Solar Energy 78 (2005) 382395

www.elsevier.com/locate/solener

Technicaleconomic analysis of wind-powered pumped

hydrostorage systems. Part I: model development
C. Bueno a, J.A. Carta
a

b,*,1

Department of Renewable Energy and Water, Technological Institute of the Canary Islands, Playa de Pozo Izquierdo s/n,
35119 Santa Luca, Gran Canaria, Canary Islands, Spain
b
Department of Mechanical Engineering, University of Las Palmas de Gran Canaria, Campus de Tara s/n,
35017 Gran Canaria, Canary Islands, Spain
Received 25 September 2002; received in revised form 7 July 2004; accepted 9 August 2004
Available online 18 September 2004
Communicated by: Associate Editor Cornelis P. van Dam

Abstract
In this paper a model is presented for the technical and economic sizing of the various components that make up
medium sized wind-powered pumped hydrostorage systems. A further aim of this model is the optimisation of the operation of such systems, thereby making full use of the synergy of the unit as a whole.
A general model is described for use as an analytical tool in implementing such systems in topographically suitable
sites with sucient wind resources. The general model developed allows for six strategies for the system operating conguration. Each strategy is based on the hypothesis that there is a centralised operator to control all the system components, except for the load systems.
The characteristics and unit energy cost of each technically feasible combination of components are determined by
applying the model. This enables the selection of the most viable composition for the system from an economic point of
view given certain technical restrictions.

2004 Elsevier Ltd. All rights reserved.
Keywords: Wind penetration; Pumped storage; Model

1. Introduction
In order to satisfy demand the energy producing
companies have traditionally used electricity generating

Corresponding author. Tel.: +34 928 45 1483; fax: +34 928

45 1484.
E-mail address: jcarta@dim.ulpgc.es (J.A. Carta).
1
ISES* Regular Member No. 0111319.

plants that employ a variety of economic and technical

characteristics. Conventional, high output, generating
plants are generally employed to cover the basic demand. Peak demand times are usually covered by smaller sized cyclically operating plants which work with
coal, oil, gas and particularly with cyclical water-pumping systems (Ramage, 2000). These latter systems use the
excess electrical production in periods of low demand to
pump water to a reservoir situated at a certain height.
When required it is then recovered by means of a turbine

0038-092X/$ - see front matter
2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.solener.2004.08.006

C. Bueno, J.A. Carta / Solar Energy 78 (2005) 382395

383

Nomenclature
maximum percentage of electrical energy demand that can be met by electrical wind-sourced energy without aecting the safety and stability of the system, %
b
parameter which can take a value between 0 and 1
d
the percentage d of peak demand which has to be covered with hydraulic energy, %
g0
eciency of the PS
g
eciency of the HP and its electrical system
g01
eciency of the pumps P1
g02
eciency of the pumps P2
u
wake factor
q
water density, kg/m3
n
reliability factor of the WT
C(k, j) cost of the unit energy supplied by the conguration system k operating in accordance with strategy i,
Euros/kW h
C(j, i)
minimum local cost of unit energy supplied which is obtained amongst the Nj congurations, Euros/
kW h
C(PCP(t)) CP production cost, Euros/kW
CF
capacity factor of the WP, %
CI
the investment costs of the electrical and control infrastructure, Euros
Cmin
minimum overall cost of the unit energy supplied, Euros/kW h
CP
conventional electrical generator set
CPS
conventional electrical power system
CS
control system
DLS(t) electrical demand in time t, kW h
DLS,max maximum peak demand, kW
ECPS(t) conventional energy of the CPS in time t, kW h
EHP(t) hydraulic energy of the HP in time t, kW h
Etotal(t) the total energy generated by a combination of components which must satisfy the demand DLS(t), kW h
EWP(t) wind energy of the WP in time t, kW h
g
acceleration due to gravity, m/s2
he
height of the rotor shaft of the WT, m
Hg
gross height, m
Hl
hydraulic losses, m
Hn
net height, m
HP
hydroelectric plant
hr
height of wind speed measurement, m
HT
hydraulic turbine and its corresponding electrical generator
Hu
useful height, m
IRR
the internal rate return, %
L
number of years over which the investment in the system is to be recovered, years
LR
lower reservoir
LS
load system
M
number of initial dierent congurations of the system which can be analysed on the hypothesis that each
subsystem can meet the demand by itself
nc
number of conventional generator sets that make up the CPS
ncc
number of CPs connected
nc0,j
initial number of CP units of type j (j = 1, . . . ,Ncp)
Ncp
number of dierent types of CP
Nht
number of dierent types of HT
number of congurations which are analysed, from an initial conguration j (j = 1, . . . ,M), in order to
Nj
determine the optimum economic conguration
Np1
number of dierent types of pumps P1
Np2
number of dierent types of pumps P2
a

