Anda di halaman 1dari 13

Applied Energy 96 (2012) 222–234

Contents lists available at SciVerse ScienceDirect

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

Optimization of specific rating for wind turbine arrays coupled to compressed


air energy storage
Samir Succar a,b,⇑, David C. Denkenberger c, Robert H. Williams b
a
Natural Resources Defense Council (NRDC), 1152 15th Street, NW, Suite 300, Washington, DC 20005, USA
b
Princeton Environmental Institute, Princeton University, Princeton, NJ 08544-1003, USA
c
Denkenberger Inventing and Consulting, 2345 Forest Ave., Durango, CO 81301, USA

a r t i c l e i n f o a b s t r a c t

Article history: A methodology is presented for jointly optimizing the wind turbine specific rating and the storage con-
Received 15 July 2011 figuration for a large-scale wind farm coupled to compressed air energy storage (CAES). By allowing the
Received in revised form 16 November 2011 wind-storage system to be optimized in an integrated, variable rating framework the levelized cost of
Accepted 5 December 2011
electricity (LCOE) can be reduced substantially. These changes also enhance the capacity factor of the
Available online 9 January 2012
wind farm, reduce the storage capacity requirements of the baseload plant and reduce the greenhouse
gas emission rate of the combined system relative to a separately optimized wind farm coupled to CAES.
Keywords:
The results of this analysis could have important implications for the competitiveness of large-scale
Compressed air energy storage
Optimization
remote wind and the applicability of energy storage as a baseload wind strategy in a carbon constrained
Power systems economics world.
Stochastic systems Ó 2011 Elsevier Ltd. All rights reserved.
Wind energy

1. Introduction coupled with CAES due to the relatively low cost and potential
for widespread availability of this type of energy storage in areas
As the share of electricity produced by wind generation contin- with high quality wind resources [10]. The role of electrical energy
ues to grow, the impacts of wind’s variability will be increasingly storage explored in this analysis is of directly coupled operation
felt on the grid [1,2]. Electrical energy storage can be an important with a single wind farm to produce baseload power. This analysis
source of flexibility to help facilitate the integration of wind into does not attempt to reflect the many other market functions that
the electric grid and can serve a wide range of grid support appli- energy storage can provide on a system [4,5,11–14], but rather to
cations as well [3–5]. Although the integration of renewable en- analyze the economic value of co-optimization of storage and wind
ergy can be facilitated by means of numerous sources of rating ratio within the context of a specific operational mode.
flexibility [6,7], coupling of wind with energy storage can be an The integration of renewables and energy storage has been
attractive strategy, especially where penetration levels of wind widely discussed and the existing literature encompasses a wide
are high and transmission capacity is constrained [8]. Furthermore, range of technologies and applications [15–21]. Several prior stud-
colocation of wind and storage enables wind to serve baseload ies have analyzed the economics, emissions and system character-
markets and maximizes utilization of transmission infrastructure istics of both standalone CAES plants [14] and wind/CAES hybrid
for wind [9]. systems [9,22–25] including optimal scheduling [26], dual mode
This analysis will focus on the economics of large-scale, electri- CAES for remote systems [27], and other novel designs [28–30].
cal energy storage for producing baseload power from wind. A There is an extensive literature of examining the optimal storage
wide array of electrical energy storage technologies exist today, sizing of the storage systems coupled to renewable energy gener-
but bulk storage technologies scalable to the 100’s of megawatts ators [15,31–37], as well as several studies looking at the optimiza-
and gigawatt–hours of energy storage capacity needed to provide tion of the power curve and specific rating of a wind turbine in the
firm power at utility scale are currently limited to pumped hydro- context of standalone wind turbine arrays [38–40]. However, a co-
electric storage (PHS) and compressed air energy storage (CAES) optimization of these systems has not been previously analyzed.
[5]. The focus of this analysis will be on wind turbine arrays Although prior wind/storage studies have optimized the configura-
tion of the CAES system [11], in these analyses the power output
characteristics of the renewable energy generator was not co-opti-
⇑ Corresponding author at: Natural Resources Defense Council (NRDC), 1152 15th
mized with the energy storage system. Rather than rely on a static
Street, NW, Suite 300, Washington, DC 20005, USA. Tel.: +1 202 513 6255; fax: +1
202 289 1060. wind turbine configuration and fixed wind turbine power curve,
E-mail address: ssuccar@nrdc.org (S. Succar). this paper co-optimizes the turbine design jointly with the CAES

0306-2619/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apenergy.2011.12.028
S. Succar et al. / Applied Energy 96 (2012) 222–234 223

Nomenclature

Parameters vout wind turbine cut-out speed


CA array loss coefficient vr wind turbine rated windspeed
CWP wind turbine array capital cost ($) a array loss exponent
ei scaling exponent of the ith capital cost block cC CAES compressor minimum loading fraction
fi capital cost fraction of the ith capital cost block cG CAES turboexpander minimum loading fraction
F levelized fuel costs ($/MWh) d2 diameter squared wind turbine array spacing
hy hours per year gC CAES storage input efficiency, ratio of stored energy to
ls CAES inventory (MWh) input energy
Mf annual fixed operations and maintenance ($/yr) gG CAES storage output efficiency, ratio of output energy to
Mv levelized variable operations and maintenance ($/MWh) stored energy
n system component index gCER CAES charging energy ratio
N number of time steps in the time series gA wind turbine array efficiency
PC CAES compressor input power sFP firm power fraction
PG CAES turboexpander output power v levelized capital charge rate
PL target power delivered to load
PT total power delivered of the combined system Variables
PW aggregate wind turbine array output hs duration of storage, hours of output at PL
PWC curtailed wind power rr wind turbine rating ratio
Pwr wind turbine rated capacity rf wind turbine array relative capacity
vavg annual mean wind speed rC CAES compressor chain relative input capacity

