Optimal Sizing of Wind-PV-Pumped Hydro Energy

Storage Systems

Solomon Ekoh, Ibrahim Unsal, Alireza Maheri

Faculty of Engineering and Environment
Northumbria University
Newcastle upon Tyne, UK
Solomon.Ekoh@northumbria.ac.uk, I.Unsal@northumbria.ac.uk, Alireza.Maheri@northumbria.ac.uk

Abstract—Storage systems are key components of standalone the economic potentials of PHS systems are reported in [6-10].
hybrid renewable energy systems due to intermittent nature of In terms of payback periods, it is reported that PHS may have
renewable resources. In design of standalone hybrid system, the payback periods as low as 2.5-5.5 years [11].
storage system needs to be optimally sized to guarantee power
quality, system reliability and cost effective energy supply. In this A large number of recent studies have been conducted on
paper, the most mature and traditional long term energy storage standalone renewable systems using PHS and micro PHS [3-
technology, the pumped hydro energy storage (PHS) is integrated 24]. Most of them, however, appear to be mainly focused on
into a standalone wind-PV system. This paper describes the the system background, simulation, modelling, sizing and
mathematical modelling of a PHS and reports a method for energy balances analysis while less attention have been given
optimal sizing of the wind-PV-PHS systems. to the method of optimal sizing of the systems. This paper
presents a method of optimal sizing of standalone hybrid wind-
Keywords—pumped hydro storage; hybrid renewable energy PV and pumped hydro energy storage system driven by not
systems; renewable energy; standalone power system only the performance evaluation but also reliability and cost.

I. INTRODUCTION II. SYSTEM MODELLING

Hybrid renewable energy systems (HRES) comprise of Figure 1 describes the operating principal of the hybrid wind-
renewable power conversion systems with or without fossil- PV-PHS system under study. Here, the PHS system has
fuelled power systems that can work as standalone or grid double-penstocks, one used for pumping water and the other
connected system. Storage and backup systems are key
discharges the water to generate electricity. Unlike the
components of a HRES to maintain the power quality and
simplified single-penstock system which requires a reversible
reliability. Storage systems can be classified as electrical based
(e.g. capacitors, ultra-capacitors and superconducting magnetic pump-turbine, the double-penstock system is mostly employed
energy storage), mechanical based (e.g. pumped hydroelectric due its ability to easily maintain steady voltage and frequency
storage, compressed air energy storage and flywheels), [3].
chemical energy storage (e.g. hydrogen storage with fuel cells),
electrochemical energy storage (conventional rechargeable
batteries such as lead-acid batteries), thermochemical energy
storage (solar fuels) and thermal energy storage (sensible heat
and latent heat storage systems) [1, 2].
PHS has a long history dating back to the early 1890s.
Typically, a PHES plant consists of two storage reservoirs of
water, located at different levels. Water is pumped from the
lower reservoir to the upper reservoir using electric power
during off-peak electricity demand. During peak-load hours the
stored water is released from the upper reservoir through the
hydro turbine into the lower reservoir. Pumped hydro is
predominantly seen as the most encouraging technology option
to increase the level of renewable energy penetration in power
systems [3, 4]. PHS plants rated power depends on pressure of
the water and its flow rate through the turbines as well as the
rated power of the pump/turbine and generator/motor units.
Different PHS plants have power ratings between 1 and 3000
MW, cycle efficiency of (~ 70 to 85%) with lifetime of more Fig. 1. Schematic of a standalone hybrid wind-PV-PHS
than 40 years [1, 2, 5]. In comparison to other storage systems,