384

NPV
nt

C. Bueno, J.A. Carta / Solar Energy 78 (2005) 382395

the net present value, Euros

number of hydraulic turbines and their corresponding electrical generators which make up the hydroelectric plant
ntc
number of HT connected
nt0,j
initial number of HT units of type j (j = 1, . . . ,Nht)
nw
number of wind turbines which make up a wind park
nw0,j
initial number of WT units of type j (j = 1, . . . ,Nwt)
Nwt
number of dierent types of WT
O&M annual operating and maintenance costs, Euros
PBP
the payback period, years
PCP(t) average hourly power supplied by the PC, kW
PCP,j
nominal unitary power of the CPs of type j, kW
PHT,j
nominal unitary power of the HTs of type j, kW
Pp1,j
nominal unitary power of the P1 pumps of type j, kW
Pp2,j
nominal unitary power of the P2 pumps of type j, kW
PS
water pumping station
PWT(v) the characteristic curve of the WT
PWT,j
nominal unitary power of the WTs of type j, kW
P min
minimum operating power of CP, kW
CP
P max
maximum operating power of CP, kW
CP
(PCPS)0 initial nominal conventional power which needs to be installed on the hypothesis that the electrical power
generated with the CPS has to cover the maximum peak demand
(PHP)0 initial nominal hydraulic power which needs to be installed on the hypothesis that the electrical power
generated with the HP has to cover the maximum peak demand, kW
(PPS)0 initial nominal pumping power which needs to be installed on the hypothesis that the electrical power
generated with the PS has to cover the maximum peak demand, kW
(PWP)0 initial nominal wind power which needs to be installed on the hypothesis that the electrical power generated with the WP has to cover the maximum peak demand, kW
(Qmax)0 initial maximum water ow that drives the turbines
Qmt
technical minimum ow of the HT, m3/h
QnT
nominal ow of the HT, m3/h
Qp(t)
ow pumped by the PS at each instant, m3/h
Qt(t)
ow used by the HP at each instant, m3/h
r
annual discount rate (opportunity cost)
S0
initial subsidy, Euros
ta
autonomous time of the UR. Time in which the HP can cover a percentage d of the peak demand, h
tc,i
instant of the start of the ith connection of the HP hydraulic system. This instant coincides with the
instant in which the volume of water in the UR (V(t)) reaches its maximum value Vmax, h
tco
time to estimate the control volume, h
td,i
instant of the ith disconnection of the hydraulic system HP, h
Dt
simulation interval (1 h)
UR
upper reservoir
V(t)
volume of water in the upper reservoir at time t, m3
Vco
control volume, or volume of water required in the upper reservoir so that the HP can cover by itself all
the electricity demand, m3
ve(t)
wind speed, extrapolated at height he of the rotor shaft of the WT, as a function of time, m/s
Vmax
maximum volume of the upper reservoir, m3
Vmin
minimum volume of water required in the upper reservoir so that the turbines can operate, m3
Vp(t)
volume of water pumped by the PS in time t, m3
vr(t)
wind speed, recorded at a reference height hr, as a function of time, m/s
Vt(t)
volume of water used by the HT, m3
WP
wind Park
WS
water storage system
WT
wind turbine

C. Bueno, J.A. Carta / Solar Energy 78 (2005) 382395

z1
z1c
z2
z2c
z0
(z1)0,j
(z2)0,j

385

number of pumps P1 which the subsystem PS has

number of pumps P1 connected
number of pumps P2 which the subsystem PS has
number of pumps P2 connected
surface roughness of the land, m
initial number of pump units P1 of type j (j = 1, . . ., Np1)
initial number of pump units P2 of type j (j = 1, . . ., Np2)

to cover the peak load periods. In this way, customer demand can be supplied at minimum cost with a highly
reliable system (Conejo, 1990).
In the last few decades, pumping with storage has
played an important part in the development of electrical power systems throughout the world (Kuan, 1989).
It is considered by engineers and designers to be an
attractive alternative, allowing a considerable amount
of energy to be stored. As the generating equipment is
highly reliable, the power output can be easily regulated
and a practically constant eciency can be maintained
in the power output range.
From the 1970s onwards, and probably due to the
complementary characteristics of pumped hydrostorage
systems and wind parks (Chabot, 2000), there have been
several projects designed with the aim of making use of
wind energy to pump water for its later hydroelectric use.
However, no general methodology has been developed in
the projects undertaken in relation to the sizing optimisation of the systems components and its operating modes.
In this paper a general model is presented for the
implementation of pumped hydrostorage systems using
wind energy. This model is suitable for any site that
meets the necessary topographical requirements and
has sucient wind resources.
The general model developed allows for six strategies
for the system operating conguration. Each strategy is
based on the hypothesis that there is a centralised operator to control all the system components, except for the
load systems.
The application of the proposed model provides the
characteristics of each technically feasible and commercially available combination, as well as the costs per unit
energy generated, the net present value (NPV), the internal rate return (IRR) and the payback period (PBP) for
each combination. This will enable a decision to be taken
on the most viable composition of the system from an economic point of view given various technical restrictions.

(WP) with nw identical wind energy conversion systems

(WT) with synchronous generators; a hydroelectric
plant (HP) with nt identical hydraulic turbines and their
corresponding electrical generators (HT); a water pumping station (PS) with two motor pump sets z1 and z2 of
dierent nominal powers; a water storage system (WS),
consisting of an upper (UR) and lower (LR) reservoir;
a control system (CS); 2 a conventional electrical power
system (CPS), consisting of nc sets (CP); and the load
system (LS). The LS is dependant on the load structure
(domestic, industrial, commercial) and the climatic characteristics of the region whose energy requirements the
system aims to cover, but which here are considered beyond the control of the CS.