system and introduces wind turbine specific rating as an additional wind turbine rating for a given application is determined by the
variable for joint co-optimization of a wind/CAES system delivering wind turbine rating ratio (rr): the ratio of the rated wind speed
baseload power. In this case, the system is optimized in an inte- to the mean speed of the wind regime (vavg). Since the choice of
grated framework such that the resulting configuration yields a turbine rating is determined by a quantity normalized to the mean
net reduction in wind turbine rating (i.e. ‘‘derating’’). wind speed at hub height, the absolute value of vr will depend on
the wind resource strength, but the rating ratio is independent of
wind regime.
2. Methodology
The effects of rating on the wind turbine power curve are shown
schematically in Fig. 2a. A decrease in the rating ratio from 1.5 to
The model consists of a wind turbine array and a CAES system
1.2 results in a wind turbine power curve that plateaus at smaller
designed to deliver baseload power with an installed capacity of
fractions of the mean wind speed, thus reducing the maximum
2000 MW [9,10]. A notated schematic of the model is provided in
power output of the turbine. This truncation of the wind turbine
Fig. 1. The system is designed to deliver wind energy directly onto
output at lower wind speeds means that the rated capacity of
the line unless the wind power output exceeds the transfer capac-
the wind turbine will be reduced. This change is largely a function
ity of the transmission line, which is equal to the target power
of a reduced size generator in the wind turbine nacelle as well as a
delivered to the load, PL. When the output of the wind turbine ar-
scaling of numerous other components (Table 1) [39,41]. This
ray rises above this transfer limit, the excess power is used to
means that a wind turbine with lower rating ratio (rr) has a smaller
power the CAES compressor train which injects high-pressure air
power block (generator, transformer, gearbox, etc.) relative to the
into a storage volume, typically an underground reservoir such as
other cost blocks, which vary as a smaller scaling exponent of
a solution mined salt dome or saline aquifer, until the CAES storage
the rated speed (ei). In particular, the fixed block components,
volume is full. If the output of the wind turbine array falls below
including the tower and blades dimensions as well as the low wind
the transmission line capacity and the storage volume is not de-
speed power output, are invariant to changes in rr.
pleted, the air is withdrawn from the CAES storage volume, natural
The cut-out speed is the speed above which the turbine is shut
gas is combusted and the combustion products are expanded
down to prevent excessive loading on the tower and blades. Above
through a CAES turboexpander train to generate electricity. The
the cut-out speed, the turbine is stopped and the blades are
combined output of the wind and CAES systems is optimized to de-
pitched to minimize loading of the structure. Thus, above this wind
liver baseload power as defined below and the size of the wind
speed the output of the wind turbine is zero. The cut-out speed is
farm and CAES components are optimized to produce energy at
assumed to be 3.5 times vavg in this case [41], although the cut-off
the lowest levelized cost. In addition, the wind turbine rating ratio
for real wind turbines in the field are based on cost tradeoffs for the
is used to adjust the power curve of the wind turbines and opti-
turbine components over their fatigue lifetimes. Since it is as-
mize the power production of the array for baseload wind/CAES
sumed that the outer dimensions of the blades and tower are
operation.
approximately independent of changes in the wind turbine rating,
the wind loading and therefore the cut-out speed are independent
2.1. Wind turbine rating ratio
of rr as well.

The power curve of a wind turbine describes the power output


as a function of wind speed. The rated speed of that turbine (vr), the 2.2. Wind turbine capital cost
wind speed at which the slope of the wind turbine power curve
goes to zero, is a critical aspect of the turbine design. Below the The change in rating has important implications for the capital
rated speed, the turbine output is nominally proportional to the cost of the turbine. Because the components of the turbine can be
cube of the wind speed and for wind speeds above rating, the out- scaled down with lower rating, the total cost per wind turbine will
put is roughly independent of the incident wind speed. The optimal decline as the rating is reduced (Fig. 2b). The functional
224 S. Succar et al. / Applied Energy 96 (2012) 222–234

Fig. 1. Model schematic.

Fig. 2. The effect of wind turbine rating ratio (rr = vr/vavg) on (a) wind turbine power curve (b) total wind turbine capital cost ($) assuming a nameplate rating of 3.45 MW at a
nominal rating ratio of 1.50 (c) wind turbine capital cost per unit rated capacity ($/kW) and (d) LCOE ($/MWh) for a wind turbine array (without storage). The LCOE minimum
occurs at rr = 1.5 which is the current rating ratio commonly used in current wind farm design.

dependence of capital components is aggregated into four blocks nominal rating (vr,0). The values for nominal capital cost (CWP,0/
each of which scales as a different power of the rating: a power Pwr,0 = $2140/kW) and nominal rating (vr,0 = 1.5  vavg,0) reflect cur-
block, a thrust block, a torque block and a fixed block. The capital rent costs for wind turbines and current rating guidelines for wind
cost of each turbine can be expressed in terms of a nominal capital turbine arrays [41,42], where Pwr,0 denotes the wind turbine rated
cost figure (CWP,0) which reflects the costs for a turbine with a capacity at nominal rating (vr = vr,0). The capital cost per kilowatt of
S. Succar et al. / Applied Energy 96 (2012) 222–234 225

Table 1 standalone wind turbine arrays (i.e. no storage) the minimum cost
Wind turbine capital cost block fractions and scaling exponents [39,41,43]. occurs at rr  1.5 for base case assumptions, as shown in Fig. 2d.
Block Components Scales Scaling Nominal Several studies have been carried out on energy storage systems
quantity exponent fraction [44,45] and wind/storage systems [21,34,46–48] including several
Power: Brake, generator, Turbine epwr = 3.0 fpwr = 0.206 focused specifically on wind with CAES [9,22,23,25,49]. Prior stud-
Cpow variable speed rated power ies have also investigated the reduction of wind turbine rating of
electronics, electrical standalone wind farms [43,50]. These studies show that a lower
connections, hydraulic
system, transformer
rating ratio might be optimal under certain market conditions. Re-
Torque: Gearbox Rated torque etor = 1.4 ftor = 0.102 cent trends in rotor diameter scaling relative to wind turbine in-
Ctor on the stalled capacity suggest that lower rating ratios with
drivetrain corresponding increases in capacity factor have been pursued in
Thrust: Blades, hub, main Load ethr = 0.7 fthr = 0.388
some commercial systems [51]. However these studies did not
Cthr frame, tower, fluctuations
foundations investigate turbine rating in the context of wind farms colocated
Fixed: Pitch mechanism, Independent efix = 0.0 ffix = 0.297 with energy storage.
Cfix bearings, low speed of rating The optimal turbine rating in a wind/CAES system will differ
shaft, nacelle cover, significantly from this standalone value due to effects of rating
control, safety system,
on wind capacity and storage size. The magnitude of this shift is
transportation, fixed
electrical interface, dependent on the cost and wind resource assumptions applied.
permits, engineering, Derating results in important changes in many characteristics of
yaw drive the system including the storage system configuration, greenhouse
gas emissions rate and LCOE.
Unless otherwise noted, cost assumptions, model framework,
wind capacity is calculated as a nominal cost times the sum over optimization constraints and synthetic time series generation fol-
‘‘i’’ capital cost blocks: low the framework described in depth by Greenblatt et al. [9]. In
addition to the rating ratio, the primary variables that govern the
X
C WP ¼ C WP;0 ðPwr =Pwr;0 Þ fi ðv r =v r;0 Þei : ð1Þ model are the storage capacity (hs) and relative capacities of the
i system components (rC, rG, rf). The latter are the ratios of the capac-
ities of the compressor train, the turboexpander, and the wind
The cost fractions fi and exponents ei are derived from prior
farm, respectively, to the nameplate capacity of the combined
studies that analyze the change in rating and capital cost of turbine
baseload system (PL). The storage capacity is the number of hours
components for numerous designs [39,41,43]. The cost block frac-
required to deplete the storage reservoir at a discharge rate of PL.
tions sum to unity:
X
fi ¼ 1; ð2Þ 3. Optimization of wind/storage with variable rating
i
As shown in Fig. 2d, the optimal rating ratio for a stand-alone
so that: wind farm with unity array efficiency is 1.5. However, the addition
of storage costs (CAES compressor, expander, balance of plant, stor-
C WP jv r ¼v r;0 ¼ C WP;0 : ð3Þ
age volume, operation and maintenance and fuel) changes the
Because Pwr is proportional to the cube of the mean wind speed, dynamics of the optimization of turbine rating substantially.
and because ei 6 3.0 for all i, the rated power of the wind turbine Since the capacity factor of a standalone wind turbine array is a
will decline faster than capital cost declines for reductions in wind function of rr, the rating will affect the amount energy produced by
turbine rating. As a result, although the cost per turbine has a po- the system. This means that the contribution of the wind turbine
sitive slope with respect to rating ratio (Fig. 2b), the cost per in- array to the 85% baseload system capacity factor will vary with rat-
stalled capacity (CWP/Pwr, $/kW) increases as the rating ratio is ing. As the turbines are derated and the wind capacity factor in-
reduced (Fig. 2c). creases, wind will account for a greater fraction of the total
output from the combined system.
The optimization model is formulated to minimize the LCOE of
2.3. Capacity factor and cost of energy the combined system summed over n system components.
" #
Because a reduction in wind turbine rating implies a lower va- X C n  v þ M n;f Pn
min þ ðMn;v þ F n Þ ð4Þ
lue of vr while the cut-out speed remains fixed, the power curve of n hy  P T PT
the ‘‘derated’’ turbine (i.e. a turbine with reduced rating, rr < rr,0)
will have a longer plateau between vr and vout (Fig. 2a). As a result, In this case, the system consists of two principle components:
the turbine will spend a larger fraction of time producing power at the wind farm and the CAES plant. This cost minimization is carried
the rated capacity of its generator, and therefore the mean power out subject to the following constraints:
output of the turbine will decrease less quickly than the rated PT;t ¼ PW;t  PC;t  PWC;t þ PG;t 6 PL ð5Þ
capacity as the turbine is derated. Hence, the capacity factor (the
ratio of mean power output to rated capacity) will increase with 1 X N
derating (Fig. 2c). P T;t P sFP ð6Þ
N  PL t¼1
In some circumstances, the increase in capacity factor offsets
the increased capital cost at lower rating values leading to an opti-
pðP T P PL Þ P sFP ð7Þ
mal rating which is lower than in the nominal case. The slope in
capital cost ($/kW) becomes increasingly negative for decreasing
values of rr such that at some rating value, it is no longer optimal
cG rG PL 6 PG;t 6 rG PL ð8Þ
to continue derating. The result is that the levelized cost of electric-
cC rC PL 6 PC;t 6 rC PL ð9Þ
ity (LCOE) exhibits a positive curvature with a local minimum. For
226 S. Succar et al. / Applied Energy 96 (2012) 222–234

