Spectral Decomposition of Demand-Side Flexibility for Reliable Ancillary Services
1. Spectral Decomposition of Demand-Side Flexibility
for Reliable Ancillary Services
International Conference on System Sciences
Kauai — January 5-8, 2015
Sean P. Meyn
Prabir Barooah and Ana Buˇsi´c
Yue Chen, Jordan Ehren, He Hao
Electrical and Computer Engineering & Florida Institute for Sustainable Energy
University of Florida
Thanks to NSF, AFOSR, and DOE / TCIPG
2. Demand-Side Flexibility for Reliable Ancillary Services
Outline
1 What Do We Want?
2 Spectral Decomposition
3 Buildings as Batteries
4 Intelligent Pools in Florida
5 Conclusions
3. Contour Map: Real Time Market - Locational Marginal Pricing Help?
$10/MWh
Nirvana @ ERCOT
What Do We Want From DR?
4. What Do We Want?
We want: Responsive Regulation
Plot of wind generation output in BPA — First week of 2015
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
GW (t) = Wind generation in BPA, Jan 2015
Scary ramps
1 / 15
5. What Do We Want?
We want: Responsive Regulation
Demand Dispatch the Answer?
A partial list of the needs of the grid operator, and the consumer:
2 / 15
6. What Do We Want?
We want: Responsive Regulation
Demand Dispatch the Answer?
A partial list of the needs of the grid operator, and the consumer:
High quality AS? (Ancillary Service)
Does the deviation in power consumption accurately track the desired
deviation target?
2 / 15
7. What Do We Want?
We want: Responsive Regulation
Demand Dispatch the Answer?
A partial list of the needs of the grid operator, and the consumer:
High quality AS? (Ancillary Service)
-15
-10
-5
0
5
10
15
20
25
30
35
6:00 7:00 8:00 9:00 6:00 7:00 8:00 9:00
Regulation(MW)
Gen 'A' Actual Regulation
Gen 'A' Requested Regulation
-20
-15
-10
-5
0
5
10
15
20
:
Regulation(MW)
Gen 'B' Actual Regulation
Gen 'B' Requested Regulation
Fig. 10. Coal-fired generators do not follow regulation signals precisely....
Some do better than others
Reg service from generators is not perfect
2 / 15
8. What Do We Want?
We want: Responsive Regulation
Demand Dispatch the Answer?
A partial list of the needs of the grid operator, and the consumer:
High quality AS?
Reliable?
Will AS be available each day?
It may vary with time, but capacity must be predictable.
2 / 15
9. What Do We Want?
We want: Responsive Regulation
Demand Dispatch the Answer?
A partial list of the needs of the grid operator, and the consumer:
High quality AS?
Reliable?
Cost effective?
This includes installation cost, communication cost, maintenance,
and environmental.
2 / 15
10. What Do We Want?
We want: Responsive Regulation
Demand Dispatch the Answer?
A partial list of the needs of the grid operator, and the consumer:
High quality AS?
Reliable?
Cost effective?
Is the incentive to the consumer reliable?
If a consumer receives a $50 payment for one month, and only $1 the
next, will there be an explanation that is clear to the consumer?
2 / 15
11. What Do We Want?
We want: Responsive Regulation
Demand Dispatch the Answer?
A partial list of the needs of the grid operator, and the consumer:
High quality AS?
Reliable?
Cost effective?
Is the incentive to the consumer reliable?
Customer QoS constraints satisfied?
The pool must be clean, fresh fish stays cold, building climate is
subject to strict bounds, farm irrigation is subject to strict constraints,
data centers require sufficient power to perform their tasks.
2 / 15
12. What Do We Want?
We want: Responsive Regulation
Demand Dispatch the Answer?
A partial list of the needs of the grid operator, and the consumer:
High quality AS?
Reliable?
Cost effective?
Is the incentive to the consumer reliable?
Customer QoS constraints satisfied?
Demand Dispatch can do all of this! (by design)
2 / 15
13. Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
Gr(t)
Gr = G1 + G2 + G3
G1
Spectral Decomposition
14. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
GW (t) = Wind generation in BPA, Jan 2015
3 / 15
15. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
GW (t) = Wind generation in BPA, Jan 2015
Scary ramps
3 / 15
16. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
G
Goal:
W (t) = Wind generation in BPA, Jan 2015
Scary ramps
GW (t) + Gr(t) ≡ 4GW
3 / 15
17. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Sca
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
G
Goal:
W (t) = Wind generation in BPA, Jan 2015
Scary
Scary
Scary
Scary
Scary
GW (t) + Gr(t) ≡ 4GW
Can the product be obtained
from generation?
