

Table 5 increasing the storage duration, which explains why the lines begin to
Performance of thermal stores in the hybrid plant. converge at large solar multiples.
Solar multiple = 2 Solar multiple = 4
The choice of heat transfer fluid for the storage media significantly
affects the LCOE. Fig. 4a illustrates the LCOE of a mineral oil (Xcel-
Solar field mirror area m2
28,318 56,636 therm 600) which has low capital cost (2.5 $/kg), is non-toxic, and has
Annual collector GWhth 28.83 57.66 low vapor pressures, so that the stores do not need to be pressurized.
energy
Annual increase in GWhe 6.97 11.37
The maximum operating temperature of the mineral oil is 300 °C.
electrical energy On the other hand, a synthetic fluid with a maximum operating
Average electricity MWhe 23.0 23.5 temperature of 400 °C is shown in Fig. 4b. This fluid is a eutectic
generateda mixture of diphenyl and biphenyl oxides (DPO) (marketed as Dow-
Average first-law % 17.7 18.5
therm A, Therminol VP1 etc.). DPO has more expensive upfront costs
efficiency
Capacity factorb % 42.0 67.6 (3.1 $/kg) compared to mineral oils. Furthermore, operating at higher
temperatures increases the vapor pressure (∼10 bar at 400 °C), re-
Storage Storage
4h 8h 4h 8h
quiring pressurized stores that are more expensive by a factor of ∼1.3.
DPO is more toxic and less environmentally friendly than mineral oils.
Total energy GWhth 6.1 7.0 9.7 17.4 Operators may be cautious of storing large volumes of DPO after a fire
discharged at the SEGS1 CSP plant in 1999 destroyed 3400 m3 of DPO and caused
Utilizationc % 52.4 30.0 83.2 74.6
considerable damage.
Average energy MWhth/day 30.0 37.7 31.8 63.4
discharged Molten salts are another widely-used heat transfer fluid in solar
(May–Jul) applications. However, they are not appropriate for this application
Utilization (May–Jul) % 93.4 59.0 99.3 99.0 because the cold tank is close to the freezing point of most molten salts.
Average energy MWhth/day 0.1 0.1 14.9 15.7
Furthermore, the requirement for electric heaters in the thermal stores
discharged
(Nov–Jan)
to keep the salts molten would increase the cost of the system.
Utilization (Nov–Jan) % 0.2 0.1 46.5 24.6 The cold tank in both cases has a temperature of 190 °C. Increasing
Total energy curtailed GWhth 0.0 0.0 10.7 6.7 the temperature difference between the stores increases the energy
a
density, thereby reducing the storage volume and cost. Fig. 4 indicates
The average electricity generated per day without solar is 22.2 MWe. that this outweighs the impact of using a more expensive HTF and a
b
The capacity factor is defined as the average additional electricity divided
pressurized vessel. For instance, the LCOE of a system with DPO, σ = 2 ,
by the design value of additional electricity (2 MWe).
c and 4 h of storage is 0.085 ± 0.011 $/kWh compared to an equivalent
The storage utilization is defined as the total energy discharged divided by
the total energy that would have been dispatched if the store had been fully plant with mineral oil which costs 0.093 ± 0.011 $/kWh. Using DPO
charged and discharged once per day. instead of mineral oil reduces the storage capital cost from 60 to 39
$/kWh.
Fig. 4 indicates that larger solar fields initially reduce the LCOE as Note that the storage capital cost per unit energy can be calculated
more energy is delivered to the hybrid plant over the course of the year. in two ways. The capital cost per unit energy capacity of the store is
However, increasing the solar multiple further leads to higher LCOEs typically quoted. However, a more representative value is given by
because of increased curtailment. Curtailment can be reduced by considering the utilization of the store—how much energy is discharged

1844
J.D. McTigue et al. Applied Energy 228 (2018) 1837–1852

Fig. 4. Effect of solar field size and

a b thermal storage sizing on the levelized
0.15 0.15
Storage duration = 0.0 h cost of energy. (a) The hot thermal
Storage duration = 2.5 h store is at 300 °C with Xceltherm 600
Storage duration = 5.0 h mineral oil for storage medium. (b)
Storage duration = 7.5 h The hot thermal store is at 400 °C with
Storage duration = 10.0 h DPO as the storage medium. The cold
LCOE, $ / kWh

LCOE, $ / kWh
store is at 190 °C in both cases. The
shaded bars indicate the mean LCOE
plus/minus one standard deviation.
0.10 0.10