Fig. 3. Power duration curves for wind/CAES systems delivering 85% capacity factor with (a) nominal rating (rr = 1.5) and (b) reduced rating (rr = 1.28).

PC;t 6 PW;t  P L ð10Þ As the wind farm capacity increases above the load level, more
wind energy becomes available both to charge the storage volume
PG;t 6 PL  PW;t ð11Þ CAES and for direct utilization (represented in Fig. 3 as the ‘‘Stored
Wind’’ and ‘‘Wind’’ components of the power duration respec-
ls;t ¼ ls;t1 þ PC;t  gC  PG;t =gG ð12Þ tively). If the wind farm output exceeds PL, the excess wind is di-
verted to CAES and used to power the compressor chain if there
gCER ¼ gG  gC ð13Þ is storage volume capacity remaining and will be curtailed other-
wise. The storage system can generate electricity at later times of
hs r G P L shortfall and the ratio between the input and output electrical en-
0 6 ls;t 6 ð14Þ ergy will be determined by the charging energy ratio (gCER) of the
gG
storage system.1 The output of CAES will thus be equal to the frac-
The constraint in Eq. (5) defines the total power delivered from tion of wind output that exceeds PL times the gCER (i.e. the area of
the Wind/CAES system as the net wind power output plus the con- the ‘‘CAES’’ component of the power duration curves in Fig. 3 will
tribution from CAES subject to a fixed upper bound. In this case, the be equal to that of the ‘‘Stored Wind’’ component times the gCER).
net wind output refers to the wind output delivered directly and The way in which the rating affects the capacity factor and tem-
excludes the curtailed wind generation and the wind generation poral variability of the wind turbine output is evident from the
used to run the CAES compressor train. power duration curve (Fig. 3). In particular, derating results in a
The constraints in (6) and (7) describe how the firm power frac- greater range of wind speeds for which the turbine produces rated
tion is applied. The firm power fraction is used to ensure that the power (Fig. 2a). This in turn implies that the turbines will spend a
capacity value of the Wind/CAES system is equivalent to that of a greater fraction of time producing power at their rated capacity.
conventional baseload unit. In this reduced form cost optimization This increased fraction corresponds to the width of the plateau in
model, the calculation of loss of load expectation that would be the left portion of the wind output of the power duration curve.
necessary to calculate the effective load capability of the unit is The change in the turbine power curve, and especially the broaden-
not feasible [52,53]. The alternative methodology adopted here is ing of the plateau between rated and cut-out speeds leads to a cor-
to apply the firm power fraction as a capacity factor constraint in responding widening of the wind portions of the power duration
(6) and as a threshold point for the power duration curve of the curve (Fig. 3). This wider power duration for wind reflects the in-
system meeting the target nameplate capacity of the system where crease in wind capacity factor depicted in Fig. 2c.
the x-axis is normalized to include values from zero to unity [9]. These increases in direct wind output lead to a reduction in the
The latter constraint is expressed in Eq. (7) in terms of a special storage output requirements for balancing wind. In addition, since
case of the cumulative probability function p(x) where the inequal- the ‘‘Stored Wind’’ component of the power duration curve in Fig. 3
ity indicates a summation with a lower bound of PL. This expres- widens with lower rating, less wind capacity above PL is needed to
sion is identical to the inverse of the system power duration achieve the same energy input to CAES. Thus, the wind farm capac-
curve and because the power duration is a monotonically decreas- ity necessary to meet the needed fraction of energy delivered by
ing function, the formulation in Eq. (7) is equivalent to a minimum CAES is reduced and the amount energy required by CAES is itself
power duration value at an x-intercept of sFP. reduced as well.
Eqs. (8)–(11) define the output constraints and minimum load- These effects provide an incentive to derate the wind turbines
ing limits for the compressor and expander trains described previ- in a wind/CAES system relative to the nominal rating for a stand-
ously by Greenblatt et al. [9]. Finally, Eqs. (12)–(14) describe the alone wind farm. Using the base case assumptions described in Ta-
storage inventory management and constraints as well as the ble 2, the LCOE minimum shifts to rr = 1.35 in the case of no array
charging energy ratio for the CAES system [10]. losses (Fig. 4a). While such a rating ratio would clearly not be opti-
mal without storage, the addition of CAES and the need to produce
3.1. Wind farm capacity and storage output baseload power provide the necessary incentives to scale the

The amount of wind energy that can be utilized directly will be 1


The charging energy ratio (gCER) for CAES is the amount of electrical output for a
determined by the rating (rr) and relative wind farm capacity (rf). given electrical input. In this case gCER = 1.5.
S. Succar et al. / Applied Energy 96 (2012) 222–234 227

Table 2
Base case parameters of the model.

