1. Environmental Change Institute
Volatile wind and flexible demand:
a balancing act?
Philipp Grünewald
5th Smart Grids & Cleanpower Conference
5 June, Cambridge
www.cir-strategy.com/events

2. What to expect
§ Something old, (Storage)
§ something new, (DSR)
§ something borrowed, (Flexibility)
§ something blue, (Peak demand)
§ and a silver sixpence in her shoe (Data)
Page 2

3. The displacement of conventional plant
Page 3Source: Transition Pathways ProjectSource: Jonathan Radcliffe (ERP) based on DECC 2050 pathway
1) decarbonise
2) electrify

4. 1: Build flexible generation and curtail excess supply
2: Build expansive networks and enable spatial arbitrage
3: Build physical electricity storage capacity
4: Enable demand side flexibility to respond to supply
Sources
of
flexibility
Flexibility Responsive
Network
Storage
Generation
Consumption
Flexible
Additional
Page 4

5. [Electricity storage] promises savings on UK energy spend of
up to £10bn a year by 2050 as extra capacity for peak load is
less necessary.
George Osborne, 9 November 2012
Up to 30 new gas power stations will be
needed by 2030
George Osborne, 5 December 2012
Page 5

6. The cost of meeting peak demand
Page 6
Price duration curve (1 year)
3.1. Storage value increase with wind deployment
Intermittent generation increases the volatility of electricity prices, which
value proposition for storage. Figure 4 shows the increasing gross value of sto
base case scenario with increasing amounts of wind.
4 8 16 32 64
38.6
62.5
116.6
125.1
Wind deployment [GW]
Grossstoragevalue[£/kW/year]
2 GW
10 GW
Figure 4: Increase in gross value of storage with increasing wind deployme
2 and 10 GW storage with 6 hour duration. Based on Grassroots 2030 scenar
At present levels of wind the value is not sufficient to stimulate further
and even at 16 GW of wind values do not increase substantially, yet. At 32 G
deployment, the gross value begins to exceed £100 per kW per year, which is
level of present technology options. With further expansion of wind capacit
2
increases the value of storage
storage
storage

7. SOURCES OF FLEXIBILITY
§ (Flexible) generation
§ (Smart) networks
§ (Responsive) demand
Page |
Figure 42: Average and marginal values for bulk and distributed storage in 2030 across different scenarios. T
marginal value is stable for most scenarios with the exception of flexible demand, which results in significant
reduced levels of deployment. (Results shown for 6 hours storage duration; for 24 hour and 48 hour results se
the Appendix)
The sensitivity of the value of storage to the presence of alternative solutions is shown
Figure 42. All cases originate from the base case scenario, which is in turn subjected
specific changes in the level of interconnection with neighbouring energy systems, flexibili
in generation, and the flexibility from the demand side. The average values again show th
all scenarios with competing options reduce the average value of storage compared to t
baseline Grassroots case.
The effect on the marginal values is shown in the lower half of Figure 42. These show ho
much storage would be invested in at a given cost within an economically efficient system
The marginal values differ only slightly between the different scenarios, with the notab
exception of flexible demand.
Page | 71
the Appendix)
The sensitivity of the value of storage to the presence of alternative solutions is shown in
Figure 42. All cases originate from the base case scenario, which is in turn subjected to
specific changes in the level of interconnection with neighbouring energy systems, flexibility
in generation, and the flexibility from the demand side. The average values again show that
all scenarios with competing options reduce the average value of storage compared to the
baseline Grassroots case.
The effect on the marginal values is shown in the lower half of Figure 42. These show how
much storage would be invested in at a given cost within an economically efficient system.
The marginal values differ only slightly between the different scenarios, with the notable
exception of flexible demand.
Page | 71
Figure 42: Average and marginal values for bulk and distributed storage in 2030 across different scenarios. The
marginal value is stable for most scenarios with the exception of flexible demand, which results in significantly
reduced levels of deployment. (Results shown for 6 hours storage duration; for 24 hour and 48 hour results see
the Appendix)
The sensitivity of the value of storage to the presence of alternative solutions is shown in
Figure 42. All cases originate from the base case scenario, which is in turn subjected to
specific changes in the level of interconnection with neighbouring energy systems, flexibility
in generation, and the flexibility from the demand side. The average values again show that
all scenarios with competing options reduce the average value of storage compared to the
baseline Grassroots case.
The effect on the marginal values is shown in the lower half of Figure 42. These show how
much storage would be invested in at a given cost within an economically efficient system.
The marginal values differ only slightly between the different scenarios, with the notable
exception of flexible demand.
Page | 71
reduced levels of deployment. (Results shown for 6 hours storage duration; for 24 hour and 48 hour results see
the Appendix)
The sensitivity of the value of storage to the presence of alternative solutions is shown in
Figure 42. All cases originate from the base case scenario, which is in turn subjected to
specific changes in the level of interconnection with neighbouring energy systems, flexibility
in generation, and the flexibility from the demand side. The average values again show that
all scenarios with competing options reduce the average value of storage compared to the
baseline Grassroots case.
The effect on the marginal values is shown in the lower half of Figure 42. These show how
much storage would be invested in at a given cost within an economically efficient system.
The marginal values differ only slightly between the different scenarios, with the notable
exception of flexible demand.
Page | 71
e sensitivity of the value of storage to the presence of alternative solutions is shown in
gure 42. All cases originate from the base case scenario, which is in turn subjected to
ecific changes in the level of interconnection with neighbouring energy systems, flexibility
generation, and the flexibility from the demand side. The average values again show that
scenarios with competing options reduce the average value of storage compared to the
seline Grassroots case.
e effect on the marginal values is shown in the lower half of Figure 42. These show how
uch storage would be invested in at a given cost within an economically efficient system.
e marginal values differ only slightly between the different scenarios, with the notable
ception of flexible demand.
Marginal values: Strbac et al., Strategic Assessment of the Role and Value of
Energy Storage Systems in the UK Low Carbon Energy Future. The Carbon Trust
reduced value for different DSR configurations is shown in Figure 5.19 on p
response, even when limited to a maximum of four hours, as shown here,
to significantly reduce the value of storage.
If the DSR capacity was 4 GW and responded to prices above £50/MW
of two-hour storage could reduce by over 50%. For longer storage durat
at this price level is lower, as shown for the example of six-hour storag
Figure 5.19 on page 124. This could suggest that in some scenarios for DS
adopt more long-duration roles, than those simulations here without any D
τ=2 hours τ=6 hours
£400/MWh
£50/MWh
πDSR
1 2 4 1 2 4 PDSR [GW]
51.4%
39.6%
20.1%
18.0%
30.1%
42.4%
Reductioningrossvalueofstorage[%]
Figure 5.19: Reduction in the value of storage as a result of demand
At a DSR resource of 4 GW and a response price of £50/MWh the gr
of storage with a 2 hour duration can half. Longer storage durations a
less affected. Example for 10 GW storage in base case scenario.
DSR
capacity
[GW]
Storage value reduction from DSR
Page 7