0.05 0.05
1 2 3 4 1 2 3 4
Solar multiple Solar multiple

Cost per discharged energy, $ / kWhdischarged

over a typical cycle. Table 6 provides the capital cost per unit energy
dispatched on average, which is significantly larger than the capital cost 160
per unit energy capacity. For instance, the DPO store capital cost per σ = 2.0
unit energy dispatched is nearly double (72.5 $/kWh) the capital cost σ = 2.5
per unit capacity (41.3 $/kWh) because of a storage utilization of
140
σ = 3.0
57.0%. σ = 4.0
Larger thermal stores lead to higher LCOEs, as the total electricity 120
generation remains roughly the same while the investment increases.
The LCOE implies that the optimal storage size is zero as it does not
reflect the value that storage provides to the plant or the grid by en- 100
abling the flexible dispatch of energy [37]. The ‘optimal’ storage size
can be found by instead considering the capital cost per unit energy 80
discharged on average, see Fig. 5. Large stores achieve economies of
scale and have low capital costs per unit capacity. However, they are
infrequently fully charged, so the capital cost per unit energy dispatch 60
is high. Conversely, small stores are frequently utilized, but have higher
capital costs per unit capacity. The ‘optimal’ storage duration is 3 h and 40
10 h for solar multiples of 2 and 3, respectively, when the storage fluid 0 2 4 6 8 10 12
is DPO. Storage duration, h
This illustrates the inherent trade-off in sizing the thermal stores.
Fig. 5. Cost per unit energy dispatched of two-tank thermal storage using DPO.
Efficient use of the storage capacity requires small stores, but cost
The hot store is at 400 °C and the cold store is at 190 °C, and the system dis-
considerations indicate that larger stores are more cost-effective (from a
charges at 8 MWth.
storage perspective). Table 6 compares the economic cost of two solar
field sizes with no storage and with the optimal duration of storage,
according to the cost per kilowatt-hour dispatched for the DPO stores. 6. Comparison with PV + storage and stand-alone CSP plants

The economic feasibility of the hybrid plant is compared to a solar

photovoltaic (PV) field with battery energy storage (BES) and a stand-

Table 6
Hybrid plant economics for different solar multiples, with and without optimally sized stores.
Solar multiple = 2 Solar multiple = 3

0 h (mineral) 3 h (synthetic) 0 h (mineral) 10 h (synthetic)

Total energy GWhth 28.8 28.8 43.2 43.2

Additional electricity GWhe 6.98 6.97 9.34 10.42
Solar field cost M$ 4.25 4.25 6.37 6.37
Solar field HTF M$ 0.230 0.286 0.345 0.430
Pump cost M$ 0.211 0.211 0.211 0.211
HX cost M$ 0.060 0.046 0.060 0.046
HTF cost M$ 0.0 0.595 0.0 1.982
Storage cost M$ 0.0 0.397 0.0 0.777
Capital cost M$ 4.750 5.785 6.986 9.816
Storage cost $/kWh 0.0 41.3 0.0 34.5
Storage utilization % 0.0 57.0 0.0 57.9
LCOE $/kWh 0.069 ± 0.011 0.081 ± 0.011 0.076 ± 0.012 0.091 ± 0.011