Parameter Symbol Base value Units Refs.


Load power PL 2.0 GW
Firm power fraction sFP 85 %
Levelized capital charge ratea v 11.5 %/yr
Natural gas pricea pNGeff 5.5 $/GJ HHV [54]
GHG intensity of natural gasa,b 66.0 kg CO2eq/GJ [58]
(55.6) (GJ/tCe)
Mean wind power densityc Pavg 650 W/m2 [41]
Mean wind speedc vavg 8.22 m/s [41]
Wind autocorrelation time H 30 h [9]
Wind weibull shape factor k 2 [9]
Wind turbine rotor diameter D 100 m [41]
Wind turbine hub height h 120 m [41]
Wind turbine base rated power Prate,0 3.0 MW [41]
Wind turbine cut-in speed vin 3.0 m/s [41]
Wind turbine cut-out speed vout 3.5 vavg m/s [41]
Wind turbine rated speed vrate,0 12.33 m/s [41]
Wind turbine rated speed ratio rrate 1.5 [41]
Wind turbine power coefficient Cp 39 % [41]
Array efficiency below rating gA 86 % [43]
Diameter squared spacing d2 50 [43]
Wind farm base capital cost CWP,0/Pwr,0 2140 $/kW [42]
Wind farm fixed O&M MW,f 28.07 $/kW-yr [54]
Wind farm variable O&M cost CW,v 0 $/MWh [54]
CAES minimum power ratio 5 % [59]
CAES charging electricity ratio gCER 1.50 [5,60]
CAES balance of plant cost ratiod 63 % [5,60]
CAES heat ratea 4220 kJ/kWh [5,60]
CAES compressor costd 280 $/kW [5,60]
CAES expander costd 310 $/kW [5,60]
CAES storage coste 2.09 $/kWh [5,61]
CAES fixed O&M MC,f 4.79 $/kW-yr [5,60]
CAES variable O&M MC,v 3.59 $/MWh [5,60]
a
Thermal content stated in LHV basis, except energy prices are on a higher heating value (HHV) basis—the norm for US energy pricing.
b
Natural gas carbon dioxide content of 55.6 kg CO2eq/GJ LHV [62] and upstream GHG emissions (10.4 kg CO2eq/GJ LHV) [58].
c
Extrapolated from 10 m reference Class 4 wind speed (5.77 m/s) using 1/7 scaling exponent to assumed hub height of 120 m.
d
Costs determined from information provided by EPRI and DOE [5,60] and commercial vendors.
e
Storage cost for solution mined domal salt or porous rock formations.

Fig. 4. LCOE as a function of rating ratio. The optimal rating ratio for wind/CAES is significantly lower than for wind power alone. Array efficiency (gA) will further shift the
location of the minimum as well: (a) gA = 1.0 (b) gA = 0.85.

turbine components in a very different way. Although the LCOE 3.2. Wake effects and array efficiency
components of the wind farm alone still reach a cost minimum
at a rating ratio of 1.5, the addition of storage costs and the The array efficiency is a parametric expression for the impact of
reduction of storage size with derating, shifts the cost minimum the reduced wind speed at downstream turbines in an array. The
to lower rating values. array efficiency for a wind turbine array will decline as the spacing
228 S. Succar et al. / Applied Energy 96 (2012) 222–234

1.5 diameter turbines) [40]. By contrast, an array efficiency of 0.7 cor-


responds to a 4 diameter isotropic spacing and a land area require-
1.45
ment of 0.16 km2 per turbine. Although local site conditions
1.4 influence turbine spacing considerably, a typical value of 50 is used
for d2 leading to a base case array efficiency of 86%. Unless other-
1.35
Optimal Rating Ratio

wise noted, this value is assumed.


1.3 The amount by which the air slows by the time it reaches the
rotor (i.e. the interference factor of the turbine) declines above vr
1.25
because the output of the turbine is constant and therefore the
1.2 power coefficient falls. This translates into a reduced downstream
η =1.0 C +20% wake that, together with atmospheric mixing, allows wind speeds
A WP,0
1.15 η =1.0 C
A WP,0 to reach the free stream velocity before reaching the downstream
ηA=1.0 CWP,0 −20%
1.1 turbines thereby eliminating wake-induced losses above vr.
ηA=0.7 CWP,0 +20%
The result is that since reductions in turbine output due to
ηA=0.7 CWP,0
1.05 wake-induced array effects only play a role for wind speeds below
η =0.7 C −20%
A WP,0
1
the rated speed in the turbine, derating reduces the impact of these
400 500 600 700 800 900 1000 1100 losses. In other words, since a wind turbine with lower rating
Wind Power Density (W/m2) spends more time in the rated wind speed regime (vr < v < vout),
and since the array efficiency goes to unity for windspeeds above
Fig. 5. Optimal ratios for wind/CAES at 80%, 100% and 120% of CWP,0 ($1241/kW at
rr = 1.5 and class 4 winds: 650 W/m2 at 120 m hub height). Rating values are vr, the energy loss due to wake effects is diminished as the turbine
computed for array efficiencies of 1.0 and 0.70. rating is reduced. Consequently, reductions in array efficiency (e.g.
due to more closely spaced wind turbines), shifts the LCOE mini-
between adjacent turbines is reduced. In the context of a simplified mum to lower rating values. For the nominal set of assumptions
step-reduction wake model [43] the array efficiency is related to in the model, the reduction of array efficiency from 1.0 to 0.85 re-
the spacing between wind turbines as follows: duces the optimal rating from 1.35 to 1.28 (Fig. 4). The other major
impact of arrays on wind turbines is increased fatigue. This effect
gA ¼ 1  C A da2 ð15Þ also decreases above vr, and so would also reduce the optimal rat-
ing, but this effect is not quantified here.
In this case, the array spacing is governed by the diameter
squared spacing parameter where total land area per turbine is gi-
ven by the product of d2 and the square of the rotor diameter. The 3.3. Wind turbine rating ratio values
array loss coefficient and exponent are derived as described by
Denkenberger et al. (2.1 and 0.7 respectively), with the result that Under the range of parameter values explored, the optimal rat-
an array efficiency of 0.85 corresponds to an isotropic spacing of ing ratio for Wind/CAES typically lies in the range 1.15–1.45
6.6 rotor diameters between adjacent turbines, or equivalently a depending on capital costs assumptions and gA (Fig. 5). The capital
land requirement of 0.43 km2 per turbine (for 100 m rotor cost of wind and CAES are key parameters in the rating ratio