Gr(t)
3 / 15
18. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
Goal: GW (t) + Gr(t) ≡ 4GW
Can the product be obtained
from generation?
Gr(t)
Gr(t)
Scary
3 / 15
19. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
Proposal:
Gr(t)
Scary
Gr = G1 + G2 + G3
3 / 15
20. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
Proposal:
Gr(t)
Scary
Gr = G1 + G2 + G3
H1 : Low pass
H2 : Band pass
H3 : High pass
Gi = HiGr
3 / 15
21. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
Proposal:
Similar to PJM’s RegA/D
Gr(t)
Scary
Gr = G1 + G2 + G3
H1 : Low pass
H2 : Band pass
H3 : High pass
Gi = HiGr
3 / 15
22. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
Gr(t)
Gr = G1 + G2 + G3
G1
3 / 15
23. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
Gr(t)
Gr = G1 + G2 + G3
G1
Not at all scary!
3 / 15
24. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
Gr(t)
Gr = G1 + G2 + G3
G1
G2
3 / 15
25. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
Gr(t)
Gr = G1 + G2 + G3
G1
G2
G3
3 / 15
26. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
Gr(t)
Nothing
Scary
Here!
G1
G2
G3
3 / 15
27. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
Gr(t)
G1
G2
G
Gas-turbine/hydro
3
3 / 15
28. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
Gr(t)
G1
G2
G
Gas-turbine/hydro
Chillers/pool pumps
3
3 / 15
29. Spectral Decomposition
Taming Scary Ramps – An Example from BPA
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06
GW
0
1
2
3
4
Gr(t)
G1
G2
G
Gas-turbine/hydro
Chillers/pool pumps
Interior commercial HVAC3
3 / 15
30. Spectral Decomposition
Control Architecture
Frequency Decomposition
Power GridControl Flywheels
Batteries
Coal
GasTurbine
BP
BP
BP C
BP
BP
Voltage
Frequency
Phase
HC
Σ
−
Actuator feedback loop
A
LOAD
Today: PJM decomposes regulation signal based on bandwidth,
R = RegA + RegD
4 / 15
31. Spectral Decomposition
Control Architecture
Frequency Decomposition
Power GridControl Flywheels
Batteries
Coal
GasTurbine
BP
BP
BP C
BP
BP
Voltage
Frequency
Phase
HC
Σ
−
Actuator feedback loop
A
LOAD
Today: PJM decomposes regulation signal based on bandwidth,
R = RegA + RegD
Proposal: Each class of DR (and other) resources will have its own
bandwidth of service, based on QoS constraints and costs.
4 / 15
36. Buildings as Batteries
Buildings as Batteries
HVAC flexibility to provide additional ancillary service
◦ Buildings consume 70% of electricity in the US
HVAC contributes to 40% of the consumption.
6 / 15
37. Buildings as Batteries
Buildings as Batteries
HVAC flexibility to provide additional ancillary service
◦ Buildings consume 70% of electricity in the US
HVAC contributes to 40% of the consumption.
◦ Buildings have large thermal capacity
6 / 15
38. Buildings as Batteries
Buildings as Batteries
HVAC flexibility to provide additional ancillary service
◦ Buildings consume 70% of electricity in the US
HVAC contributes to 40% of the consumption.
◦ Buildings have large thermal capacity
◦ Modern buildings have fast-responding equipment:
VFDs (variable frequency drive)
6 / 15
39. Buildings as Batteries
Buildings as Batteries
HVAC flexibility to provide additional ancillary service
◦ Buildings consume 70% of electricity in the US
HVAC contributes to 40% of the consumption.
◦ Buildings have large thermal capacity
◦ Modern buildings have fast-responding equipment:
VFDs (variable frequency drive)
6 / 15
41. Buildings as Batteries
Buildings as Batteries
Tracking RegD at Pugh Hall — ignore the measurement noise
In one sentence: Ramp up and down power consumption, just 10%, to
track regulation signal.
7 / 15
42. Buildings as Batteries
Buildings as Batteries
Tracking RegD at Pugh Hall — ignore the measurement noise
In one sentence: Ramp up and down power consumption, just 10%, to
track regulation signal.
Result:
7 / 15
43. Buildings as Batteries
Buildings as Batteries
Tracking RegD at Pugh Hall — ignore the measurement noise
In one sentence: Ramp up and down power consumption, just 10%, to
track regulation signal.