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:41:32 UTC from IEEE Xplore. Restrictions apply.
A. Renewable Power Modelling meters (actual head between the upper reservoir and the turbine
The wind power generated by the wind turbine is given by minus the head loss) and KT ,max is the maximum turbine
[25]: efficiency.
2) Upper and Lower Reservoir
1 Depending on the location of the system, if there is no
PWT 3
UVhub AWT C PK EG 
2 natural upper and lower reservoir, two manmade reservoirs
needs to be added to the system. The optimum capacity for the
where PWT is the power output of the wind turbine, U is the air upper and lower reservoirs depends on the design methodology
density (kg/m3), Vhub is the wind speed at hub elevation (m/s), used in sizing. Using a deterministic method based on worst-
2 case scenario and margin of safety, the size of upper and lower
AWT is the wind turbine rotor area (m ), K EG is the overall
reservoirs depend on the peak demand load and the
efficiency of the electrical components and gearbox, and C P is performance of other components in the system. Here it is
the rotor power coefficient [25]. assumed that the upper and lower reservoirs have the same
For the PV array, the power generated is given by: capacities ( VLR VUR ).
The upper reservoir is sized using worst case scenario of no
PPV IAPV K PV  renewable power for a period of autonomy of Ta , in which the
demand load is constantly at ( 1  MOS L ) times of its peak
where I is the solar irradiance (kW/m2), APV is the PV value. This worst-case scenario implies that the water flow rate
array area (m2) and K PV is the PV and converter efficiency (-). from the upper reservoir is at its maximum value of
QUR,max QT ,max . Using another margin of safety, MOSUR to
Furthermore, the power PR generated by the renewables is
take into account the evaporation loss, loss due to leakages
given by (self-discharge rate) and assuming that the upper reservoir
cannot be fully drained, the volume of the upper reservoir can
PR PWT  PPV 
be obtained by:

See reference [25] for more details on wind turbine and PV VUR (1  MOSUR )Ta QT ,max 
power modelling.
3) Pump
B. PHS Power Modelling and Sizing While the size of the turbine is dictated by the peak demand
1) Turbine/Generator Unit load, the pump size depends on the minimum charging time
Having selected the turbine type, its nominal power output Tcharge , which is a system performance-related parameter to be
PT ,nom can be selected based on the peak demand load L peak ,
selected by the designer.
assuming there is no renewable power ( Pd L peak ).

PT ,nom (1  MOS L ) L peak KT ,maxK G (4) PP,nom UgH P QP,max K P,max 

where, KT , max and K G are efficiency of turbine at full load and where, H P is the effective hydraulic head of the upper reservoir
the efficiency of the generator respectively, MOS L is the in meters (actual head between the upper reservoir and the
demand load margin of safety. Turbine efficiency depends on pump plus the head loss) and K P, max is the maximum
the power ratio, the actual turbine output power to its nominal efficiency of the pump. Hence, the maximum water flow rate
value Pt PT ,nom . The hydraulic head is then selected from the from the pump is:
turbine selection chart to ensure it falls within the given range
of available head as well as the elevation difference of the
proposed site. QP, max VLR Tcharge 

The maximum volumetric flow rate of water into the

It should be noted that since the power system is isolated,
turbine Qmax (m3/s) at PT ,nom during discharge is calculated by:
the only source of power for pumping water to the upper
reservoir is the excess power Pe, sch during a scheduled pumping
PT ,nom
 time TP, sch .
QT ,max
UgH T

where, U is water density, g is the acceleration due to gravity, Pe, sch max( PR (t )  L(t ))  t  TP, sch 
H T is the effective hydraulic head of the upper reservoir in

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:41:32 UTC from IEEE Xplore. Restrictions apply.
 With this consideration, one can modify Equation 7 to If L  P R  KT ,maxK G,max PT ,nom : PP 0 , PT L  P /K K
R T G
,
Equation 10 and summarise the size optimisation as shown in SOCUR (t  't )  SOCUR (t )
the algorithm of Figure 2.
If L  P R ! KT ,maxK G,max PT ,nom : PP 0 , PT KT ,maxK G,max PT ,nom ,