Wind/CAES with Nominal Rating


(rated speed = 1.50 * avg wind speed)
4.20 GW Wind Farm
1,217 3.5MW Turbines
596 sq. km

13.2 TWh/y
9.02 TWh/y Direct 14.9 TWh/y
Wind Output
3. in

61% Wind, 39% CAES


92 d
W

TW Inp
h/ t

0.23 TWh/y Dumped 106 kgCO2e/MWh


y
u

CAES
St=165hr 5.88 TWh/y CAES Output
Comp =
Natural Gas Input: 2.20GW
23.9e6 GJ LHV / y

Wind/CAES with Optimized Rating


(rated speed = 1.29 * avg wind speed)

3.35 GW Wind Farm


1,518 2.2 MW Turbines
744 sq. km 9.80 TWh/y Direct 14.9 TWh/y
13.3 TWh/y Wind Output 66% Wind, 34% CAES
CA 3.40 91.7 kgCO2e/MWh
ES TW
Wi h/y
nd
Inp
0.12 TWh/y Dumped ut CAES 5.10 TWh/y CAES Output
St=137hr
Natural Gas Input: Comp=
1.35 GW
20.7e6 GJ LHV / y

Fig. 6. Effect of reduced turbine rating on wind/CAES system energy balance.


S. Succar et al. / Applied Energy 96 (2012) 222–234 229

Table 3
Wind/CAES System Configuration Summary With and Without Derating.

Wind/CAES with nominally Wind/CAES with optimally rated Wind/CAES, optimal rating, 20% reduction
rated wind turbines (rr = 1.5) wind turbines (rr = 1.29) in wind capital cost (rr = 1.24)
Wind transmitted directly (TW h/y) 9.02 9.80 10.09
Wind input to CAES (TW h/y) 3.92 3.40 3.21
Curtailed wind (TW h/y) 0.23 0.12 0.22
CAES output (TW h/y) 5.88 5.10 4.81
Total annual generation (TW h/y) 14.9 14.9 14.9
CAES Fuel input (106 GJ LHV/y) 23.9 20.7 19.5
Wind farm capacity (GW) 4.20 3.35 3.21
Wind turbine rated power (MW) 3.45 2.21 1.95
Number of turbines 1217 1518 1651
Wind farm capital cost ($/kW) 2140 2903 2540
Wind farm area (km2) 596 744 809
CAES compressor chain (GW) 2.20 1.35 1.21
CAES turboexpander (GW) 1.46 1.32 1.23
Storage energy capacity (GW h) 331 275 228
CAES capital cost ($/kW)2 1670 1410 1346
GHG emissions rate (kg CO2eq/MW h) 106 91.7 86.5
LCOE ($/MW h) 107.2 105.2 90.9

optimization because the optimization of rr is a tradeoff between plant components and subsurface components) by approximately
capital cost increase (at low values of rr) and wind capacity factor 16%. Capital costs for CAES coupled to nominally rated wind is
decrease (at high values of rr). $1670/kW compared to $1410/kW for the re-optimized system
As the wind capital cost increases, the cost minimum shifts to (Table 3).
higher rating (Fig. 5). Changes to CAES capital cost or financing in- The relationship between the rating and the size of the storage
puts affecting the levelized capital charge rate have similar effects cavity, turboexpander capacity and the wind farm capacity all
on the optimal rating as well. The increase in wind power density show approximately linear behavior (Fig. 7). Furthermore, the rela-
has an effect on LCOE equivalent to that of reducing the wind tur- tionships between the storage component sizes and the rating ratio
bine capital cost and therefore increasing resource strength (i.e. appear to be independent of the wind capital cost. However there
moving to higher wind classes) has a similar effect as reductions is a strong relationship between these trends and the array effi-
in wind capital cost. ciency of the wind farm. Because the power output of the wind
farm diminishes with reductions in array efficiency, the size of
4. Results wind and storage components must be increased to meet the re-
quired system capacity factor. As a result, the trend line for
4.1. CAES configuration gA = 1 results in the smallest values for rf, rG and hs. The reduction
of gA to 0.85 and 0.7 increase the optimal values for the wind and
Incorporating turbine rating within the wind/storage optimiza- storage component sizes as well as the slope of the trend line
tion leads to substantial scaling down of CAES components relative (Fig. 7).
to what is required to achieve the same system output with a nom- Another way to view benefit of derating is to consider the wind/
inally rated wind turbine array. As mentioned above, the relative CAES system capacity factor as a function of storage size as shown
wind capacity (rf) of the system is substantially reduced because in Fig. 8. This shows that a wind/CAES system can achieve higher
of derating (Section 4.1). Since the input power of the compressor capacity factor with the same storage size by using lower rated tur-
chain is fixed at the difference between the wind capacity and the bines. Hence, given a fixed storage capacity, a wind farm with re-
load power (i.e. rC = rf  1), the reduction in rf implies substantial duced rr will be able to deliver baseload power with a smaller
reduction in rC as well. In the base case, the reduction of wind tur- installed capacity of wind (or alternatively can produce greater
bine rating from rr = 1.5 to 1.29 reduces the CAES system compres- output with the same wind farm capacity).
sor capacity from an rC value of 1.10 to 0.67.
Because the derated turbines have higher capacity factor, the 4.2. Energy balance and emissions
CAES output requirement needed to meet 85% capacity factor is re-
duced from 5.88 to 5.10 TW h/y (Fig. 6, Table 3) thus allowing the Since the turbines in the derated wind farm operate at rated
CAES storage volume to be reduced from 165 h of storage power a larger fraction of the time, the reduction of wind turbine
(331 GWh) to 137 h (275 GWh). This reduction of storage volume specific rating results in increased wind capacity factor. As a result,
size could be especially important if storage size were geologically systems with reduced rr will be able to deliver more wind-derived
constrained. This reduction in CAES output requirement also al- electricity directly to the load. For the systems compared in Fig. 6,
lows for a reduction in the turboexpander capacity from an rg of the optimally rated wind farm supplies 66% of the electricity
0.73 at nominal rating to 0.66 for the re-optimized system. (9.80 TW h/y at rr = 1.29) compared to 61% (9.02 TW h/y) transmit-
Together, these changes in expander, compressor and storage ted directly for the nominal rating case. Smoother output also
volume size reduce the total capital cost of the system substan- means that less electricity needs to be diverted into storage (the
tially. Under base case conditions, re-optimization of wind turbine power duration curves in Fig. 3). This further underscores the effect
rating reduces the CAES system capital costs2 (including balance of of rating on storage system and wind capacity described above.
These results are summarized in Table 3.
2
The capital cost in this case is expressed in dollars per kilowatt of system capacity This reduction in storage size also means that the system re-
(i.e. PL) rather than in terms of the expander capacity of the storage plant because (1) quires less fuel (Fig. 6). This translates directly into a corresponding
the system is evaluated as a component of an integrated generation system rather
than as an autonomous storage unit and (2) the load power provides a constant basis
reduction in the CO2-equivalent GHG emission rate for the wind/
for comparison of the capital costs compared to the expander capacity which in this CAES system as a whole (from 106 to 92 kg CO2eq/MWh). By com-
case is allowed to vary freely. parison, wind balanced by natural gas generation have a GHG
230 S. Succar et al. / Applied Energy 96 (2012) 222–234