Result:
PJM RegD Measured
Power
(kW)
0
1
-1
0 10 20 30 40
Time (minute)
7 / 15
44. Buildings as Batteries
Pugh Hall @ UF
How much?
−4
−2
0
2
4
−10
0
10
0 500 1000 1500 2000 2500 3000 3500
−0.2
0
0.2
Time (s)
Fan Power Deviation (kW )
Regulation Signal (kW )
Input (%)
Temperature Deviation (◦
C)
One AHU fan with 25 kW motor:
> 3 kW of regulation reserve
Pugh Hall (40k sq ft, 3 AHUs):
> 10 kW
Indoor air quality is not affected
8 / 15
45. Buildings as Batteries
Pugh Hall @ UF
How much?
−4
−2
0
2
4
−10
0
10
0 500 1000 1500 2000 2500 3000 3500
−0.2
0
0.2
Time (s)
Fan Power Deviation (kW )
Regulation Signal (kW )
Input (%)
Temperature Deviation (◦
C)
One AHU fan with 25 kW motor:
> 3 kW of regulation reserve
Pugh Hall (40k sq ft, 3 AHUs):
> 10 kW
Indoor air quality is not affected
100 buildings:
> 1 MW
8 / 15
46. Buildings as Batteries
Pugh Hall @ UF
How much?
−4
−2
0
2
4
−10
0
10
0 500 1000 1500 2000 2500 3000 3500
−0.2
0
0.2
Time (s)
Fan Power Deviation (kW )
Regulation Signal (kW )
Input (%)
Temperature Deviation (◦
C)
One AHU fan with 25 kW motor:
> 3 kW of regulation reserve
Pugh Hall (40k sq ft, 3 AHUs):
> 10 kW
Indoor air quality is not affected
100 buildings:
> 1 MW
That’s just using the fans!
8 / 15
48. Buildings as Batteries
Buildings as Batteries
What do you think?
Questions:
Capacity? Tens of Gigawatts from commercial buildings in the US
9 / 15
49. Buildings as Batteries
Buildings as Batteries
What do you think?
Questions:
Capacity? Tens of Gigawatts from commercial buildings in the US
Can we obtain a resource as effective as today’s spinning reserves?
9 / 15
50. Buildings as Batteries
Buildings as Batteries
What do you think?
Questions:
Capacity? Tens of Gigawatts from commercial buildings in the US
Can we obtain a resource as effective as today’s spinning reserves?
Yes!! Buildings are well-suited to balancing reserves,
and other high-frequency regulation resources
9 / 15
51. Buildings as Batteries
Buildings as Batteries
What do you think?
Questions:
Capacity? Tens of Gigawatts from commercial buildings in the US
Can we obtain a resource as effective as today’s spinning reserves?
Yes!! Buildings are well-suited to balancing reserves,
and other high-frequency regulation resources
much better than any generator
9 / 15
52. Buildings as Batteries
Buildings as Batteries
What do you think?
Questions:
Capacity? Tens of Gigawatts from commercial buildings in the US
Can we obtain a resource as effective as today’s spinning reserves?
Yes!! Buildings are well-suited to balancing reserves,
and other high-frequency regulation resources
much better than any generator
How to compute baselines?
9 / 15
53. Buildings as Batteries
Buildings as Batteries
What do you think?
Questions:
Capacity? Tens of Gigawatts from commercial buildings in the US
Can we obtain a resource as effective as today’s spinning reserves?
Yes!! Buildings are well-suited to balancing reserves,
and other high-frequency regulation resources
much better than any generator
How to compute baselines?
Who cares? The utility or aggregator is responsible for the equipment –
the consumer cannot ‘play games’ in a real time market!
9 / 15
54. Buildings as Batteries
Buildings as Batteries
What do you think?
Questions:
Capacity? Tens of Gigawatts from commercial buildings in the US
Can we obtain a resource as effective as today’s spinning reserves?
Yes!! Buildings are well-suited to balancing reserves,
and other high-frequency regulation resources
much better than any generator
How to compute baselines?
Who cares? The utility or aggregator is responsible for the equipment –
the consumer cannot ‘play games’ in a real time market!
What open issues do you see?