1845
J.D. McTigue et al.
alone CSP plant. PV systems have been widely deployed in recent years
– particularly in California due to lower capital costs. For hybrid sys-
tems to be considered a worthwhile investment, they must achieve
LCOEs competitive with conventional renewable technologies. The four
hybrid plant systems described in Table 6 are compared to PV + BES
and CSP systems that produce the equivalent annual energy.
Photovoltaic and battery cost assumptions are summarized in
Table 7. Photovoltaic cell capital costs are estimated from NREL’s An-
nual Technology Baseline [38]. Batteries are Lithium-ion batteries
which have a lifetime of approximately 10 years [39]. Since the LCOE
analysis occurs over a 30-year period it is assumed that the batteries are
replaced at the 10th and 20th year of operation, and the total battery
cost is discounted appropriately. Li-ion batteries are assumed to be
operated to a maximum depth of discharge of 80% [40] – which ef-
fectively increases the required battery size capacity by 25%.
The PV field energy output was modelling using the ‘System Advisor
Model’ (SAM) [27]. The PV panels are assumed to be 1-axis tracking,
with total system losses of 14.1%, an inverter efficiency of 96% and a
DC-to-AC ratio of 1.2. (These are the default values in SAM). The PV
field LCOE is evaluated with the FCR method and has the same eco-
nomic assumptions as the hybrid LCOE calculations, so that the results
are directly comparable.
The nominal PV power capacity is 2 MWe which equals the nominal
design point of the hybrid plant which generates an additional 2 MWe.
The PV field is sized so that it produces the same annual quantity of
energy as the equivalent hybrid plant. As a result, the PV field that
corresponds to a hybrid plant with a solar multiple of 2 has a rated
power of 3.225 MWe while the PV field that corresponds to a hybrid
plant with a solar multiple of 3 has a rated power of 4.82 MWe. It is
assumed that all the power produced by the PV field can be absorbed by
the grid. Thus, no power is curtailed.
When storage is not included PV arrays have similar costs to the
hybrid-geothermal system, as shown in Table 8. However, batteries
have higher capital costs (200–800 $/kWh) and shorter lifetimes
(10 years) than thermal storage [40,41]. Consequently, the LCOE of the
hybrid plant with storage is lower than the PV + BES system. For in-
stance, the hybrid plant with σ = 2 and 3 h of storage has an LCOE of
0.081 ± 0.011 $/kWh, compared to the equivalent PV field value of
0.148 ± 0.066 $/kWh. Similarly, the hybrid plant with σ = 3 and 10 h
of storage has an LCOE of 0.091 ± 0.01 $/kWh, compared to the
equivalent PV field value of 0.254 ± 0.130 $/kWh. The wide variance
in battery costs leads to a large spread of possible LCOEs, but it is worth
noting that even the lowest cost batteries (200 $/kWhe) lead to systems
with higher LCOEs than the hybrid system (0.0821 and 0.125 $/kWhe,
for the 3 h and 10 h storage, respectively). For comparison, other stu-
dies have calculated the LCOE of PV-BES to be 0.082 $/kWh [42] for a
200 MWe solar field with 10 h of storage, and 0.16 ± 0.05 $/kWh and
0.19 ± 0.05 $/kWh [41] for a 100 MWe solar field with 3 and 9 h of
storage, respectively.
These results suggest there is a compelling economic argument to
consider hybrid plants as a competitor to PV systems with batteries.
A major factor in the high LCOE of the PV + BES system is the high
cost of batteries, the requirement to replace them periodically, and the
limited maximum depth of discharge. Furthermore, the above
PV + BES analysis is somewhat optimistic for several additional rea-
sons. The effects of battery degradation over time have been neglected,
but this affects the lifetime and can reduce energy capacities over time
[43]. The round-trip efficiency is assumed to be 100%, while current
estimates place the value in the range of 85–95% for Li-ion batteries. In
this simplified analysis it is assumed that all electricity generated by the
PV field can be dispatched to the grid. Storage is employed to provide
flexibility but would reduce the annual electricity dispatched. The
LCOEs therefore represent a lower bound which would increase as the
benefits of storage were taken advantage of and more accurate battery
models were employed. Of course, the LCOE does not account for the
price electricity is sold at and is not necessarily the most suitable metric
1846
Applied Energy 228 (2018) 1837–1852
for systems including storage.
Similarly to battery degradation, the HTF in the hybrid plant
thermal stores degrades over time and has to be replaced. The above
calculations assumed a replacement rate of 1% per year. However, if
the plant is operated ineffectively then degradation rates could be
higher. For a make-up rate of 10% per year (equivalent to replacing the
fluid once every ten years) the LCOE is 0.089 ± 0.011 $/kWh for DPO
with 3 h of storage and σ = 2 , and 0.109 ± 0.012 $/kWh for DPO with
10 h of storage and σ = 3, which is still more cost-effective than
PV + BES.
The LCOE of a stand-alone CSP plant is also evaluated for compar-
ison. The same solar collectors and thermal storage are used as in the
hybrid plant case. The system uses a synthetic oil that is operated be-
tween the same temperatures as the hybrid plant. The power block
converts solar heat to electricity with a 35% efficiency and costs ap-
proximately 1500 $/kWe (this includes the turbine, generator and
condenser) [38]. The solar multiple of the CSP plant is fixed at 2.5, and
the power rating of the turbine is adjusted to produce an annual
quantity of energy equivalent to the hybrid plant. The stand-alone CSP
plant is affected more severely by curtailment than the hybrid plant.
This is because the geothermal power plant can absorb a greater
quantity of excess solar energy than an independent CSP field. As a
result, the hybrid plant has a lower LCOE than the stand-alone CSP
plant when storage is not included. When storage is included, the stand-
alone CSP’s LCOE decreases since less energy is curtailed and more
electricity is produced over the course of the year. However, the results
in Table 8 demonstrate that the hybrid plant is still a more cost-effective
option, due to the large capital cost of the power block and condenser in
the stand-alone CSP plant.
It is worth noting that the concentrating solar field has been as-
sumed to have relatively conservative cost and performance para-
meters. For instance, the cost of the solar field is 150 $/m2 which is
higher than costs proposed by some manufacturers. Furthermore, linear
Fresnel reflectors typically have poorer optical performance than
parabolic trough technologies, and thereby produce less energy over
the course of a year.
These results suggest that a hybrid geothermal-solar-storage system
provides a relatively cost-effective way to add renewable, dispatchable
power to the electricity grid compared to existing renewable alter-
natives.
7. Flexible power generation
Storage increases a plant’s flexibility as power can be dispatched
during hours of greatest demand. Dispatchable power generation is
increasingly important as renewable penetration increases. For in-
stance, large quantities of solar power on the Californian grid has led to
a surplus of power during the afternoon [44]. This results in the so-
called “duck curve” [45] whereby firm capacity must be rapidly dis-
patched during the evening hours. The abundance of intermittent
generation has led to volatile electricity prices, which can become ne-
gative [46]. Consequently, power generation units should be flexible;
reducing power delivery during the afternoon and ramping at high rates
during the evening. Combining baseload production (geothermal) with
variable production (CSP) and thermal storage provides the flexibility
to operate in such a marketplace. A hybrid system which includes
Table 7
Cost assumptions for a photovoltaic solar array with Lithium-ion battery sto-
rage.
Low Medium High Reference
PV capital cost $/kWe 1000 1300 1700 [38]
Operations cost $/kWe 14 14 14 [27]
Storage capital cost $/kWhe 200 500 800 [40,41]
J.D. McTigue et al. Applied Energy 228 (2018) 1837–1852

