A proposal to use real time pricing to manage the electrical grid as a step toward distributed...
-
Upload
david-haynes -
Category
Documents
-
view
213 -
download
0
Transcript of A proposal to use real time pricing to manage the electrical grid as a step toward distributed...
Procedia Computer Science 8 (2012) 370 – 375
1877-0509 © 2012 Published by Elsevier B.V.doi:10.1016/j.procs.2012.01.074
Available online at www.sciencedirect.com
ProcediaComputerScience Procedia Computer Science 00 (2012) 000–000
www.elsevier.com/locate/procedia
Available online at www.sciencedirect.com
New Challenges in Systems Engineering and Architecting Conference on Systems Engineering Research (CSER)
2012 – St. Louis, MO Cihan H. Dagli, Editor in Chief
Organized by Missouri University of Science and Technology
A proposal to use real time pricing to manage the electrical grid as a step toward distributed control
David Haynesab, Steven Cornsb aAclara, 945 Hornet Dr., Hazelwood MO, 63042, USA
bMissouri University of Science and Technology (MS&T), 205 Engr. Mgt., 600 W., 14th St., Rolla MO, 65409-0370, USA
Abstract
An alternative method of managing the electrical grid is presented. Rather than using a centralized control paradigm to balance generation against the load, a distributed approach is proposed. It is believed that this design improves the robustness of the grid by using an architecture that allows for the automatic formation of local micro grids. The self-organization of these micro-grids makes it possible to identify participants in an ad hoc fashion after a storm for purposes of disaster recovery. The micro-grid can automatically grow to encompass all participants, and join with other grids to encompass the entire region. © 2012 Published by Elsevier Ltd. Selection Keywords: Power Systems, Power Distribution Control, Power Distribution Economics, Power Generation Dispatch, Power Generation Reliability, Power Generation Scheduling, Power System Control, Power System Economics, Power System Stability.
1. Background
The electric distribution grid today primarily relies on centralized power generation. Network operators monitor the load and stability of the distribution network and carefully control the level of generation to match the load. Many regions expect the future to exhibit an increased use of distributed resources such as small wind turbines and roof-top solar panels. [1] New ways should be considered to coordinate generation, pay for power, and the infrastructure used to transport this power. Furthermore, communication must occur to invoke energy storage as needed. This paper discusses the issues in greater detail and proposes a new method inspired by natural systems [2, 3] to address them.
a Corresponding author. Tel: +1-314-895-6452 E-mail address: [email protected]
371David Haynesa and Steven Corns / Procedia Computer Science 8 (2012) 370 – 375
Haynes/ ProcediaComputer Science 00 (2012) 000–000
Fig. 1 – Usecase diagram showing the roles and relationships within the status quo
2. Proposal
It is proposed that a tariff be created which allows network operators to attempt to balance generation against the load using Real Time Pricing (RTP). What makes this proposal unique is:
1.) The use of analog information in the power sinewave to adjust the Real Time Price between announcements.
2.) The sharing of consumer preference information with the bid reconciliation function in order to eliminate much of the guesswork involved in forecasting the effects of a given price selection.
It is proposed that consumption and generation preference profiles be uploaded to a hierarchical domain structure for local bid reconciliation. A local price for energy would be computed with each domain serving as an aggregator for the members beneath it. The domain itself is empowered to buy or sell energy on behalf of the net aggregation of the needs of its membership. When energy is purchased, a corresponding percentage of ancillary services must also be purchased.
2.1. Guiding Principles
A number of guiding principles were developed to guide the development of self-organizing rule sets: Local problems should be solved locally.
uc Background
Generation Transmission Distribution
Generates electricity and sells it onto "the
grid."
Transports bulk power from generation sites to
distribution networks.
Joins indiv idual consumers together on a common distribution grid, and serv es the
collectiv e load with power transmitted from remote generating plants.
ISO/RTO
Manage network Local utility
Perform Contingency Analysis
Collect Rev enue
Perform Billing
Pay suppliers
LinesCoGenCo
Sell energy to consumers
«include»
«include»
«precedes»
372 David Haynesa and Steven Corns / Procedia Computer Science 8 (2012) 370 – 375
Haynes/ Procedia Computer Science 00 (2012) 000–000
The sale of energy should be allowed to freely operate according to the laws of supply and demand. ○ Energy is more valuable during a time of crisis shortage than when it is available in plenty. ○ Power with high quality is worth more than power with poor quality.
2.2. Rules
For every kWh of energy sought by a consumer, for every interval of time, the system must also secure 0.5 kWh of spinning reserve and 0.5 kWh of non-spinning reserve.
All generation is eligible to participate in frequency correction. There are no distance penalties involved in generators who participate in frequency correction. Generators will earn at the latest posted RTP rate times the inter-period frequency price-multiplier.