Fig. 7. Rating integrated in system optimization; storage parameters and rating ratio for 85% capacity factor baseload wind/CAES systems.

emission rate of 227 to 267 kg CO2eq/MWh while natural gas com-


100 bined cycle and coal IGCC emit approximately 440 and 901 kg
CO2eq/MWh respectively [54].
90 The resulting emissions rate is even lower than coal IGCC with
carbon capture and storage (approximately 144 kg CO2eq/MWh
80
System Capacity Factor (%)

[55]), indicating that wind/CAES with variable rating can have a


significant impact in mitigating GHG emissions for baseload power
70
applications.
The CAES duration curves in Fig. 9 show that utilization of the
60
storage volume is not a strong function of rating ratio and is only
somewhat dependent on the characteristics of the wind regime.
50
rr = 1.05
Although the nominally rated and derated systems in Fig. 9a and
40 r = 1.20
9b differ in terms of the frequency of compressor dispatch, the rate
r
r = 1.35 of storage volume utilization is nearly identical and the approxi-
r
30 r = 1.50 mate linearity of the inventory duration curve indicates that the
r
r = 1.65 CAES system was dispatched such that the frequency across stor-
r
20 age inventory values was comparable. In contrast, changes in wei-
0 50 100 150 200
bull shape factor (k) in Fig. 9c and Fig. 9d show that a wind regime
Hours of Storage Capacity
with a narrower probability distribution (i.e. higher value of k) will
Fig. 8. System capacity factor as a function of storage cavity size for various rating spend a relatively shorter amount of time near the lower end of the
values. inventory range and a smaller fraction of time with an empty
S. Succar et al. / Applied Energy 96 (2012) 222–234 231

Fig. 9. CAES duration curves above show the CAES storage inventory duration curve, the CAES compressor train duration curve, and the turboexpander train duration curve
for (a) nominal rating (b) derating under base case conditions (c) derating with Weibull shape factor k = 1.5 (d) derating with Weibull shape factor k = 3 (e) derating with wind
autocorrelation time of 10 h and (f) derating with wind autocorrelation time of 90 h. All values are normalized to the maximum value of the relevant quantity.

reservoir. This is likely a consequence of winds with narrower dis- few changes in the CAES duration curves, although the high h run
tribution functions producing steadier output. Lastly, changes in does result in less time with an empty reservoir. In addition, as
the wind autocorrelation time in Fig. 9e and Fig. 9f show relatively noted in earlier studies, higher autocorrelation times also result
232 S. Succar et al. / Applied Energy 96 (2012) 222–234

170
ηa=1 rr=1.5
5 η =1, r opt
a r
160
ηa=0.85 rr=1.5
0 η =0.85, r opt

%Δ COE
a r
150
η =0.7 r =1.5
a r
−5 ηa=0.7, rr opt
140

−10
130 500 750 1000
COE ($/MWh)

W/m2

120

110

100

90

80

70
500 600 700 800 900 1000
Wind Power Density W/m2

Fig. 10. COE for the constant rating case (rr = 1.5, dashed lines) and variable rating (rr optimized, solid lines). Inset shows percentage change in wind/CAES COE resulting from
integration of wind turbine rating in wind/CAES optimization.

Capacity Factor
200 1
C Price
Caes FC
CaesO&M 0.8
180
CaesScc
CaesPcc
WindO&M 0.6
160 WindCap

0.4

140
0.2
CAES
Wind
Cost of Energy ($/MWh)

120
0
0 50 100 150 200
Greenhouse Gas Price ($/tCO )
2
100

Parameter Values
2
80 rf
r
g
1.5 rc
60
h
s
rr
40 1

20 0.5

0 0
0 50 100 150 200 0 100 200
Greenhouse Gas Price ($/tCO2) Greenhouse Gas Price ($/tCO2)

Fig. 11. Wind/CAES COE, Capacity Factor and Configuration as a function of CO2-equivalent GHG emissions price.

in increased storage requirements [9,56]. In this case, the storage 4.3. Cost of energy
capacity requirement increased to 452 GWh for h = 90 h, compared
to 275 GWh for h = 30 h. In most respects however, the CAES sys- Over the range of wind power density considered, re-optimiza-
tem configuration did not significantly change in response to these tion of wind turbine rating results in a $1.0–5.4/MWh decrease in
changes in wind regime. LCOE relative to a system with nominal rating at nominal array
S. Succar et al. / Applied Energy 96 (2012) 222–234 233

efficiency (Fig. 10). This represents a 0.84–5.6% decrease in gener- 5. Conclusions