8. Demand aggregation in the UK
§ Collaboration with Kiwi Power
§ Access to data for over 500 sites
§ Half-hourly profile data (68 sites)
§ Recorded response events (266 sites)
§ Analysis of load profiles and load characteristics
§ Distinct profiles demand a sector by sector approach
§ Some sectors can not be generalised based on their profiles (e.g.
manufacturing)
Page 8

9. ‘Demand response’ from stand-by generation
(Telecoms sector)
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
Figure 4: Stand-by generation responses to demand response requests
Sample of 97 sites.
10
Page 9
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
sites and hotels do engage in demand
within agreed bounds.
Figure 6 shows significant load red
the distribution of responses to a loa
at around 25% load reduction, relativ
very few instances show little or no r
Contrasting the level of response
response, shows that the former contr
reductions. Stand-by generation ma
effectiveness and reliability. In the ne
response resource from demand turn

10. ‘Demand response’ from
Page 10
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
sites and hotels do engage in demand
within agreed bounds.
Figure 6 shows significant load red
the distribution of responses to a loa
at around 25% load reduction, relativ
very few instances show little or no r
Contrasting the level of response
response, shows that the former contr
reductions. Stand-by generation ma
effectiveness and reliability. In the ne
response resource from demand turn
stand-by generation
(Telecoms)
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
Load reduction
-100% -75% -50% -25%
Figure 6: Distribution of load reduction (relative to baseline) during demand response trial
in the hotel sector.
12
Load turn down
(Hotels)

11. Untapped potential?
§ Presently demand response plays a minor
role in response provision.
§ Although policy makers desire a “level
playing field” for all options, generation
based solutions have shaped the existing
framework.
§ Could the demand response contribution be
improved through regulatory changes?
Page 11Based on: National Grid.

12. Short
Term
Opera7ng
Reserve
(STOR)
Page 12
• “a
service
for
the
provision
of
addi7onal
ac7ve
power
from
genera7on
and/or
demand
reduc7on”
• 3MW
or
more
of
genera7on
or
steady
demand
Deliver
within
4
hours
from
instruc7on
• Provide
for
at
least
2
hours
• 20
hours
recovery
• Ability
to
provide
STOR
at
least
3
7mes
a
week
Based on: National Grid. STOR Market Information Report

13. What if conditions were relaxed?
2m
20m
4h
24h
0
0.5
1
1.5
2
2.5
3
10h
4h
2h
15/20m
Longer response time
à Time to “preload”
à Greater capacity
Longer response duration
à Rising/falling warehouse temperature
à Reduced capacity
Capacityresourcefactor
(STOR=1)
Response capacity availability from UK warehouses
(illustrative example)
STOR
CM? Possibly worth
between
0.4 and 1 GW

14. Plato’s cave
Page 14

15. Plato’s cave
Page 15
Researchers
Consumers

16. Page 16
Conclusions
• Significant technical and commercial
potential for various flexibility options
• Current regulatory structures favour
generation based solutions
• A better understanding of demand
response potential could be gained
through a combination of empirical
studies and detailed analysis

17. Page 17
Thank you.