3. System operating strategies

The general model developed allows for six strategies
for the system operating conguration. Each strategy is
based on the hypothesis that there is a centralised operator to control all the system components, except for the
loads LS.
3.1. Strategy 1
There are two independent electrical systems:
(a) WP is isolated from the conventional grid (switches
a and b open and switch c closed: see Fig. 1). Electrical energy EWP(t) is supplied exclusively to cover
the electrical consumption DPS(t) of the pumps, Eq.
(1), which are connected and disconnected from the
system in accordance with the energy variations of
the wind park. The proposed operational mode for
the WPPS system is similar to that analysed by
Carta et al. (2003) for the case of a WP connected
to a modular reverse osmosis desalination plant.
2

2. General conguration of the system

Fig. 1 shows an outline of the general conguration
of the proposed system. It consists of a wind park

In Fig. 1 the control network lines are not shown so as not

to muddle up the diagram. However, an attempt has been made
to reect the existence of a relation between the CS and the
various subsystems through the use of numbers located on the
CS and these subsystems.

386

C. Bueno, J.A. Carta / Solar Energy 78 (2005) 382395

Fig. 1. General diagram of the proposed system.

(b) HP and CPS are connected to the same grid and

cover the demand DLS(t) in time t, Eq. (2), with
the following order of priority. Firstly, the control
system attempts to cover the demand with hydraulic
energy EHP(t), provided the volume (V(t)) of water
in the upper reservoir is higher than the minimum
technical value (Vmin) which allows the turbines to
be driven (V(t) P Vmin). Secondly, the control system covers the demand with conventional energy
ECPS(t) from the CPS. This will be connected if
the HT are unable to cover all the demand.
DPS t EWP t

DLS t 1
bEHP t bECPS t

where Eq. (2) is constrained by Eq. (3)

b 0 if tc;i 6 t < td;i  ! EHP t > DLS t

b 0 if EHP t P DLS t
0 < b < 1 if 0 < EHP t < DLS t

there is a control volume Vco. This Vco is established

to enable the subsystem HP to satisfy by itself the maximum demand (DLS,max) 3 during a given time tco. The
HP are disconnected whenever V(t) 6 Vco. Fig. 2 shows
how no energy from hydraulic resources is fed into the
electrical grid during the periods a when the UR is
being lled. During these periods the demand is covered
exclusively by fossil-fuel sourced energy. During the
periods b, when the UR is being emptied, the demand
is satised exclusively by energy from hydraulic resources. With this approach it is possible to forecast
the periods of connection (tc,i) and disconnection (td,i)
of the HP, making maintenance operations easier and
minimising the number of connections/disconnections
of the HP and the CPS. So, in this strategy Eqs. (1)
and (2) of Strategy 1 are also valid, though now Eq.
(2) is restricted by Eq. (4) and not by Eq. (3).
b 1 if td;i
1 6 t < tc;i  ! EHP t 0

b 1 if EHP t 0
3.2. Strategy 2
This strategy only diers from Strategy 1 in the ith
stages of initial connection (tc,i) of the HP, and will only
be executed when the volume V(t) of the upper reservoir
reaches its maximum value (V(t) = Vmax), and when

when
tc;i ! V t tc;i V max ; td;i ! V t td;i < V co ;
td;i
1 ! V t td;i
1 < V co

Or a percentage d of the maximum demand.

C. Bueno, J.A. Carta / Solar Energy 78 (2005) 382395

387

Fig. 2. Schematic operation of hydroelectric (HP) controlling reservoir volume.

3.3. Strategy 3
All the generating subsystems are connected to the
same electrical grid (switches a and b closed and switch
c open, Fig. 1). As was mentioned in the introduction to
this paper, the aim is for as much of the electricity demand (DLS) as possible to be covered by renewable-sourced energy (wind and hydraulic). Consequently, it is
important to remember that the direct feeding of windsourced energy into a small or medium sized system
can have serious drawbacks. When there is a high degree
of wind energy penetration the variations in the active
power generated by such levels (due to variations in
the wind speed) cause imbalances between the power
output and demand. These imbalances can result in
frequency and voltage variations that can lead to potentially dangerous operating conditions (Gavanidou et al.,
1993; Papadopoulos et al., 1991). Therefore, in order to
avoid serious problems (which could aect the safety
and stability of the system) the wind-sourced energy
has to be limited to a specic percentage of the conventional synchronous capacity connected to the grid,
which in turn depends on the load of the system (Kabouris and Contaxis, 1992; Gavanidou et al., 1993). A dynamic study that takes into account wind speed
variations and other operating conditions, such as the
size of the system, is required to determine the percentage of wind-sourced energy which can be fed into a particular electrical system. In the model used in this paper,
a is dened as the maximum percentage of electrical
energy demand that can be met by wind-sourced energy
without aecting the safety and stability of the system.