ation cost for baseload wind without a price on GHG emissions.
The cost saving associated with variable rating increase as wind The joint optimization of a wind/CAES baseload plant results in
capital costs decline or if lower capital charge rates emerge. a system configuration that is substantially different from stand-
The impact of derating on the LCOE depicted in Fig. 10 reveals alone systems. The fixed capital costs associated with storage
that the benefit from turbine derating is greatest in the case of high and the ability to transmit more wind energy directly using der-
wind power density (or alternatively, low wind capital cost) and ated turbines provides incentives to rate wind turbines substan-
low array efficiency (i.e. more prominent wake effects due to closer tially lower than for a stand-alone wind farm. Derating results in
turbine spacing). a decreased levelized cost of electricity, reduced storage require-
Because the wind turbines have a reduced cost per kilowatt at ments, a greater percentage of wind transmitted directly, and a
higher wind classes, the optimal value of rr is lower (above regard- lower greenhouse gas emissions rate. This methodology will be
ing optimal rating relative to the cost of wind capital). Comparing especially relevant in cases where storage capacity is constrained
these results with the trends for rr reveals that the greatest cost or climate policy is enacted. Future reductions in wind capital cost
reduction from derating is realized under conditions that favor will magnify the benefits available from derating thus further
the greatest degree of derating: lower wind capital cost (or higher enhancing wind’s viability in baseload markets in a carbon con-
wind power density) and lower array efficiency (Fig. 5). strained world.
The wind farm capital cost figure used in this analysis assumes
current costs of $2140/kW for nominally rated turbines based on
recent market data [42]. However, current market conditions and Acknowledgements
supply dynamics in the industry have resulted in wind turbine cost
reductions relative to the turbine prices implicit in current project The authors would like to thank Jeffery Greenblatt and Carsten
costs used in this analysis [42,57]. Furthermore, continued down- Hein Westergaard for their helpful input and discussions in devel-
ward cost pressure from increased international competition and oping and refining this work. The findings in this article are a result
other market factors could plausibly result in further cost reduc- of research conducted at the Princeton Environmental Institute at
tions [57]. In light of these developments, a 20% wind capital cost Princeton University with generous financial support from BP
reduction case is presented in Table 3 for comparison. This results and the William & Flora Hewlett Foundation.
in further derating (rr = 1.24) with corresponding changes in sys-
tem configuration, cost and performance.
In prior studies, the system was optimized in a nominal rating References
framework that results in three static configurations that do not
change as a function of GHG emissions price (pGHG) [9]. Because [1] Holttinen H, Meibom P, Orths A, Hulle Fv, Lange B, O’Malley M et al., Impacts of
large amounts of wind power on design and operation of power systems,
varying the wind turbine rating has an impact on the system emis-
results of IEA collaboration. In: Presented at the 8th International Workshop on
sions rate, this assumption does not hold for system optimization LargeScale Integration of Wind Power into Power Systems as well as on
in a variable rating framework. As discussed above, derating in- Transmission Networks of Offshore Wind Farms, Bremen, Germany; 2009.
[2] Smith JC, Milligan MR, DeMeo EA, Parsons B. Utility wind integration and
creases the fraction of system output delivered directly from wind
operating impact state of the art. IEEE Trans Power Syst 2007;22:900–8.
thus reducing contribution from backup (natural gas fired capacity [3] Denholm P, Ela E, Kirby B, Milligan M. The Role of Energy Storage with
or CAES). Therefore derating results in a reduction in the GHG Renewable Electricity Generation, National Renewable Energy Laboratory,
emission rate. This in turn affects the dynamics of baseload wind Golden, CO NREL/TP-6A2-47187; January 2010.
[4] Eyer J, Corey G. Energy storage for the electricity grid: benefits and market
optimization as a function of the CO2-equivalent GHG emissions potential assessment guide: a study for the DOE energy storage systems
price. program. Sandia National Laboratories, Albuquerque, New Mexico SAND2010-
In the case of a Wind/Gas system in which wind is backed by 0815; February 2010.
[5] Rastler D. Electricity energy storage technology options: a white paper primer
natural gas fired capacity rather than storage, the only relevant on applications, costs and benefits. Electric Power Research Institute, Palo Alto,
variable is the relative capacity of the wind farm. Because oversiz- CA 1020676; December 2010.
ing of the farm is not cost-effective until very high values of pGHG, [6] Milligan M, Porter K, DeMeo E, Denholm P, Holttinen H, Kirby B. Wind power
myths debunked. IEEE Power Energy Mag 2009;7:89–99.
the configuration of the system is static over the relevant range of [7] Mathiesen BV, Lund H. Comparative analyses of seven technologies to facilitate
GHG emissions price [9]. The introduction of turbine rating allows the integration of fluctuating renewable energy sources. IET Renew Power
the system to choose the optimal mix of gas and wind without cur- Generation 2009;3:190–204.
[8] Denholm P, Sioshansi R. The value of compressed air energy storage with wind
tailing wind output by oversizing the wind turbine array.
in transmission-constrained electric power systems. Energy Policy
However, the gas backup system does not change in configura- 2009;37:3149–58.
tion or cost as the wind output becomes more consistent with low- [9] Greenblatt JB, Succar S, Denkenberger DC, Williams RH, Socolow RH. Baseload
wind energy: modeling the competition between gas turbines and compressed
er rating. The gas backup is modeled as a conventional natural gas
air energy storage for supplemental generation. Energy Policy
fired plant with a static configuration, in contrast to the CAES sys- 2007;35(Mar):1474–92.
tem that can vary in expander, compressor or storage volume [10] Succar S, Williams RH. Compressed air energy storage: theory operation and
capacity in response to changes in wind output. applications. Princeton University, Princeton, NJ: Princeton Environmental
Institute; 2008. April.
The COE reductions for Wind/CAES brought about by derating [11] Drury E, Denholm P, Sioshansi R. The value of compressed air energy storage in
grow with increased values of pGHG despite relatively small frac- energy and reserve markets. Energy 2011;36:4959–73.
tional change in the configuration variable values (Fig. 11). The [12] Walawalkar R, Apt J, Mancini R. Economics of electric energy storage for
energy arbitrage and regulation in New York. Energy Policy 2007;35:2558–68.
change in rating values over the full range of GHG emissions prices April.
considered is small (1.2 < rr < 1.3) relative to the variation observed [13] Lund H, Salgi G, Elmegaard B, Andersen AN. Optimal operation strategies of
for Wind/Gas (1.1 < rr < 2.0) and the rating ratio remains below the compressed air energy storage (CAES) on electricity spot markets with
fluctuating prices. Appl Therm Eng 2009;29:799–806.
nominal (rr = 1.5) for all values of pGHG (Fig. 11). However, the [14] Lund H, Salgi G. The role of compressed air energy storage (CAES) in future
reduction in GHG emissions rate relative to the nominally rated sustainable energy systems. Energy Convers Manage 2009;50:1172–9.
system means that the economic benefit of derating increases as [15] Zhou W, Lou C, Li Z, Lu L, Yang H. Current status of research on optimum sizing
of stand-alone hybrid solar-wind power generation systems. Appl Energy
pGHG is raised.
2010;87:380–9.
234 S. Succar et al. / Applied Energy 96 (2012) 222–234