9 / 15
56. Intelligent Pools in Florida
Example: One Million Pools in Florida
How Pools Can Help Regulate The Grid
1,5KW 400V
Needs of a single pool
Filtration system circulates and cleans: Average pool pump uses
1.3kW and runs 6-12 hours per day, 7 days per week
10 / 15
57. Intelligent Pools in Florida
Example: One Million Pools in Florida
How Pools Can Help Regulate The Grid
1,5KW 400V
Needs of a single pool
Filtration system circulates and cleans: Average pool pump uses
1.3kW and runs 6-12 hours per day, 7 days per week
Pool owners are oblivious, until they see frogs and algae
10 / 15
58. Intelligent Pools in Florida
Example: One Million Pools in Florida
How Pools Can Help Regulate The Grid
1,5KW 400V
Needs of a single pool
Filtration system circulates and cleans: Average pool pump uses
1.3kW and runs 6-12 hours per day, 7 days per week
Pool owners are oblivious, until they see frogs and algae
Pool owners do not trust anyone: Privacy is a big concern
10 / 15
59. Intelligent Pools in Florida
Example: One Million Pools in Florida
How Pools Can Help Regulate The Grid
1,5KW 400V
Needs of a single pool
Filtration system circulates and cleans: Average pool pump uses
1.3kW and runs 6-12 hours per day, 7 days per week
Pool owners are oblivious, until they see frogs and algae
Pool owners do not trust anyone: Privacy is a big concern
Randomized control strategy is needed.
10 / 15
65. Conclusions
Conclusions
What we want: Responsive Regulation and Happy Consumers
High quality Ancillary Service
Reliable on many time scales
Cost effective? marginal cost zero?
Is the incentive to the consumer reliable?
Customer QoS constraints satisfied? damn straight!
12 / 15
66. Conclusions
Conclusions
What we want: Responsive Regulation and Happy Consumers
High quality Ancillary Service
Reliable on many time scales
Cost effective? marginal cost zero?
Is the incentive to the consumer reliable?
Customer QoS constraints satisfied? damn straight!
Demand Dispatch can do all of this!
12 / 15
67. Conclusions
Conclusions
What we want: Responsive Regulation and Happy Consumers
Is the incentive to the consumer reliable?
A caveat: There is the “tragedy of the commons”
– perhaps not so in Hawaii?
12 / 15
68. Conclusions
Conclusions
What we want: Responsive Regulation and Happy Consumers
Is the incentive to the consumer reliable?
A caveat: There is the “tragedy of the commons”
– perhaps not so in Hawaii?
Need for Research in Engineering and Economics
12 / 15
70. Conclusions
Control Techniques
FOR
Complex Networks
Sean Meyn
Pre-publication version for on-line viewing. Monograph available for purchase at your favorite retailer
More information available at http://www.cambridge.org/us/catalogue/catalogue.asp?isbn=9780521884419
Markov Chains
and
Stochastic Stability
S. P. Meyn and R. L. Tweedie
August 2008 Pre-publication version for on-line viewing. Monograph to appear Februrary 2009
π(f)<∞
∆V (x) ≤ −f(x) + bIC(x)
Pn
(x, · ) − π f → 0
sup
C
Ex[SτC(f)]<∞
References
14 / 15
71. Conclusions
References: Demand Response
A. Brooks, E. Lu, D. Reicher, C. Spirakis, and B. Weihl. Demand dispatch. IEEE Power
and Energy Magazine, 8(3):20–29, May 2010.
H. Hao, T. Middelkoop, P. Barooah, and S. Meyn. How demand response from commercial
buildings will provide the regulation needs of the grid. In 50th Allerton Conference on
Communication, Control, and Computing, pages 1908–1913, 2012.
H. Hao, Y. Lin, A. Kowli, P. Barooah, and S. Meyn. Ancillary service to the grid through
control of fans in commercial building HVAC systems. IEEE Trans. on Smart Grid,
5(4):2066–2074, July 2014.
S. Meyn, P. Barooah, A. Buˇsi´c, Y. Chen, and J. Ehren. Ancillary service to the grid using
intelligent deferrable loads. ArXiv e-prints: arXiv:1402.4600 and to appear, IEEE Trans. on
Auto. Control, 2014.
D. Callaway and I. Hiskens, Achieving controllability of electric loads. Proceedings of the
IEEE, vol. 99, no. 1, pp. 184–199, 2011.
P. Xu, P. Haves, M. Piette, and J. Braun, Peak demand reduction from pre-cooling with
zone temperature reset in an office building, 2004.
D. Watson, S. Kiliccote, N. Motegi, and M. Piette, Strategies for demand response in
commercial buildings. In Proceedings of the 2006 ACEEE Summer Study on Energy
Efficiency in Buildings, August 2006.
15 / 15