Table 8
LCOE comparison of hybrid geothermal-solar-storage with photovoltaic cells
with battery storage, and a stand-alone CSP plant with thermal storage.
Annual Storage LCOE, $/kWhe
energy duration,
generation, h Hybrid plant PV + BES
GWhe
6.98 0 0.067 ± 0.011 0.062 ± 0.014
3 0.081 ± 0.011 0.148 ± 0.066
9.34 0 0.076 ± 0.012 0.062 ± 0.014
10 0.091 ± 0.011 0.254 ± 0.130
dispatchable storage can provide other benefits to the grid, such as
frequency control.
Since dispatchable power generation is a valuable commodity it
should be priced accordingly. This section investigates the price that
stored energy should be dispatched at for the hybrid system to be
economically viable.
The hybrid plant electricity is sold at a flat rate of 0.09 $/kWe.
Electricity produced by discharging the stores is sold at a multiple of the
flat rate, called a “price multiplier”. For a system with a given storage
duration and solar multiple, the price multiplier required for the system
to achieve an internal rate of return of 10% is calculated. (This value of
IRR represents a scenario where a system looks profitable. The IRR is
calculated using the methodology in [36] and described in Appendix C).
Results are presented in Fig. 6. The curves indicate that there is an
optimal storage size: as discussed above, small quantities of storage are
expensive per kWh and therefore require high price multipliers. In-
creasing the storage size increases the economic return, although ex-
tremely large stores are underutilized and again require high price
multipliers. For solar multiples of 2, the optimal storage size is around
3–4 h. Increasing the solar multiple increases the optimal storage
duration to around 10 h, as thermal energy availability has increased.
Using a synthetic fluid typically requires lower price multipliers to
achieve a profitable IRR than mineral oils, which is consistent with the
above analysis. Profitable IRRs can be achieved with price multipliers in
the region of 1.5–1.75.
This analysis assumes that all dispatched energy from storage re-
ceives the increased price. As such it is not a reflection of the current
electricity market where prices vary throughout the day and across the
year. In such a scenario, large storage systems may be of little use, since
higher prices may only be available for one or two hours of the day (and
may coincide with times of high solar availability). Analyzing this
scenario requires time of delivery pricing (including estimates for
future years), as well as a control system for the optimal dispatch of
storage. Such analyses may result in optimal storage sizes and utiliza-
tions that differ from the above results. The present analysis assumes
that storage provides value to the grid in various ways that may not be
directly quantifiable. Therefore, the price multipliers reflect the average
Stand-alone CSP price that stored energy must be dispatched at for the system to be
viable.
0.102 ± 0.013
0.096 ± 0.011
8. Alternative storage systems
0.102 ± 0.013
0.093 ± 0.011
Two-tank liquid stores are the most readily deployable storage
system for CSP plants, but several other pre-commercial storage tech-
nologies exist that may be more compact and cost effective. Options
include solid sensible storage, phase change materials (PCMs) or ther-
mochemical storage [47]. Solid sensible storage is currently the most
cost-effective option with the lowest technological complexity [48],
and includes concrete blocks to which heat is transferred through pipes
[49], or packed beds: a storage vessel filled with solid particles through
which the HTF percolates [50].
For a packed bed, the energy stored is given by:
E = β (1−ε ) ρs V [(cs T )h−(cs T )c ] (15)
where ρs is the solid density, cs is the solid heat capacity, and ε is the
void fraction (the fraction of space not filled by particles), T is the
temperature, V is the volume, and β is a simple factor that accounts for
the fact that the thermal front in the store is not a step function but has
a finite gradient. This means that the full storage is not utilized and
should be oversized in order to store the required energy. Subscripts h
and c refer to the hot and cold temperatures, respectively.
The packing material is an iron oxide such as magnetite (Fe3O4)
with an average cs = 800 kJ/kg K, ρs = 5000 kg/m3, ε = 0.4 and
β = 0.75. The cost of the packing is estimated to be 500 $/m3 [51], and
the pressure vessel cost is calculated with the correlations in Appendix
C, and uses DPO as the HTF.
Two-tank liquid storage with 4 h of discharge (32 MWh) operating
between 400 °C and 190 °C requires each tank to have a volume of
330 m3, and a total capital cost of 39 $/kWh. However, a packed bed
with the same energy capacity and operating temperatures would be
315 m3 and 14 $/kWh. Moreover, the HTF volume is reduced from
300 m3 to 126 m3, thereby reducing the severity of fire risk. However,
the discharging temperature varies with time and the round-trip effi-
ciency may be affected by irregular charge and discharge cycles [52].
Numerous schemes have been proposed to mitigate these problems,
such as the inclusion of PCMs [53], the segmentation of the bed into
layers [54,55], or mixing the discharging flow with a cooler fluid to

a b Fig. 6. Curves showing the price multiplier

required for a hybrid system with a given
4.0 4.0 storage duration to achieve an internal rate
Mineral oil, SM = 2 Synthetic oil, SM = 2 of return of 10%. The price multiplier is
Mineral oil, SM = 3 Synthetic oil, SM = 3 the increased price that stored energy is
dispatched at. Shaded bands show the un-
certainty. (a) Mineral oil storage, with the
Price multiplier

Price multiplier

3.0 3.0 hot store at 300 °C. (b) Synthetic oil sto-
rage, with the hot store at 400 °C.

2.0 2.0

1.0 1.0
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Storage duration, h Storage duration, h