2.3. Assumptions
Some of the undergirding assumptions related to the principle components should be described to better understand this paradigm. It should be understood that like today’s system, it is proposed that a centralized authority continue to exist to read revenue meters, bill customers, and pay suppliers. To perform this service, as well as operate the grid, a ubiquitous, secure, fault-tolerant communication system exists. This communication system has a knowledge of the electrical assets and may very well communicate over the powerline it supports. This paper introduces “Energy Management Nodes” (EMNodes) which are deployed in the field alongside important assets. Upon installation, the EMNodes are configured with engineering data, connectivity data, and location data regarding the asset they protect.
Generators have a certain price at which it becomes interesting for them to spin-up and contribute power to the grid. Owners of generation will weigh operational and lifecycle costs to determine their bid price. These costs include asset financing costs, as well as the price and availability of fuel. We assume that generator outcomes may be modeled as a Boolean go/no-go decision. Large generators will continue to play an important role in the future of the grid. Large generators have considerable inertia and slow ramp rates, so they require advanced notice in order to get up to speed. Ancillary services will therefore continue to be important in future grids. Energy storage is considered to be behave like a “generator” when making power and like a “consumer” when using power.
Consumers want what they want when they want it. This leads to price elasticity. Consumer decision inputs are a function of the price of energy, the day of the week, the week of the year (season), the weather, consumer affluence, and personal preferences. Unlike generator decisions which are “Boolean”, consumer outcomes are more complex and may be modeled with Fuzzy rules.
Lines companies will maintain their lines with regular maintenance, and make improvements as warranted by the need as funds allow. It is assumed that the substation EMNode will make voltage (tap changer and volt/VAr) adjustments (at the sub and along lengthy feeders) as appropriate so that local voltages along a feeder are an accurate reflection of the energy availability of the local region.
When participants get what they want, they have money to do other things. They will use some of this money to maintain their equipment, invest in improved efficiencies, and make capital investments to address the problems at the top of their list of priorities.
2.4. Domains
The domains are illustrated in Fig. 2a, and EMNode functionality described in Fig. 2b.
2.4.1. Participant Domain Every EMNode is responsible to announce its intentions (forecasted net energy consumption or
production) to its superior EMNode. The EMNode at the participant level is responsible to announce its
373David Haynesa and Steven Corns / Procedia Computer Science 8 (2012) 370 – 375
Haynes/ ProcediaComputer Science 00 (2012) 000–000
intentions to the service transformer EMNode. If power and ancillary services are not found at a price the consumer is willing to pay for, the service contactor will open – disconnecting the participant from the grid for the interval. Similarly, if the participant has generation, and no customers are found for any of its services, the service contactor will be opened for the interval.
Fig. 2 – (a) Proposed Domains (in UML Notation) (b) Usecase functionality of Energy Management Node
2.4.1.1. Load Forecast The participant node is expected to announce a 24 hr. forecast for energy. This should be based on a
number of factors including: historical energy usage patterns, planned activities, weather forecast, and
class EMNode
ISO Domain
Utility Domain
Substation Domain
Feeder Domain
Serv ice Transformer
Domain
Participant Domain
Premises Area Network
Independent Power Producer
Energy Storage Facility
Industrial CustomerAgricultural
CustomerCommercial
Customer
Residential Customer
0..*
1
1..*
1
1..*
11..*
1
1..*
1
1..*
0..*
0..*1..*
0..*0..*1..*
1..* 0..* 0..*
0..*1..*
uc EMNode
EM Node
Monitor asset health
Perform engineering contingency analysis
Aggregate downstream bids
Communicate to other nodes (upstream,
downstream, and peer)
Compute downstream bid Log price messages
Participant EMNode
Forecast loadIdentify price preference
behav ior
374 David Haynesa and Steven Corns / Procedia Computer Science 8 (2012) 370 – 375
Haynes/ Procedia Computer Science 00 (2012) 000–000
current temperature.
2.4.1.2. Prioritization If the price of power is low, a consumer would likely run every appliance to their heart’s content. As
the price of energy rises, a consumer is forced to make tradeoffs between their physical and financial comfort. When a crisis prevails and energy is scarce, the cost of power would be expected to be high. This energy might come from backup emergency generators and be very limited. A homeowner might elect to run their refrigerator on such power, but not much more. Similar decisions are made in the hands of the power producer. For this paper we will assume that prices can be categorized into five levels as seen in Fig. 3a. This figure represents data generated by a survey supplied by one of the authors for his home.
Fig. 3 – (a) Example Homeowner Price Preferences (b) Price Multiplier
2.4.2. Service Transformer Domain and Domains Above Every Energy Management Node:
Protects assets immediately beneath it. Declares the mode of operation (e.g. normal or emergency) for its domain. Aggregates the loads and generation bids of the participants within its domain. Is empowered to act on behalf of the participants in its domain to buy and sell according to aggregated
needs. (This allows the EMNode to “go shopping” for energy as an agent representing their domain.) Performs an engineering contingency check to confirm the viability of the planned powerflows.