a is an input parameter in this model and can take a value of between 0% and 100%.
With Strategy 3 the potential drawback of a high
direct wind penetration in the power grid (a high)
can be resolved by the use of a HP system. This enables the percentage of renewable-sourced energy to be
increased (Schafer and Simond, 1998; Mansoor,
2000).
In this strategy the CS controls the dierent subsystems in accordance with the following procedures:
(a) A minimum percentage (100
a) of the energy
demanded by the LS will be covered by the HP
and/or the CPS in the following sequence, Eq. (6),
(Eq. (6) is conditioned by Eqs. (7) and (9)). First,
by the HP if the volume of water in the upper reservoir (V(t)) is sucient; secondly, by the HP in parallel with the CPS if the water level in the upper
reservoir is not sucient; and thirdly by the CPS
alone if the level of water in the upper reservoir is
lower than the technical minimum, Vmin required
to drive the turbines.
(b) A maximum percentage a of the electricity demand
will be covered by wind-sourced energy if there is
sucient wind. The CS will attempt to connect the
pumps in succession with the aim of reaching an
electrical balance, Eq. (8), provided there is sucient energy and provided the wind energy is higher
than aDLS(t)/100.

DLS t 1
bEHP t bECPS t EWP t

388

C. Bueno, J.A. Carta / Solar Energy 78 (2005) 382395

1
bEHP t bECPS t P 100
aDLS t=100

EWP t 6 aDLS t=100

DPS t EWP t
aDLS t=100 if EWP t > aDLS t=100

DPS t 0 if EWP t 6 aDLS t
8

b 0 if EHP t EWP tDLS t;


aDLS t
EWP t 6
100

0 < b < 1 if 0 < EHP t EWP t < DLS t;


aDLS t
EWP t 6
100

9
DLS t ECPS t >
=

if td;i
1 6 t < tc;i

a 0%
>
;
DPS t EWP t

10

In the emptying stage demand is covered in the same

way as in Strategy 4.
In summary, in periods a (Fig. 2)when the UR is
being lledenergy from hydraulic resources is not fed
into the grid and the demand is covered exclusively by
fossil-fuel sourced energy; in periods bwhen the
UR is being emptieddemand is covered by energy
from both hydraulic and wind resources.
3.6. Strategy 6

b 1 if EHP t 0
9
3.4. Strategy 4
The non-controlled load demand is covered by dierent electricity generation subsystems, Eq. (6), depending
on whether the upper reservoir has reached its maximum
volume, Vmax or not. If the upper reservoir has not
reached its maximum volume (b = 1 in Eq. (4)), a minimum percentage (100
a) of demand will be covered by
the CPS and a maximum percentage a by the WP, Eq.
(7), provided there is sucient wind. If the wind energy
is higher than aDLS(t)/100 the pumps are connected in
succession to drive the water from LR to UR in order
to attempt to balance the electrical system, Eq. (8).
When the volume of water in the upper reservoir has
reached its maximum value (t = tc,i) the percentage of
the demand which was being covered by the CPS
will be covered by the HP (b = 0 in Eq. (4)). The advantage of this approach compared to Strategy 3 is that
it is possible to forecast the periods of connection/
disconnection of the CPS and the HP, making maintenance easier and reducing the number of connections/
disconnections.
During the periods a (Fig. 2) of the lling of the
UR, energy from hydraulic resources is not fed into
the grid, and so demand is covered by fossil-fuel and
wind-sourced energy. During the periods b of the
emptying of the UR, demand is met by energy from
hydraulic resources and wind-sourced energy.
3.5. Strategy 5
This strategy is a particular instance of Strategy 4
with the only dierence being in the lling stage of the
upper reservoir, when all non-controlled demand is covered by the CPS and all the wind-sourced energy is used
for water pumping, Eq. (10).

This deals with a particular instance of Strategy 4

with the only dierence being in the emptying stage of
the upper reservoir, when all the demand is covered by
the HP, while all the wind-sourced energy is used for
water pumping, Eq. (11).
In the lling stage demand is covered in the same way
as for Strategy 4.
In summary, in periods a (Fig. 2)when the UR is
being lledenergy from hydraulic resources is not fed
into the grid and the demand is met by wind and fossilfuel sourced energy; in periods bwhen the UR is
being emptiedthe demand is met exclusively by energy
from hydraulic resources.
9
DLS t EHP t >
=
a 0%
11
if tc;i 6 t < td;i
>
;
DPS t EWP t
In all of the above strategies the volume of water V(t)
stored in the reservoir in time t is given as a function
of the existing volume V(t
Dt), of the volume
pumped Vp(t) and of the volume used by the HT
Vt(t):
V t V t
Dt V p t
V t t with
V tV min _ V t P V co

12

4. Initial sizing methodology

The initial starting point data for the sizing of the
system are essentially as follows: the evolution in time
of the hourly electrical power demand, the gross height
between the reservoirs, the number of autonomous running hours of the UR, the control time and the capacity
factor of the WP (Fig. 3).
The power input/output of the various subsystems
must rst be dened in order to initiate the execution
of the model that has been developed. The starting point
for the determination of these initial power estimates is

C. Bueno, J.A. Carta / Solar Energy 78 (2005) 382395

389

determine the initial nominal hydraulic power (PHP)0

which must be installed, and so
P HP 0 P DLS;max ! Qmax 0
P HP 0 =gqH u
H l g
P HP 0 =ggqH n 