1847
J.D. McTigue et al.
control the mixture’s temperature. Evaluating the influence of this
transient behavior on the integrated plant performance will determine
the true cost savings.
9. Conclusions
In this article, an existing double-flash geothermal power plant is
retrofitted with concentrating solar power to increase the power gen-
eration. Solar heat is added in five different configurations, and the
most suitable approach involves reheating the brine after the first flash
tank. The solar heat added to this brine is converted to electricity with
an efficiency of 24.3%. This approach has a reduced risk of depositing
minerals on the heat exchangers or pipework compared to other
methods. Retrofitting in this way is cost effective since the existing
power block, pipework, and condenser can be used. Various other ap-
proaches exist, such as using solar heat to power a high-temperature
topping cycle. Heat could be rejected into the geothermal plant to im-
prove generation. A topping cycle will increase the capital cost but will
achieve higher efficiencies. Furthermore, storage systems at higher
temperatures are likely to be more compact and cost effective.
Incorporating two-tank liquid thermal storage into the hybrid plant
increases the flexibility of the plant, at increased capital cost. Increasing
the temperature difference between the tanks reduces the volume, and
thus the cost. This is an effective approach even if more expensive
storage fluids and pressurized vessels are required. The levelized cost of
electricity (LCOE) was calculated for several storage durations and solar
field sizes and was found to be significantly lower than equivalently
sized photovoltaic arrays with battery storage because thermal stores
are cheaper and have longer lifetimes than batteries. For instance, a
system with 3 h of storage had an LCOE 28% lower, while a system with
10 h of storage was 47% lower. The hybrid plant also had a lower LCOE
than a conventional stand-alone concentrating solar power (CSP) plant.
Hybrid geothermal-solar-storage systems therefore provide a relatively
cost-effective way to add renewable, dispatchable power to the elec-
tricity grid compared to existing renewable alternatives. Concrete
stores and packed-beds are likely to provide cheaper, more compact
storage than liquid tanks, albeit at increased complexity and risk.
Storage provides numerous benefits to the grid and can dispatch
Appendix A. Thermal properties of heat transfer fluids
See Tables 9 and 10.
Appendix B. Proof showing 1st law efficiency decreases as solar heat is added
The first law efficiency of a system with no solar heat input is
Wo
η1, o =
Q
The first law efficiency of the hybrid plant including a solar heat input is
W
η1 =
Q + Qsol
The solar conversion efficiency is
Table 9
Thermal properties of mineral oil Xceltherm 600.
Temperature Heat Vapor
capacity pressure
°C kJ/kg K kPa
25 2.0 0.0
100 2.3 0.0
200 2.6 1.1
300 2.9 19.0
400 – –
1848
Applied Energy 228 (2018) 1837–1852
power at times of high electricity prices. Calculations indicate that the
price needs to be roughly double the typical flat-rate price of 0.09
$/kWe for storage to provide internal rate of returns over 10%. These
price points are small compared to the price fluctuations that are cur-
rently observed in the Californian energy market. However, exploiting
price differentials may not be an option for some current geothermal
plant operators if they are committed to an existing power purchase
agreement which specifies a fixed electricity price.
The results of this study suggest that hybridizing concentrating solar
power with geothermal energy and thermal storage can be a cost-ef-
fective strategy to add dispatchable renewable energy to the grid.
Numerous existing geothermal plants could potentially benefit from the
addition of solar heat to increase power output. This article suggests an
effective way to integrate the two systems without requiring the in-
stallation of a topping cycle. Future studies should consider more
practical aspects – such as long-term trends in geothermal well per-
formance, quantify the risk and mitigation of scaling in components,
and the effect of electricity price fluctuations on operating strategies
and plant design.
Acknowledgements
This work was authored in part by Alliance for Sustainable Energy,
LLC, the manager and operator of the National Renewable Energy
Laboratory for the U.S. Department of Energy (DOE) under Contract No.
DE-AC36-08GO28308. Funding provided by U.S. Department of Energy
Office of Energy Efficiency and Renewable Energy, Geothermal
Technologies Office and Solar Energy Technologies Office. The views
expressed in the article do not necessarily represent the views of the
DOE or the U.S. Government. The U.S. Government retains and the
publisher, by accepting the article for publication, acknowledges that
the U.S. Government retains a nonexclusive, paid-up, irrevocable,
worldwide license to publish or reproduce the published form of this
work, or allow others to do so, for U.S. Government purposes.
The authors gratefully acknowledge Dr. Mike Erbes for assisting
with the off-design steam turbine modeling and the Naval Geothermal
Program Office, including Dr. Andrew Sabin, Mr. David Meade, Mr.
Michael Lazaro, and Ms. Kelly Blake, for their contributions.
Viscosity Density Thermal
conductivity
2 3
mm /s kg/m W/m K
31.0 848.0 0.14
3.0 804.0 0.13
0.8 743.2 0.12
0.4 682.5 0.11
– – –
J.D. McTigue et al. Applied Energy 228 (2018) 1837–1852

Table 10
Thermal properties of synthetic fluid, Therminol VP1 (DPO).
Temperature Heat Vapor Viscosity Density Thermal
capacity pressure conductivity
2 3
°C kJ/kg K kPa mm /s kg/m W/m K

25 1.6 0.0 3.6 1060.6 0.14

100 1.8 0.5 1.0 998.3 0.13
200 2.0 10.1 0.4 912.6 0.11
300 2.3 241.5 0.3 816.4 0.10
400 2.6 1086.0 0.2 694.2 0.08

W −Wo
ηsol =
Qsol
The first law may therefore be written as
W + ηsol Qsol
η1 =
Q + Qsol
Differentiating with respect to Qsol leads to
∂η1 η Q−Wo
= sol
∂Qsol (Q + Qsol )2
In order for the first law efficiency to increase with solar heat addition, the first derivative must be greater than zero, this leads to the condition
Wo
ηsol > = η1, o
Q

Appendix C. Cost correlations

Cost correlations for storage vessels: The pressure vessel contains the storage media. Pressure vessel cost estimates were obtained from PCL
Construction by Coso Operating Company and fitted well with a correlation derived from Peters and Timmerhaus [56] (when costs were inflated to
current values) for unpressurised vessels. The cost of the vessel depends on its volume V and is given by:
Cves = 7351·V 0.557
Pressurizing the store significantly increases the cost, and the following correlation was obtained from EconExpert [57] (an online tool for capital
cost estimation) for pressures over 7 bar:
Cpres = 0.922 + 0.0335P−0.0003P 2 + 1 × 10−6P 3

The total cost of the pressure vessel is therefore given by CPV = Cpres·Cves .
Cost correlations for heat exchangers: Correlations from EconExpert for a floating-head, shell-and-tube heat exchanger constructed from
carbon steel.
CHX = β (18944.44 + 280.71A−0.06601A2 )