PP,nom min^UgH PQP,max K P,max , Pe,sch ` SOCUR (t  't )  SOCUR (t ) , U L  P R  PT

x Case L  PR ! 0 and SOCUR SOCUR,min : PP 0 , PT 0,

0-Input hourly averaged renewable resources and demand SOCUR (t  't ) SOCUR,min , U L  PR
load Vw , I and L respectively
in which, 't 1 hour, and P R and L represent the hourly-
1-Set performance and size optimisation parameters: Tcharge ,
averaged renewable power and load respectively. SOCUR at
Ta , MOS L and MOS UR time t  't is given by:
2-Select wind turbine and PV panel size AWT and APV
3-Using Equations 1 through 3 to find hourly averaged Qin,UR  Qout,UR 't  SOC
SOCUR (t  't )

renewable power PR VUR
UR (t )

4-Size turbine-generator unit:

4.1. Select turbine hydraulic head H T (unless it is fixed In case of power deficit, the generator power output
in case of natural reservoirs) PG ,out should be equal to the power deficit PG ,out
Pd . Having
4.2. Select turbine type efficiencies, the power deficit can be related to the turbine
4.3. Using the efficiency-power ratio chart for the input power and consequently the water flow rate to the
selected turbine type, calculate PT, nom from Equation 4 turbine:
4.4. If H T is within the envelope of selected turbine set
Pd 
hT 0 (head loss as fraction of head) and go to Step PT ,in
KT K G
4.5; otherwise go to Step 4.1
4.5. Use Equation 5 with (1  hT ) H T to find QT, max
4.6. Size pipe between upper reservoir and turbine PT ,in
QUR,out 
4.7. Using QT, max and the size of pipe, calculate hT, new ; UgH T
if hT, new  hT  H go to Step 4.8
Similarly, during a power excess period:
otherwise hT m hT, new , go to Step 4.5
4.8. If (1  hT ) H T is within the envelope of selected PP,out K P PP 
turbine go to Step 5, otherwise go to Step 4.1
5-Find VUR and VLR VUR using Equation 6
6- Size pump unit using Equations 8 through 10 and similar PP,out 
QUR ,in
steps/method as described in Step 4 for the turbine unit. UgH P
7-Evaluate design candidate by conducting power, reliability
and cost analysis The state of the charge of the upper reservoir SOCUR is
8-Stop if optimisation termination criteria met, otherwise go
subjected to the constraints below:
to Step 2

Fig. 2. Size optimisation algorithm SOCUR,min d SOCUR d 1 

where, bottom limit of the SOCUR,min is required to sustain

III. POWER ANALYSIS
a minimum head.
A dispatch strategy is required to conduct system power
analysis. In this study, the following dispatch strategy is used.
IV. CASE STUDY
x Case PR  L ! 0
In this section, the algorithm of Figure 2 is explained
If SOCUR  1 : PP ^
min PP,nom , P R  L `, PT 0, through a case study. The focus is on Steps 4 through 6 and the
SOCUR (t  't ) ! SOCUR (t )
power analysis of Step 7. Reliability and cost analysis can be
conducted as for any other standalone HRES configuration
If SOCUR 1 : PP 0 , PT 0 , SOCUR (t  't ) SOCUR (t ) 1 (e.g. see [25 and 26]).

x Case L  PR ! 0 and Typical daily distribution of resource and demand load for
SOCUR ! SOCUR,min
the site for which the system is designed are shown in Figures

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:41:32 UTC from IEEE Xplore. Restrictions apply.
3 to 5 (Step 0). The wind speed of Figure 3 Vref is given at calculated: wind turbine hub elevation hhub 25m ,
elevation href 3.5m . The surface roughness length of the site K EG 0.93 and K PV 0.16 . The model of C P as a function of
is also given z0 0.03 . wind speed at hub elevation Vhub , as given in [25], is used to
find the wind turbine power coefficient. Given Vref at href and
the surface roughness length of the site z 0 , Vhub can be easily
obtained by using logarithmic law [25]. Figures 6 and 7 show
the renewable power as well as the power deficit and surplus
(Step 3).