13

where (Qmax)0 is the maximum water ow that drives the

turbines; g is the acceleration due to gravity; q the water
density; Hu the useful height; Hl the hydraulic losses,
which are estimated at 0.5% of the gross height Hg; 4
Hn the net height; and g the eciency of the HP and
its electrical system.
The ow used by the HP at each instant Qt(t) can be
determined from Eq. (13), using the demand curve,
DLS(t), as a starting point. After establishing a specic
number of hours for the autonomous running of the
upper reservoir (ta) and a control time (tco), the maximum required volume Vmax of the UR can be calculated,
Eq. (14)

Z t a

Qt t dt V co ; V co tco DLS;max
V max P max
0

14
After determination of the net pumping height Hn, calculation of the ow (Qmax)0 which the HT require for
maximum demand, and the establishment of the eciency of the motor pumps g 0 , it is possible to obtain
the initial nominal pumping power, Eq. (15), and the
minimum initial installable nominal power of the WP,
Eq. (16), estimating a capacity factor (CF) of the WP
(Hau, 2000).
P PS 0 P qgH g H l Qmax 0 =g0

15

P WP 0 P P PS 0 =CF=100

16

Fig. 3. Flowchart to determine initial power of each subsystem.

The initial nominal power (PCGS)0 which has to be installed must be able to cover by itself all the demand
and, therefore, (PCPS)0 P DLS,max must be satised.
If dierent models of the components of each subsystem are to be tested (with dierent technical and economic characteristics) then, after determination of the
initial power estimates of the dierent subsystems, an
initial conguration of the system will be generated for
each combination of components, Eq. (17). In Eq.
(17), j varies between one and the number of dierent
models of components. The numbers of dierent models
of wind turbines, hydraulic turbines, higher power
pumps, lower power pumps, and conventional generator
sets, are represented by Nwt, Nht, Np1, Np2 and Ncp,
respectively.

the requirement that each of the subsystems must be

able to satisfy the demand on its own.
The hypothesis that the power output of the HP has
to cover the maximum peak demand, DLS,max is used to

4
The useful height and the gross height are considered to be
the same in the pre-sizing stage.

390

C. Bueno, J.A. Carta / Solar Energy 78 (2005) 382395





nw0;j P INT P WP 0 =P WT;j ; nt0;j P INT P HP 0 =P HT;j ;


nc0;j P INT P CPS 0 =P CP;j ;


z1 0;j P INT P PS 0 =P p1;j ;
nh
i
o
z2 0;j P INT P PS 0
z1 0;j P p1;j =P p2;j
17
As can be deduced from this design description, for a
given degree of autonomy and a given height between
reservoirs, the number M of initial dierent congurations of the system which can be analysed will be given
by Eq. (18).
M

N wt !
N ht !
N p1 !
N p2 !
N cp !
N wt
1! N ht
1! N p1
1! N p2
1! N cp
1!
18

5. Algorithm of the procedure

As can be seen in Fig. 4, the algorithm of the procedure starts with the reading of certain basic data including: the energy demand DLS(t) over one full year
(t = 1, . . . , 8760 h); the autonomous time of the reservoirs; ta the gross height between reservoirs Hg; the percentage of wind penetration a; the percentage d of the
peak demand which has to be covered with hydraulic energy, 5 etc.
The rst calculation is to determine the initial nominal powers of each subsystem, as shown in Fig. 3. The
number M of initial congurations is determined, Eq.
(18), using a database reading of the characteristics of
the various models of components of each subsystem
(WP, HP, PS, CPS).
Following this, the numbers of initial components are
determined (nw0,j, nt0,j, nc0,j, (z1)0,j, (z2)0,j) of each subsystem, Eq. (17), for each number j of initial congurations.
Due to the hypothesis involved in obtaining such presizing, these do not necessarily have to coincide with the
optimum economic conguration. Therefore, the proposed model analyses the Nj possible combinations of
components which can be established for each initial
conguration j, Eq. (19). This analysis is carried out for
each of the six previously described strategies (i = 16).
h
i
h
i


 
z1 0;j !
z2 0;j !
nw0;j !
nt0;j !
i h
i
 
 h
Nj 
nw0;j
1 ! nt0;j
1 ! z1
1 ! z2
1 !
0;j
0;j
19
In the loop k = 1 . . . Nj the algorithm rejects the technically unfeasible combinations and determines the costs
5

ta and d can be incrementally modied by the model,

though this has not been represented in the owchart in Fig. 4
to avoid complications.

per unit energy supplied of the remaining combinations.