β = 1 for P < 10 bar

β = 1.0011 + 0.001P for P ⩾ 10 bar

where P is the pressure, and A is the heat transfer area. A is calculated using the log-mean temperature difference method, where
Q
A=
U ·LMTD
where Q is the heat transferred, U is the overall heat transfer coefficient (assumed to have a constant value of 1000 W/m2 K) and LMTD is the log-
mean temperature difference, given by
ΔTh−ΔTc
LMTD =
log(ΔTh)−log(ΔTc )

where ΔTh is the temperature difference on the hot side of the heat exchanger, and ΔTc is the temperature difference on the cold side of the heat
exchanger.
Solar field costing: The solar field cost is assumed to be in the range of 100–200 $/m2 following discussions with solar collector manufacturers.
HTF costing: Following contact with manufacturers and suppliers, Xceltherm 600 was priced at 2.5 $/kg and Therminol VP1 was priced at 3.1
$/kg for bulk quantities (over 19 m3 or 5000 gallons). The HTF degrades over time and is replaced at a rate of 1% per year over the lifetime of the
plant.
Pump costing: This correlation was derived from EconExpert for a centrifugal pump made from cast steel. P is the discharge pressure, and Ẇ is
the rated power.

Cpump = 5648.8P 0.4305Ẇ 0.4121

Price of electricity: The price of electricity was set at 0.09 $/kWh following discussions with industrial partners.

1849
J.D. McTigue et al. Applied Energy 228 (2018) 1837–1852

Sensitivity analysis: The cost of each component was assumed to follow a normal distribution C ∼ N (μ, σ 2) , which was truncated to prevent
negative costs. A Monte Carlo simulation with 10,000 runs was undertaken. For each run, each component normal distribution was sampled, and the
economic metrics evaluated to develop a distribution of costs. The mean value of each distribution was that given by the equations above. The value
of the standard deviations are given in Table 11.
Economic assumptions: Economic parameters for calculating the LCOE and IRR are summarized in Table 12 and are the default values used in
the System Advisor Model (SAM). The LCOE is calculated using the fixed charge rate (FCR), as in Eq. (14). This relatively simple method is used to
reduce the number of uncertain variables in the economic calculations. SAM is also widely employed, so using the FCR method facilitates comparison
with other studies. More detailed LCOE calculations are possible, such as those discussed in [58]. These advanced models can include details such as
shadow pricing of capital costs, re-investments, and scrap values.

C.1. Calculation of the fixed charge rate (FCR)

The values in Table 12 are used to calculate the fixed charge rate, which is used to evaluate the LCOE from Eq. 14. The FCR is given by:
FCR = (PFF)(CRF)
where PFF is the project financing factor, and CRF is the capital recovery factor. PFF is given by
1−T (PVDEP)
PFF =
1−T
where T is the tax rate and PVDEP is the present value of depreciation, which is given by:
N
Dj
PVDEP = ∑ (1+WACC) j (1 + i) j
j=0

where Dj is the depreciation rate in year j, i is the inflation rate, and WACC is the weighted average cost of capital, given by:
1 + (1−d )((1 + RROE)(1 + i)−1) + d (1−T )((1 + RINT)(1 + i)−1)
WACC = −1
1+i
where d is the project debt fraction, and RROE is the real return on investment, given by:
1+IRR
RROE = −1
1+i
where IRR is the internal rate of return. RINT is the real debt interest rate:
1+NINT
RINT = −1
1+i
where NINT is the nominal debt interest rate. The CRF is given by:
WACC
CRF = N
1− ( 1
1 + WACC )
where N is the lifetime of the plant in years, which is assumed to be 30 years.

C.2. Internal rate of return

The internal rate of return (IRR) of an investment that has a series of future cash flows Fi is the rate of return that sets the net present value (NPV)
of the cash flows equal to zero.
N
Fi
∑ (1 + IRR)i
= NPV = 0
i=0

where the cash flow F in year i is given by

Fi = Ri−Ii−Mi−(Ri−Di−Mi ) T
where R is the revenue, I is the investment, M is the O&M costs, D is the depreciation, and T is the tax rate. For the hybrid plant, the revenue is the
marginal revenue above that which would have been produced by the geothermal power plant.

Table 11
Parameters used in sensitivity analysis.
Mean value, $ Standard Upper limit Lower limit
deviation

Solar field 150 0.20 µ 10 µ 0.30 µ

HTF – 0.10 µ 5.0 µ 0.30 µ
Pressure vessel – 0.25 µ 20 µ 0.30 µ
Heat exchanger – 0.20µ 10 µ 0.30 µ
Pump – 0.20 µ 10 µ 0.30 µ
Electricity price 0.09 0.20 µ 1.50 µ 0.20 µ

Dash indicates costs are calculated from correlations above.