[16] Mason J, Fthenakis V, Zweibel K, Hansen T, Nikolakakis T. Coupling PV and [39] Malcolm DJ, Cotrell J. The influence of specific rating on the cost of wind
CAES power plants to transform intermittent PV electricity into a dispatchable energy. In: Presented at the ASME Wind Energy Symposium Technical Papers
electricity source. Progr Photovoltaics: Res Applic 2008;16:649–68. AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, United States; 2004.
[17] Black M, Strbac G. Value of bulk energy storage for managing wind power [40] Denkenberger D, Greenblatt JB, Socolow RH. Optimal wind turbine rated speed
fluctuations. IEEE Trans Energy Convers 2007;22(March):197–205. taking into account array effects and capacity factor. Electric Power 2004.
[18] Garcia-Gonzalez J, de la Muela RMR, Santos LM, Gonzalez AM. Stochastic joint [41] Malcolm DJ, Hansen AC. WindPACT turbine rotor design study. National
optimization of wind generation and pumped-storage units in an electricity Renewable Energy Laboratory, Golden, CO2002.
market. IEEE Trans Power Syst 2008;23:460–8. [42] Wiser R, Bolinger M. 2010 Wind Technologies Market Report. US Department
[19] Sioshansi R, Denholm P. The value of concentrating solar power and thermal of Energy June 2011.
energy storage. IEEE Trans Sust Energy 2010;1:173–83. [43] Denkenberger D. Optimal wind turbine rated speed taking into account array
[20] Beaudin M, Zareipour H, Schellenberglabe A, Rosehart W. Energy storage for effects, the larger system, and climate impacts., Master of Science and
mitigating the variability of renewable electricity sources: an updated review. Engineering, Department of Mechanical and Aerospace Engineering,
Energy Sust Dev 2010;14:302–14. Princeton University, Princeton, NJ; 2005.
[21] Sorensen B. Fluctuating power-generation of large wind energy converters, [44] Sharma A, Chiu HH, Ahrens FW, Ahluwalia RK, Ragsdell KM. Design of
with and without storage facilities. Solar Energy 1978;20:321–31. optimum compressed air energy-storage systems. Energy 1979;4:201–16.
[22] Cavallo AJ. High-capacity factor wind energy systems. J Solar Energy Eng, Trans [45] Ahrens FW, Sharma A, Ragsdell KM. Computer aided optimal design of
ASME 1995;117:137–43. compressed air energy storage systems. Trans ASME J Mech Des
[23] DeCarolis JF, Keith DW. The economics of large-scale wind power in a carbon 1980;102:437–45.
constrained world. Energy Policy 2006;34(March):395–410. [46] Korpaas M, Holen AT, Hildrum R. Operation and sizing of energy storage for
[24] Swider DJ. Compressed air energy storage in an electricity system with wind power plants in a market system. Int J Electr Power Energy Syst
significant wind power generation. IEEE Trans Energy Convers 2003;25(October):599–606.
2007;22(March):95–102. [47] Paatero JV, Lund PD. Effect of energy storage on variations in wind power.
[25] Arsie I, Marano V, Rizzo G, Moran M. Energy and economic evaluation of a Wind Energy 2005;8(October–December):421–41.
hybrid power plant with wind turbines and compressed air energy storage. In: [48] Castronuovo ED, Lopes JAP. Optimal operation and hydro storage sizing of a
2006 ASME power conference, Atlanta, GA, United States; 2006. wind-hydro power plant. Int J Electr Power Energy Syst
[26] Daneshi H, Srivastava AK, Daneshi A. Generation scheduling with integration 2004;26(December):771–8.
of wind power and compressed air energy storage. In: 2010 IEEE PES [49] Arsie I, Marano V, Nappi G, Rizzo G. A model of a hybrid power plant with wind
transmission and distribution conference and exposition: smart solutions for turbines and compressed air energy storage. In: 2005 ASME Power Conference,
a changing world; 2010. Chicago, IL, United States; 2005. p. 987–1000.
[27] Zafirakis D, Kaldellis JK. Autonomous dual-mode CAES systems for maximum [50] Bossanyi EA, Maclean C, Whittle GE, Dunn PD, Lipman NH, Musgrove PJ.
wind energy contribution in remote island networks. Energy Convers Manage Efficiency of wind turbine clusters. Precision Engineering; 1980. p. 401–6.
2010;51:2150–61. [51] Bolinger M, Wiser R. Understanding trends in wind turbine prices over the past
[28] Nakhamkin M. Novel compressed air energy storage concepts. In: Presented at decade. Lawrence Berkeley National Laboratory, Environmental Energy
the electricity storage association meeting 2006: Energy Storage in Action, Technologies Division LBNL-5119E, October 2011.
Knoxville, Tenn.; 2006. [52] Garver LL. Effective load carrying capability of generating units. IEEE Trans
[29] Nakhamkin M, Wolk RH, van der Linden S, Pate M. New compressed air energy Power Apparatus Syst 1966;PAS-85:910–9.
storage concept improves the profitability of existing simple cycle, combined [53] Milligan M, Porter K. Wind capacity credit in the United States. In: 2008 IEEE
cycle, wind energy, and landfill gas power plants. In: Proceedings of the ASME Power & energy society general meeting, 20–24 July 2008, Piscataway, NJ, USA,
Turbo Expo 2004, Vienna, Austria; 2004. p. 103–10. 2008. p. 5.
[30] Tuschy I, Althaus R, Gerdes R, Keller-Sornig P. Compressed air energy storage [54] EIA. Annual Energy Outlook 2011 With Projections to 2035, Office of
with high efficiency and power output. VDI Berichte; 2002. p. 57–66. Integrated Analysis and Forecasting, Energy Information Administration, US
[31] Ekren O, Ekren BY. Size optimization of a PV/wind hybrid energy conversion Department of Energy, Washington, D.C. DOE/EIA-0383(2011); 2011.
system with battery storage using simulated annealing. Appl Energy [55] Chiesa P, Consonni S, Kreutz T, Williams R. Co-production of hydrogen,
2010;87:592–8. electricity and CO2 from coal with commercially ready technology. PartA:
[32] Kellogg WD, Nehrir MH, Venkataramanan G, Gerez V. Generation unit sizing Performance and emissions. Int J Hydrogen Energy 2005;30(July):747–67.
and cost analysis for stand-alone wind, photovoltaic, and hybrid wind/PV [56] Cavallo AJ, Keck MB. Cost effective seasonal storage of wind energy. Houston,
systems. IEEE Trans Energy Convers 1998;13:70–5. TX, USA: American Society of Mechanical Engineers, Solar Energy Division
p
[33] Brown PD, Pe ßas Lopes JA, Matos MA. Optimization of pumped storage (Publication) SED; 1995.
capacity in an isolated power system with large renewable penetration. IEEE [57] BNEF. Coming to America? New Asian Turbine Exports to the United States.
Trans Power Syst 2008;23:523–31. Bloomberg New Energy Finance 21 April, 2011.
[34] Bae HM, Devine MD. Optimization models for economic design of wind power- [58] Wang MQ. GREET 1.5 – Transport fuel-cycle model. Center for Transportation
systems. Solar Energy 1978;20:469–81. Research, Energy Systems Division, Argonne National Laboratory, Argonne, IL
[35] Mitchell K, Nagrial M, Rizk J. Simulation and optimisation of renewable energy ANL/ESD-39, vol. 2; August 1999.
systems. Int J Electr Power Energy Syst 2005;27:177–88. [59] Macchi E, Lozza G. Study of thermodynamic performance of caes plants,
[36] Brunetto C, Tina G. Optimal hydrogen storage sizing for wind power plants in including unsteady effects. In: Gas Turbine Conference and Exhibition,
day ahead electricity market. IET Renew Power Generation 2007;1:220–6. Anaheim, CA, USA; 1987. p. 10.
[37] Takahashi R, Tamura J, Fukushima T, Sasano E, Shinya K, Matsumoto T. A [60] EPRI-DOE. Handbook of energy storage for transmission and distribution
determination method of power rating of energy storage system for smoothing applications. EPRI, DOE, Palo Alto, CA, Washington, DC; 2003.
wind generator output. In Proceedings – the 12th international conference on [61] EPRI-DOE. Energy storage for grid connected wind generation applications,
electrical machines and systems, ICEMS 2009; 2009. EPRI, DOE, Palo Alto, CA, Washington, DC 1008703; 2004.
[38] Torres JL, Prieto E, Garcia A, De Blas M, Ramirez F, De Francisco A. Effects of the [62] EIA. Annual Energy Outlook 2007 With Projections to 2030. Office of
model selected for the power curve on the site effectiveness and the capacity Integrated Analysis and Forecasting, Energy Information Administration, US
factor of a pitch regulated wind turbine. Solar Energy 2003;74:93–102. Department of Energy, Washington, D.C. DOE/EIA-0383(2007); February 2007.

Anda mungkin juga menyukai