In other words, a simulation of the annual operation of
the system is carried out in accordance with strategy i,
for each possible combination k, of the Nj combinations
which can be established from an initial combination j
(as described in Section 3 of this paper). The total energy
Etotal(t) generated by this combination k of components
must satisfy the demand DLS(t) at any moment t; if this
technical restriction is not met the combination k will be
eliminated. If this restriction is met, the combination k
of components is evaluated from an economic point of
view and the cost of the unit energy supplied, C(k, i),
the net present value, the internal rate return and the
payback period are determined.
Once the Nj combinations of components of the system have been analysed, a search is carried out to determine the system sizing requirements. Using as a starting
point the initial combination j, these must satisfy the annual energy demand at minimum cost, C(i, j), when
operating under strategy i.
When the M congurations have been evaluated the
composition of the system with the least cost per unit energy supplied can be determined for each strategy i, as
well as the global optimum strategy and its composition.
The algorithm of the model developed uses the Wind,
Hydraulic and Diesel Power Output Models, and Pumps
Power Input Model to carry out the annual system performance simulation.
5.1. Wind power output model
The Wind Power Model requires as basic input data
the hourly wind speed evolution ve(t), the height he of the
rotor shaft of the WT, and the wind powerspeed curves
PWT(v), (Fig. 5).
Wind power output model takes as a starting point
the hypothesis that the available wind speeds vr(t) have
been measured at a reference height hr lower than the
height he. A logarithmic law is then used (Mikhail
and Justus, 1981) to quantify the vertical wind speed
prole, Eq. (20), assuming that the land is at, homogeneous and with a surface roughness z0 (Frost et al.,
1978).
ve t vr tlnhe =z0 = lnhr =z0 

20

The electrical power output of a WT at time t,PWT(t), is

determined from the power curve, PWT(v), as a function
of the wind speed v(t) (Fig. 5).
WT manufacturers give the PWT(v) curve of their
machines in discrete form (in n points). So, in order to
determine the power generated by the WT when the wind
speed v is between two points j and j + 1 a linear
approximation is made between them, Eq. (21)
P WT v P WT vj1
P WT vj v
vj =vj1
vj
P WT vj
21

C. Bueno, J.A. Carta / Solar Energy 78 (2005) 382395

391

Fig. 4. Algorithm of the optimization procedure.

To estimate the power produced by the WP it is assumed

that all the WT generate the same power. However, the
WP power output is corrected using a reliability factor
n of the WT and a wake factor n (Manwell et al., 2002).

So, the hourly energy generated by the WP will be given

by Eq. (22).
EWP t nw P WT tDtnu

22

392

C. Bueno, J.A. Carta / Solar Energy 78 (2005) 382395

Qt t
nt QnT P
0
<

23

If the result of Eq. (23) is null or positive a comparison is

made, Eq. (24), of the available electrical energy from
hydraulic resources and the demand, DLS(t).
DLS t=jggqH n Dt
QnT 6
0
>

Fig. 5. Power curves of some of the wind turbines.

where nw is the number of WTs which make up the

WP.

24

If the result of Eq. (23) is negative, a comparison is

made, Eq. (25), of the ow Qt(t) and the technical minimum ow of the HP system. If Eq. (25) is negative or
null for all j (number of hydraulic turbines connected)
it is concluded that it is impossible to supply energy
from hydroelectric resources (EHP(t) = 0). If Eq. (25) is
positive for some j, then the number (ntc) of HT which
can be connected is determined, as well as the electrical
energy from hydraulic resources which is available during 1 h; this is compared, Eq. (24), with the demand,
DLS(t).

5.2. Hydraulic power output model

0
Qt t
jQmt P
<

In order to determine whether the subsystem HP

(consisting of nt turbines) is able to supply, at an instant
t, a particular energy demand DLS(t), the CS proceeds
along the following basic steps (Fig. 6):

If the result of Eq. (24) is positive, the number of turbines which can be connected ntc is determined, as well
as the hydraulic power which can be supplied for 1 h,
Eq. (26), and the energy loss of the HP subsystem which
has to be covered by other subsystems, if possible.

Step 1. Reading of the water ow Qt(t) which can be

extracted from the UR at the instant t. This ow
depends on the balance between the ow input
and output.
Step 2. Comparison of the ow Qt(t) and the nominal
ow of the HP system, Eq. (23).

EHP t ntc ggqH n QnT Dt

25

26

If the result of Eq. (24) is negative or null, the ow required to cover the demand is compared with the technical minimum ow of the HP system, Eq. (27). If Eq. (27)
is negative or null for all j, it is concluded that it is
impossible to supply energy from hydroelectric re-

Fig. 6. Flowchart of the hydraulic model.

C. Bueno, J.A. Carta / Solar Energy 78 (2005) 382395

393

Fig. 7. Flowchart of the pumping model.

sources (EHP(t) = 0). If Eq. (26) is positive for some j,

then the number (ntc) of HTs which can be connected
is determined; with, in this case, all the demand requested from the HP being covered.
DLS tjggqH n Dt
Qmt 6
0
>

27

Step 3. Memorisation of the new volume of water in the

reservoir and the number of turbines connected,
ntc.
5.3. Pumps power input model
In order to determine the ow pumped by the subsystem PS (consisting of z1 and z2 pump sets of dierent
capacity), at the instant t, when it is fed by an energy
EWP(t), the CS proceeds along the following steps (Fig. 7):
Step 1. Reading of the water ow Qp(t) which can be
pumped at the instant t. This ow depends on
the balance between the input and output ow
in the UR.
Step 2. Calculation of the number of pumps of highest
capacity (type 1) which could operate as a function of the wind energy supply. The result will
be restricted by the number (z1) of pumps which
the subsystem PS has available, Eq. (28).