1850
J.D. McTigue et al.
Table 12
Economic factors used in LCOE calculations.
Economic parameters
Tax rate, T, %
Inflation rate, i, %
Project debt fraction, d, %
Internal rate of return, IRR, %
Nominal debt interest rate, NINT, %
Depreciation rate, Dj, %
Note that the equation for IRR is a Laguerre polynomial with N roots. For the case of an initial cash investment followed by cash inflows, there is
only one positive root, and the remainder are negative or imaginary.
References
[1] Wendt DS, Mines GL. Use of a geothermal-solar retrofit hybrid power plant to mi-
tigate declines in geothermal resource productivity. Trans – Geotherm Resour
Counc 2014;38.
[2] Snyder DM, Beckers KF, Young KR, Johnston B. Analysis of geothermal reservoir
and well operational conditions using monthly production reports from Nevada and
California. Trans – Geotherm Resour Counc 2017;41.
[3] “National Solar Radiation Database.” [Online]. Available: < https://nsrdb.nrel.
gov/ > . [Accessed: 12-Nov-2017].
[4] Wendt DS, Mines GL. Stillwater hybrid geo-solar power plant optimization analyses.
Trans – Geotherm Resour Counc 2015;39:891–900.
[5] DiMarzio G, Angelini L, Price W, Chin C, Harris S. The Stillwater triple hybrid power
plant: integrating geothermal, solar photovoltaic and solar thermal power genera-
tion. In: Proceedings world geothermal congress; 2015, April, pp. 1–5.
[6] Kuyumcu OC, Özalevli OSCC, Baker DK, Somek SK. Design and implementation of
the Gümüşköy hybrid geothermal and solar thermal power system. Trans –
Geotherm Resour Counc 2014;38:811–6.
[7] Zhou C, Doroodchi E, Moghtaderi B. An in-depth assessment of hybrid solar –
geothermal power generation. Energy Convers Manag 2013;74:88–101.
[8] Ayub M, Mitsos A, Ghasemi H. Thermo-economic analysis of a hybrid solar-binary
geothermal power plant. Energy 2015;87:326–35.
[9] Heberle F, Hofer M, Ürlings N, Schroder H, Anderlohr T, Brüggemann D. Techno-
economic analysis of a solar thermal retrofit for an air-cooled geothermal Organic
Rankine Cycle power plant. Renew Energy 2017;113:494–502.
[10] Lentz A, Almanza R. Solar – geothermal hybrid system. Appl Therm Eng
2006;26:1537–44.
[11] Handal S, Alvarenga Y, Recinos M. Geothermal steam production by solar energy.
Trans – Geotherm Resour Counc 2007;31.
[12] Alvarenga Y, Handal S, Recinos M. Solar steam booster in the Ahuachapan geo-
thermal field. Trans – Geotherm Resour Counc 2008;32:2–9.
[13] Miguel Cardemil J, Cortés F, Díaz A, Escobar R. Thermodynamic evaluation of
solar-geothermal hybrid power plants in northern Chile. Energy Convers Manag
2016;123:348–61.
[14] Mathur PN. An assessment of solar-geothermal hybrid system concepts. The
Aerospace Corporation, report No: ATR-79(7773-01)-2; 1979.
[15] Ghasemi H, Sheu E, Tizzanini A, Paci M, Mitsos A. Hybrid solar – geothermal power
generation: optimal retrofitting. Appl Energy 2014;131:158–70.
[16] Astolfi M, Xodo L, Romano MC, Macchi E. Geothermics technical and economical
analysis of a solar – geothermal hybrid plant based on an Organic Rankine Cycle.
Geothermics 2011;40(1):58–68.
[17] Bonyadi N, Johnson E, Baker D. Technoeconomic and exergy analysis of a solar
geothermal hybrid electric power plant using a novel combined cycle. Energy
Convers Manag 2018;156:542–54.
[18] SimTech, “Process Simulation Environment (IPSEpro)”, < http://www.
simtechnology.com/CMS/index.php/ipsepro > ; 2017.
[19] DiPippo R. Geothermal power plants. 4th ed. Butterworth-Heinemann; 2016.
[20] Dixon LS, Hall CA. Fluid mechanics and thermodynamics of turbomachinery.
Elsevier Science; 2010.
[21] Cooke DH. On prediction of off-design multistage turbine pressures by Stodola’s
ellipse. J Eng Gas Turbines Power 1985;107:596–606.
[22] Denton JD. Loss mechanisms in Turbomachines. J Turbomach
1993;115(October):2017.
[23] Spencer RC, Cotton KC, Cannon CN. A method for predicting the performance of
steam turbine-generators 16,500 kW and larger; 1974.
[24] Fuller LC, Stovall TK. A computer code for the performance of regenerative su-
perheated steam-turbine cycles. In: ORNL-5547, NASA CR-159540; 1979.
[25] Choo YK, Staiger PJ. New Features and Applications of PRESTO, a Computer Code
for the Performance of Regenerative, Superheated Steam Turbine Cycles. In: NASA
TP 1954 c.1; 1982.
[26] Lazalde-Crabtree H. Design approach of steam-water separators and steam dryers
for geothermal applications. Geotherm Resour Council Bull 1984.
[27] System Advisor Model Version 2.17.9.5 r2 (SAM 2017.9.5 r2). National Renewable
Energy Laboratory, Golden, CO. < https://sam.nrel.gov/content/downloads > .
[Accessed July 1, 2018].
1851
Applied Energy 228 (2018) 1837–1852
Assumed values
40
2.5
60
10
8
20, 32, 20, 14, 14; 0 thereafter
[28] Zhu G. Development of an analytical optical method for linear fresnel collectors. Sol
Energy 2013;94:240–52.
[29] Zhu G, Turchi C. Solar field optical characterization at stillwater geothermal/solar
hybrid plant. J Sol Energy Eng 2017;139(June):1–10.
