Pumped storage in systems with very high wind penetration
y g
p
Aidan Tuohyy
SEI / ERC Storage Seminar
N Nov 2009
Overview

Introduction
 Why storage is considered beneficial


Test System/Model Used
Results
 When, if ever does pumped storage become a good option on a thermal/wind system?
 Examination of system operation with and without storage
 Examination of cost effects
Pumped Storage with wind 

Wind is inherently variable and uncertain
Wi
d i i h
l i bl d i
Storage seen as good option – ‘make wind dispatchable’




Wi d t d h bl i t b d h Wind stored when blowing to be used when needed
d d
However, pumped storage has high capital costs, needs specific geology, and has associated efficiency losses
Examine system with and without pumped storage (
(including removing already existing) for increasing wind
g
g
y
g)
g
Storage used for system to reduce costs during the day



Not for profit, but to benefit system
Economic reasons – not specifically to help wind or maintain reliability Savings need to be large enough to justify additional capital costs
Pumped Storage today

A
Approx 100GW installed worldwide
GW i
ll d ld id





Mainly in places with good natural geology
Much of the available sites would have been used
Storage, in many systems, would be too expensive
Maybe with wind (or other variable RES) storage becomes Maybe with wind (or other variable RES), storage becomes attractive again
Flexible plant, quick to react and can store excess wind to use when needed


One of flexibility measures available Also Demand Side Resources, Interconnection, flexible plant
Model Used

Wilmar Planning Tool
 Originally used to study Nordic system
 Adapted for All Island 2020 Grid Study  Main functionality is in Scenario Tree Tool (STT) and y
Scheduling Model (SM)
 Outputs taken from STT used as inputs to SM
p
p
 Scenario trees, other inputs and outputs ‐ Access databases
 GAMS language used for scheduling model (Cplex solver))
WILMAR
• 2 main parts: • Scenario Tree Tool
•Scheduling Model
• Uses rolling planning to minimise expected costs (fuel carbon startup)
(fuel, carbon, startup) while meeting load under various constraints (min (
up time, start up time etc)
Test system

Applied to possible Irish System in 2020
 Portfolio (5) produced in Grid Study as base
(5) p
y
 9.6 GW peak / 3.5 GW min
 Interconnection to Britain –
1000MW
 British units grouped
g p
together, deterministic
Annual run, with outages
etc modelled
d ll d
• Run for 6, 9, 12 GW (34%, 51%, 68% of energy) wind installed
R f 6 9 12 GW (34% 51% 68% f
) i di
ll d
• 500MW / 10 hr storage replaces conventional plant
Adding wind to system
25.00%
Capacity Credit
20.00%
15.00%
10.00%
5.00%
0.00%
6
9
12
Installed Wind (GW)

Remove units to ensure reliability remains approximately the same
Operation of Storage
Curtailment
Penetration Level achieved
70%
65%
60%
55%
50%
45%
40%
35%
30%
6
7
8
9
10
11
Wind Installed (GW)
Maximum Amount
With Storage
g
Without Storage
g
12
Operation of interconnection
Change in net export
4.00%
2.00%
% of IRL dem
mand
0.00%
-2.00%
6
7
8
9

11
12
-4.00%
-6.00%
-8.00%
-10.00%
Wind Installed (GW)
Without Storage

10
With Storage
GB System cheaper –
GB S
t h
storage allows more import from GB t
ll
i
t f
GB – regardless of wind
Ireland net exporter at 11GW
CO2 Emissions 

Reduction compared to no storage case
Both GB and IRL emissions – increase in one area may cause decrease in another
Costs
Effect of wind uncertainty
15
Costs by region

Saving for Ireland outweighed by increase for GB until high enough wind level reached – would be reflected in high enough wind level reached interconnection payments
Justification of building storage



These system cost savings need to be enough to justify building storage for the system
Net Present Value of 20 yrs of savings need to be g
greater than difference in cost between storage g
and unit it replaces
Different parameters for discount rate lifetime Different parameters for discount rate, lifetime etc examined
Capital Costs justified


Discount Rate / Savings move line left/right
/
g
/ g
Capital cost / lifetime moves justification line up/down
Net Savings
500
Net savin
ngs (€m)
400
300
200
100
0
‐100
49
50
51
52
53
54
55
‐200
Penetration Level
Y axis = (NPV of Additional Capital Costs of storage vs CCGT (with 40 yr lifetime) ‐ €0.25m) + (NPV of Additional Fuel Costs, 20 yrs 5% DR) – (Saved Capital Costs of Wind Turbines – €1.3m)
56
When is storage a sensible option?

F hi For this particular system:
i l  Depends somewhat, but not very significantly, on lifetime, discount rates received etc
lif
ti
di
t t i d t
 If storage costs same as CCGT – 7GW
 If storage costs double, with same lifetime and high If t
t d bl ith lif ti d hi h DR – 11.5GW
 Answer will lie somewhere between 2
 Storage projects vary significantly depending on civil works
 If lifetime, DR and capital costs known, this shows level of wind necessary
Saving on installed wind
Amount of additional wind needed to get to desired penetration if no storage
1800
Ad
dditional M
MW
1600
1400
1200
1000
800
600
400
200
0
30
35
40
45
50
Penetration (%)
55
60
65
Fuel Costs
Fuel Costs by Penetration
13500
13400
13300
13200
With Storage
13100
Without Storage
Without Storage
13000
12900
12800
12700
30.00%
35.00%
40.00%
45.00%
50.00%
55.00%
60.00%
65.00%
Reliability Effects
Load not met
Hours nott met
30
25
20
15
6GW
10
12GW
5
0
0
2
4
6
8
10
12
Hours of energy in store

Daily optimization used – smaller store cannot be relied upon as much
Conclusions

Storage may start to be sensible option for system S
b ibl i f at high wind levels
 However, 50% is extremely high on a synchronous %i
l hi h
h
system (40% seen as very ambitious by 2020)
 Other options for flexibility may be better
 Demand Side resources, interconnection, better market operation, better forecasting
p
,
g
 Storage good for reducing curtailment in island system
 Storage may make profit at lower level of wind –
g
y
p
looked at system costs only
 Or reduce need for transmission expansion
Conclusions (2)

Uncertainty of wind makes the flexibility of storage more useful
 If perfect wind foresight assumed, storage not as good

Plant mix/Size of System/Wind Characteristics will all have effect
 Results here for island system with high penetration of gas, low interconnection, good but variable wind