Fig. 3. Site wind speed Vref at elevation href 3.5m

Fig. 6. Daily distribution of renewable power and the demand load

Fig. 4. Site solar irradiance

Fig. 7. Daily distribution of power deficit and surplus

Selecting Kaplan turbine ( see Fig. 8), using Equations 4

and 5 with peak load L peak 15kW , load margin of safety of
MOS L 5% , turbine full load efficiency KT ,max (Fig. 9),
0.962
generator efficiency KG 0.95 and hydraulic height H T 30m ,
the turbine nominal power and maximum flow rate are
calculated as PT ,nom 17kW and QT ,max 0.06m 3 / s respectively.
These values are well inside the Kaplan turbine working
envelop (see Fig. 8).
Fig. 5. Demand load
Using Equation 6 with an autonomy period of Ta 2hr and
Performance and size optimisation parameters are selected capacity margin of safety of MOSUR 5% , the capacity of the
as: Tcharge 5hr , Ta 2hr and MOS L MOS UR 5% (Step 1). upper and lower reservoirs are obtained as
The size of wind turbine and PV panel are selected as VUR VLR 470m3 .
2
RWT 5m and APV 80m (Step 2). In addition to these two Using Tcharge 5hr and a full load pump efficiency of 85%,
primary design variables, renewable power can be obtained Equations 8 through 10 suggest a nominal power of
only if a number of secondary parameters are also decided or

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:41:32 UTC from IEEE Xplore. Restrictions apply.
PP,nom 23kW which is the maximum available power surplus.
In both turbine and pump calculations, the head loss is assumed
to be 5% of the actual elevation difference. By selecting the
pump, based on the actual available power surplus and power
deficit, one can easily obtain the water flow to and from the
upper reservoir and therefore obtain the SOC of the upper
reservoir. The design candidate is a feasible solution if in all
cases the constraint SOCUR,min d SOCUR is satisfied.

Fig. 11. Upper reservoir SOS

V. SUMMARY AND CONCLUSION

Stochastic nature of renewable resources makes storage and
backup systems key components for standalone hybrid
renewable energy systems, which need to be designed and
sized optimally alongside other components of the system to
guarantee power quality, system reliability and cost
effectiveness. In this paper, a sizing algorithm for the pumped
hydro energy storage is introduced and through a case study is
Fig. 8. Typical envelopes for various types of turbines
described.

REFERENCES
[1] Luo, X., et al., Overview of current development in electrical energy
storage technologies and the application potential in power system
operation. Applied Energy, 2015. 137: p. 511-536.
[2] Chen, H., et al., Progress in electrical energy storage system: A critical
review. Progress in Natural Science, 2009. 19(3): p. 291-312.
[3] Ma, T., et al., Technical feasibility study on a standalone hybrid solar-
wind system with pumped hydro storage for a remote island in Hong
Kong. Renewable Energy, 2014. 69: p. 7-15.
[4] Rehman, S., L.M. Al-Hadhrami, and M.M. Alam, Pumped hydro energy
storage system: A technological review. Renewable and Sustainable
Energy Reviews, 2015. 44: p. 586-598.
[5] Pumped-hydro energy storage: potential for transformation from single
dams. JRC IET scientific and technical report.EUR 25239 EN. Institute
Fig. 9. A typical efficiency curve for Kaplan turbine for Energy and Transport. Published on 24th February 2012.
[6] Hino T, L.A., Pumped storage hydropower developments. Compre
Renewable Energy, 2012. ;6:405–34.
[7] Mitteregger, A. and G. Penninger, Austrian pumped storage power
stations supply peak demands. World Pumps, 2008. 2008(500): p. 16-21.
[8] Nazari, M.E., M.M. Ardehali, and S. Jafari, Pumped-storage unit
commitment with considerations for energy demand, economics, and
environmental constraints. Energy, 2010. 35(10): p. 4092-4101.
[9] Kear, G. and R. Chapman, ‘Reserving judgement’: Perceptions of
pumped hydro and utility-scale batteries for electricity storage and
reserve generation in New Zealand. Renewable Energy, 2013. 57: p.
249-261.
[10] Ma, T., H. Yang, and L. Lu, Feasibility study and economic analysis of
pumped hydro storage and battery storage for a renewable energy
powered island. Energy Conversion and Management, 2014. 79: p. 387-
397.
[11] Connolly, D., et al., The technical and economic implications of
Fig. 10. Turbine/pump power during power deficit/surplus integrating fluctuating renewable energy using energy storage.
Renewable Energy, 2012. 43: p. 47-60.
[12] Aihara R, Y.A., Nomiyama F, Kosugi N., Impact of operational
scheduling of pumped storage power plant considering excess energy
and reduction of fuel cost on power supply reliability in a power system
with a large penetration of photovoltaic generations In: power System