[30] Burkholder F, Kutscher CF. Heat loss testing of Schott’s 2008 PTR70 parabolic
trough receiver. In: NREL Tech. Rep. TP-550-45633, no. May, 2009, p. 58.
[31] Bell IH, Wronski J, Quoilin S, Lemort V. Pure and pseudo-pure fluid thermophysical
property evaluation and the open-source thermophysical property library coolprop.
Ind Eng Chem Res 2014;53:2498–508.
[32] McTigue J, Castro J, Mungas G, Kramer N, King J, Turchi C. Retrofitting a geo-
thermal plant with solar and storage to increase power generation. Trans –
Geotherm Resour Counc 2017;41.
[33] González-Roubaud E, Pérez-Osorio D, Prieto C. Review of commercial thermal
energy storage in concentrated solar power plants: steam vs. molten salts. Renew
Sustain Energy Rev 2017;80(March, 2016):133–48.
[34] Nakao Y, Mugikura Y, Ogata K, Katsuki N. Development of hybrid geothermal
power plants in Japan. Trans – Geotherm Resour Council 2017;41.
[35] Kawahara Y, et al. Laboratory experiments on prevention and dissolution of silica
deposits in a porous column (1): solid deposition due to silica particle aggregation
and inhibition by acid dosing. Trans – Geotherm Resour Counc 2012;36:1–4.
[36] Short W, Packey DJ. A manual for the economic evaluation of energy efficiency and
renewable energy technologies. NREL Tech. Rep. NREL/TP-462-5173, no. March;
1995.
[37] Denholm P, Eichman J, Margolis R. Evaluating the technical and economic per-
formance of PV plus storage power plants. Technical report NREL/TP-6A20-68737;
2017.
[38] NREL, “2017 Annual Technology Baseline”; 2017. < http://www.nrel.gov/
analysis/data_tech_baseline.html > . [Accessed July 1, 2018].
[39] Lazard. “Lazard’s Levelized Cost of Storage Analysis – Version 3.0”; 2017.
[40] Lai CS, Jia Y, Lei L, Xu Z, Mcculloch MD, Po K, et al. Comprehensive review on
large-scale photovoltaic system with applications of electrical energy storage.
Renew Sustain Energy Rev 2017;78(November, 2016):439–51.
[41] Feldman D, Margolis R, Stekli J, Feldman D, Margolis R, Stekli J. Exploring the
Potential Competitiveness of Utility-Scale Photovoltaics plus Batteries with
Concentrating Solar Power, 2015 – 2030, NREL Technical Report TP-6A20-66592;
2016.
[42] Lazard. Levelized Cost of Energy Analysis – version 11.0”; 2017.
[43] Lai CS, Mcculloch MD. Levelized cost of electricity for solar photovoltaic and
electrical energy storage. Appl Energy 2017;190:191–203.
[44] Fu R, Feldman D, Margolis R, Woodhouse M, Ardani K. U.S. Solar Photovoltaic
System Cost Benchmark : Q1 2017, NREL Technical Report TP-6A20-68925; 2017.
[45] Denholm P, Connell MO, Brinkman G, Jorgenson J. Overgeneration from Solar
Energy in California : A Field Guide to the Duck Chart, Technical report NREL/TP-
6A20-65023; 2015.
[46] California ISO; “California ISO Q3 2017 Report on Market Issues and Performance”;
2017.
[47] Pardo P, Deydier A, Anxionnaz-Minvielle Z, Rouge S, Cabassud M, Cognet P. A
review on high temperature thermochemical heat energy storage. Renew Sustain
Energy Rev 2014;32:591–610.
[48] Zhang H, Baeyens J, Caceres G, Degreve J, Lv Y. Thermal energy storage: recent
developments and practical aspects. Prog Energy Combust Sci 2016;53:1–40.
[49] Laing D, Steinmann WD, Viebahn P, Gräter F, Bahl C. Economic analysis and life
cycle assessment of concrete thermal energy storage for parabolic trough power
plants. J Sol Energy Eng 2010;132(4):41013.
[50] Zanganeh G, Pedretti A, Zavattoni S, Barbato M, Steinfeld A. Packed-bed thermal
storage for concentrated solar power – pilot-scale demonstration and industrial-
scale design. Sol Energy 2012;86(10):3084–98.
[51] McTigue J. Analysis and optimisation of thermal energy storage. [PhD thesis].
University of Cambridge; 2016. https://doi.org/10.17863/CAM.7084.
[52] Bruch A, Fourmigué JF, Couturier R. Experimental and numerical investigation of a
pilot-scale thermal oil packed bed thermal storage system for CSP power plant. Sol
Energy 2014;105:116–25.
[53] Zanganeh G, Commerford M, Haselbacher A, Pedretti A, Steinfeld A. Stabilization of
the outflow temperature of a packed-bed thermal energy storage by combining
rocks with phase change materials. Appl Therm Eng Sep. 2014;70(1):316–20.
[54] Crandall DM, Thacher EF. Segmented thermal storage. Sol Energy
J.D. McTigue et al. Applied Energy 228 (2018) 1837–1852

2004;77(4):435–40.
[55] White AJ, McTigue JD, Markides CN. Analysis and optimisation of packed-bed
thermal reservoirs for electricity storage applications. Proc Inst Mech Eng Part A J
Power Energy 2016;230(7):739–54.
[56] Peters MS, Timmerhaus KD. Plant design and economics for chemical engineers.
McGraw-Hill, Inc.; 1990.
[57] Vasudevan PT, Ulrich T. EconExpert; 2000. [Online]. Available: < http://www.
ulrichvasudesign.com/cgi-bin/cgiwrap.cgi/econ/econnew.pl > . [Accessed: 24-
Jan-2018].
[58] Zhao Z, Chen Y, Douglas J. Levelized cost of energy modeling for concentrated solar
power projects: a China study. Energy 2017;120:117–27.

1852