Int EWP t=P p1 Dt 6 z1
28
Step 3. Determination of the number (z1c = j) of pumps
of type 1 which have to be connected to satisfy
the ow Qp(t), Eq. (29).



P p1 jg01 = qgHg H l
Qp t 6 0

29

If the result of Eq. (29) is positive for all number of

pumps j, then the number of type 1 pumps which can
operate will be null (z1c = 0). Calculation is then made
of whether a particular number (z2c 6 z2) of type 2
pumps can operate, Eqs. (30) and (31).
If the result of Eq. (29) is negative, there will be
z1c = j pumps of type 1 that can operate. Calculation is
then made as to whether a particular number (z2c 6 z2)
of type 2 pumps can operate, Eqs. (30) and (31).


30
z2c Intf EWP t
z1c P p1 Dt =P p2 Dtg 6 z1


P p2 ig02 P p1 z1c g01 =qgHg H l 
Qp t 6 0

31

Step 4. Determination of the total pumped ow QPS(t),

the wind-sourced energy consumed in pumping
and the wind energy which has to be eliminated
(Carta et al., 2003) since it cannot be used for
water pumping. This waste of wind-sourced
energy can be due to two main reasons: (a)
The upper reservoir is full, (b) the consumption
of the pump sets does not match the prole of
the wind energy supply. Also, the new volume
of water in the reservoir and the number of
pumps z1c and z2c connected must be
memorised.
5.4. Diesel power output model
In order to determine whether the CPS subsystem
(consisting of nc conventional power generating sets) is

394

C. Bueno, J.A. Carta / Solar Energy 78 (2005) 382395

the generator, and (b) from the fuel price (Kirchmayer, 1958). There are dierent proposals
to relate fuel consumption and production
(Lipman and Ineld, 1990; Muselli et al.,
1999; Ashari and Nayar, 1999). 7 In this work
a linear relation has been used between the percentage of fuel consumption and the percentage
of the load seen by the diesel generator
(Carta and Gonzalez, 2001), indicating the
presence of a consumption at zero load of
40% of the full load and a minimum load of
40% of the full load (Hunter and Elliot,
1994). The emission of CO2 (in grams) in the
atmosphere is determined in proportion to the
fuel consumption (in grams), with a proportionality constant of 3.15. 8
5.5. Economic model

Fig. 8. Flowchart of the model of conventional electrical

generation.

able to supply, at an instant t, a particular energy demand DLS(t), the CS proceeds along the following basic
steps (Fig. 8): 6
Step 1. Determination, subject to the production limits
max
P min
CP , P CP ) of the conventional generator sets,
of the number of sets (j = ncc) which need to
be connected to satisfy the demand, Eq. (32)
P min
CP jDt

6 DLS t 6

P max
CP jDt

32

Step 2. Selection of the number of sets which meet Eq.

(32) at minimum cost. It is assumed that for
each conventional generator there is a function
which connects the production in kW, PCP(t),
and the cost in Euros/kW h which this production entails, C(PCP(t)). This relation is derived
(a) from the specic consumption curve
which expresses the amount of fuel consumed
per hour and the electricity production in
kW, and is a measurement of the eciency of
6

For simplicity, in this owchart the case has been modelled

for all the generator sets being equal. However, the model
permits the use of diesel sets of dierent capacities.

The optimisation model uses the investment costs of

the following: the systems commercial components
(market unit sale price); the civil works (cost in terms
of the size of the civil works); 9 the electrical and control
infrastructure, CI (costs proportional to the investment
costs of the components); 10 and the annual operating
and maintenance costs, O&M. 11
Taking into account the M initial congurations, the
aim is to determine the strategy and system conguration which provides the minimum cost Cmin Eq. (33),
per unit energy supplied.
C min minC i;j



min CI
S 0 r=1
1 r
L  O&M i;j
Z 8760

DLS t dt ; i 1; 6; j 1; M 33
0

With the hydraulic generators there is a similar relation

between the ow of water used by the HT and the production of
electrical power (Galiana and Conejo, 2002).
8
Average relation determined for the following diesel oil
(whose sulfur content is less than or equal to 1%. Density:
0.9 kg/l (15 C). Inferior caloric power: 9800 kcal/kg. Superior caloric power: 10,400 kcal/kg. Viscosity kinematic:
8 10
6 m2/s. (50 C). Ask: 0.02% m/m.).
9
For example, the investment costs in the reservoirs can be
introduced through a function which relates the cost per unit
volume in terms of the volume, the costs of the hydraulic plant
in terms of the ow and gross height, etc.
10
For example, the costs associated with the wind park can
be estimated from the investment cost in wind turbines, on the
hypothesis that the investment cost in wind turbines entails
some 70% (modiable) of the investment cost in the wind park.
11
An annual increase in maintenance costs is accepted to
take into account the possible worsening condition of the
machines with the passage of time.