2.5. Formulaic Energy Pricing
Local prices are computed and broadcast (separately) for consumers and producers. Prices between broadcasts (sub intervals) are computed using equations (1) (2) and (3) and plotted in Fig. 3b.
ConsumerSubIntervalPrice(t) = PublishedConsumerIntervalPrice(t) * AnalogPriceMultiplier(t); (1) ProducerSubIntervalPrice(t) = PublishedProducerIntervalPrice(t) * AnalogPriceMultiplier(t); (2)
������������������������ ����1
���� ���������������������� � 1����������������������1 � ��� � 1���
���� ��������������������� � ���1 �
������1
������ ���������������������� � 1��1�������������������� � ��� � 1��1
���� ��������������������� � ���� �
��; (3)
Emergency
CriticalPeak
Peak
Off‐Peak
Abu
ndant
0
500
1000
1500
2000
0 4 812
1620
Price Level
Energy (W
h)
TimeBucket
Home Summary
EmergencyCriticalPeakPeakOff‐PeakAbundant
0123456789
0.50
0.58
0.67
0.75
0.83
0.92
1.00
1.08
1.17
1.25
1.33
1.42
1.50
Multiplier
Per Unit Value
FreqMult VoltageMult
Product
375David Haynesa and Steven Corns / Procedia Computer Science 8 (2012) 370 – 375
Haynes/ ProcediaComputer Science 00 (2012) 000–000
Where “Vpu” is the per unit voltage measured at the endpoint, and “Fpu” is the per unit frequency.
3. Closing Comments
The proposal allows for the ac hoc creation of a micro grid with membership consisting of the interconnected survivors after a storm. This is a form of short-term self-organization in which viable network alternatives are automatically discovered and put in place prior to the arrival of repair crews. By pre-establishing prices, and performing engineering analysis on proposed powerflows, the network (small or large) is able to match consumers needs with the available generation. The novel inter-period price formulas allow the price to reflect the local situation better than other Real Time Pricing commonly in use. In a free market, such formulas should attract additional generation where it is needed and discourage it where it is not. These formulas eliminate the common fixed-rate tariff, and allow for regional differences in prices to exist. The removal of fixed prices can be somewhat painful for certain consumers, but it forces the real issues to be addressed. In the short term, dynamic prices allow demand to match supply. In the long term, it could lead to the appropriate build-out of infrastructure:
1.) Investments to take advantage of low energy costs where unusually low energy prices exist. 2.) Investments to combat high energy prices where high prices exist. 3.) Investments in energy storage in order to increase the value of non-dispatchable generation.
Under these rules, owners of non-dispatchable generation will likely find that they are unable to maximize the return on their investment without also making a corresponding investment in energy storage.
4. Future Development
Opportunities exist to leverage computational intelligence and/or expert systems at several levels in the proposed system. This could include from Load forecasting from the perspective of the EMNode (over their respective domains.) to anticipate the upcoming costs. We could also develop algorithms for the automation of network operation in which energy and ancillary services are secured through bid reconciliation. Several options exist for the optimization of pricing formulas [4, 5, 6], or a custom method could be tailored to this problem. This type of optimization could also be applied to the development of communication protocols, since it is quite possible that no single protocol will suffice to serve all domains [7].
5. References
[1] J. McDowall, “Integrating energy storage with wind power in weak electricity grids,” Journal of Power Sources, vol. 162, no. 2, p. 959, 22. 2006.
[2] J. J. Bartholdi III, D. D. Eisenstein, and Y. F. Lim, “Self-organizing logistics systems,” Annual Reviews in Control, vol. 34, no. 1, pp. 111-117, Apr. 2010.
[3] A. Adamatzke and J. Jones, “Road planning with slime mould: If Physarum built motorways it would route M6/M74 through Newcastle,” Elsevier, vol. In process, Dec. 2009.
[4] K. Shoa, “Smart Grid Dynamic Pricing: Behavior Change Easier Said than Done,” Smart-Grid.TMCNet, 09-Jun-2010.. [5] R. Sexton, N. Johnson, and A. Konakayama, “Consumer Response to Continuous-Display Electricity-Use Monitors in a time-
of-Use Pricing Experiment,” Journal of Consumer Research, vol. 14, Jun. 1987. [6] R. O’Neill, E. Fisher, B. Hobbs, and R. Baldick, “Towards a complete real-time electricity market design,” Springer, May.
2008. [7] D. D. Haynes, “A Case for Optimized Protocols in the Creation of a Smarter Grid,” IEEE Transactions on Power Delivery,
vol. 25, no. 3, pp. 1476-1482, Jul. 2010.