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:41:32 UTC from IEEE Xplore. Restrictions apply.
 Technology (POWERCON), 2010 International Conference on;, 2010: [20] Junhui Z, G.K., Caisheng W, Gene L, Chih-Ping Y, A Hybrid
p. 1-6. Electric/Hydro Storage Solution for Standalone Photovoltaic
[13] Aihara R, Y.A., Nomiyama F, Kosugi N., Optimal operation scheduling Applications in Remote Areas. In: Power and energy society general
of pumped storage hydro power plant in power system with a large meeting IEEE;, 2012: p. 1-6.
penetration of photovoltaic generation using genetic algorithm In: [21] Ma, T., et al., Pumped storage-based standalone photovoltaic power
PowerTech, 2011 IEEE Trondheim; 2011., 2011: p. 1-8. generation system: Modeling and techno-economic optimization.
[14] Li, Y., et al., The research of wind-light complementary based on Applied Energy, 2015. 137: p. 649-659.
pumped storage power system, in Advanced Materials Research. 2012. [22] Ruisheng Li, B.W., Xianwei Li, Fengquan Zhou, Yanbin Li, Design of
p. 1132-1136. Wind-solar and Pumped-storage Hybrid Power Supply System. In: 2010
[15] Margeta J, G.Z., Theoretical settings of photovoltaic-hydro energy 3rd IEEE international conference on computer science and information
system for sustainable energy production. Sol Energy 2012. ;86:972–82. technology (ICCSIT);, 2010: p. 402-405.
[16] Margeta, J. and Z. Glasnovic, Feasibility of the green energy production [23] Ren Yan, L.Y., Simulation and optimization of hybrid wind-solar-
by hybrid solar + hydro power system in Europe and similar climate pumped-storage power system. In: 2011 International conference on
areas. Renewable and Sustainable Energy Reviews, 2010. 14(6): p. electric information and control engineering (ICEICE);, 2011: p. 729-
1580-1590. 733.
[17] Tao M, C.L., Lin L, Hongxing Y., Study on stand-alone hybrid solar [24] Manolakos D, P.G., Papantonis D, Kyritsis S, A simulation-optimisation
pumped storage systems. In: 11th International conference on programme for designing hybrid energy systems for supplying
sustainable energy technologies (Set-2012), 2012. Vancouver, Canada; electricity and fresh water through desalination to remote areas: Case
September 2–5 study: the Merssini village, Donoussa island, Aegean Sea, Greece.
Energy, 2001. ;26:679–704.
[18] Pirooz Javanbakht, S.M., M. Godoy Simões, Transient Performance
Analysis of a Small-Scale PV-PHS Power Plant Fed by a SVPWM [25] Maheri, A., A critical evaluation of deterministic methods in size
Drive Applied for a Distribution System. Energy conversion congress optimisation of reliable and cost effective standalone hybrid renewable
and exposition (ECCE). IEEE;, 2013: p. 4532–9. energy systems. Reliability Engineering & System Safety, 2014. 130: p.
159-174.
[19] Manolakos, D., et al., A stand-alone photovoltaic power system for
remote villages using pumped water energy storage. Energy, 2004. [26] Maheri, A., Multi-objective design optimisation of standalone hybrid
29(1): p. 57-69. wind-PV-diesel systems under uncertainties. Renewable Energy, 2014.
66: p. 650-661.

Authorized licensed use limited to: Anelis Plus consortium. Downloaded on January 11,2021 at 10:41:32 UTC from IEEE Xplore. Restrictions apply.