C. Bueno, J.A. Carta / Solar Energy 78 (2005) 382395

where r is the annual discount rate (opportunity cost), L

is the number of years over which the investment in the
system is to be recovered and S0 is an initial subsidy.
Likewise, the net present value (NPV), the internal
rate of return (IRR) and the payback period (PBP) are
determined (Canada et al., 1996; Sherman et al., 1983)
of the possible combinations for each operating strategy.
In this way, it is possible to determine the most viable
option from an economic point of view whilst complying
with certain given technical restrictions.

6. Conclusions
A general model has been presented for the technical
and economic sizing of the various components that
make up medium sized wind-powered pumped hydrostorage systems.
The application of the proposed model provides the
characteristics of each technically feasible and commercially available combination, as well as the costs per unit
energy generated, the net present value (NPV), the internal rate return (IRR) and the payback period (PBP) for
each combination. Accordingly, this will enable a decision to be taken on the most viable composition of the
system from an economic point of view given various
technical restrictions.
Furthermore, the algorithm of the model developed
simulates the operation of each feasible combination
by making energy balance calculations for each of the
8760 h in a year. Therefore, the model developed in this
paper can be used as an analytical tool in implementing
such systems.

References
Ashari, M., Nayar, C.V., 1999. An optimum dispatch strategy
using set points for a photovoltaic (PV)dieselbattery
hybrid power system. Solar Energy 66, 19.
Canada, J.R., Sullivan, W.G., White, J.A., 1996. Capital
Investment Analysis for Engineering and Management.
Prentice-Hall, New Jersey, pp. 1566.
Carta, J.A., Gonzalez, J., 2001. Self-sucient energy supply for
isolated communities: winddiesel systems in the Canary
Islands. The Energy Journal 22, 115145.
Carta, J.A., Gonzalez, J., Subiela, V., 2003. Operational
analysis of an innovative wind powered reverse osmosis
system installed in the Canary Islands. Solar Energy 75,
153168.
Chabot, B., 2000. A long term wind power prospect from
hydropower retrospect and prospect: scenarios and lessons.
In: WIP (Eds.), Proceedings of Wind Power for the 21st
Century, Kassel, Germany, pp. 1922.

395

Conejo, A.J., 1990. Optimal reservoir utilisation of

pumped hydrostorage plants in probabilistic production
costing models intended for generation expansion planning.
Ph.D. Thesis, Department of Electric Power Systems, Royal
Institute of Technology, Stockholm, Sweden.
Frost, W., Long, B.H., Turner, R.E., 1978. Engineering handbook on the atmospheric environmental guidelines for use in wind turbine generator development.
NASA TP-1359, NASA Lewis Research Center, Cleveland,
OH.
Galiana, F.D., Conejo, A.J., 2002. Operacion del sistema de
generacion. In: Exposito, G. (Ed.), Analisis y Operacion de
Sistemas de Energa Electrica. McGraw-Hill, Spain, pp.
261310.
Gavanidou, E.S., Bakirtzis, A.G., Dokopoulos, P.S., 1993. A
probabilistic method for the evaluation of the performance
and reliability of winddiesel energy systems. IEEE Transactions on Energy Conversion 8, 197206.
Hau, E., 2000. Wind-Turbines. Springer, Berlin, pp. 413416.
Hunter, R., Elliot, G., 1994. WindDiesel Systems. Cambridge
University Press, Cambridge.
Kabouris, J., Contaxis, G.C., 1992. Autonomous system
expansion planning considering renewable energy sources.
A computer package. IEEE Transactions on Energy Conversion 7, 374381.
Kirchmayer, L.K., 1958. Economic Operation of Power
Systems. Wiley, New York.
Kuan, T., 1989. Basic planning analysis of pumped-storage.
Ph.D. Thesis, Civil Engineering Department, Colorado
State University.
Lipman, N.H., Ineld, D.G., 1990. WindDiesel Systems.
Wind Energy Conversion Systems. Prentice-Hall, New
Jersey, pp. 330356.
Mansoor, S., 2000. Behaviour and operation of pumped storage
hydroplants. Ph.D. Thesis, University of Wales, Bangor,
UK.
Manwell, J.F., McGowan, J.G., Rogers, A.L., 2002. Wind
Energy Explained. Theory, Desing and Application. Wiley,
New York, pp. 370425.
Mikhail, A.S., Justus, C.G., 1981. Comparison of height
extrapolation models and sensitivity analysis. Wind Engineering 5, 91107.
Muselli, M., Notton, G., Louche, A., 1999. Design of hybrid
photovoltaic power generator, with optimization of energy
management. Solar Energy 65, 143157.
Papadopoulos, M., Malatestas, P., Hatziargyriou, N., 1991.
Simulation and analysis of small and medium size power
systems containing wind turbines. IEEE Transactions on
Power Systems 6, 14531458.
Ramage, J., 2000. Hydroelectricity. In: Boyle, G. (Ed.),
Renewable Energy. Power for a Sustainable Future. Oxford
University Press, New York, pp. 183226.
Schafer, D., Simond, J., 1998. Adjustable speed asynchronous
machine in hydropower plants and its advantages for the
electric grid stability. International Council on Large
Electric Systems (CIGRE), Paris session (France).
Sherman, J.M., Gresham, M.S., Fergason, D.L., 1983. Wind
Systems Life Cycle Cost Analysis. Windbooks, New York,
pp. 179.