Demand Response and Ancillary Service Strategy using Dynamic Game Model for Aggregator Business Market in Smart Grids

 An aggregator acts as a mediator or broker to a group of customers to participate in demand responses (DRs) in an electricity market. Key green policies have encouraged green financial loans to construct distributed generations (DGs) as a virtual power plant. Hence, an innovative business market integrates load management and ancillary services into a demand and generation aggregator. This controllable entity can reduce demands with incentive and contractual controls and dispatch the ancillary power. This study proposes DR programs and ancillary services using the dynamic game model in an aggregator business market. This strategy impartially dispatches reserve energy sources with customers’ dynamic demands, such as onsite generations and storage energy. The aim is to achieve the demand reduction from the main grid, further increasing system flexibility to activate active-duty DGs. A small customer group is selected to demonstrate the effectiveness of the proposed model. Benefits for billing charges may arise. Index Terms Demand Response, Distributed Generation, Dynamic Game Model, Aggregator Business Market.


Introduction
Demand response (DR) is a strategy or program employed to make changes in customer usage and coordinates with power utilities or electricity market conditions, diving into direct load control (DLC) and indirect load control (ILC) in demand side management [1,2].In a DLC strategy, utilities directly send command signals for loads to change their operation conditions with the aim of shifting peak demand and matching power balance or providing the demand of using renewable energy and other distributed generations (DGs).An ILC strategy uses price-based or incentive payments such as time-of-use, real-time pricing, and peak time pricing information to reduce customer demands.These conventional methods are performed by power utilities and independent system operators, which can reduce peak period (PP) demand or shift demand to the off-peak period.An innovative business model called "aggregator program [3][4][5]" combines a group of customers into a single purchasing unit and pools DG resources to achieve higher demand reductions from the main grid, as shown in Figure 1.It offers DR strategies or programs during PP demand and provides available incentives in this innovative electricity market.
An aggregator (power purchaser) acts as a mediator or broker between end users and power utilities in the • Forecast: customer demands, weather conditions, operation conditions.
• Monitor: customer usage 24 hours per day.  1, aggregator programs perform DR to control electrical appliances such as machines, heaters, air conditioners, and pumps via a communication network and control devices.Therefore, an aggregator can directly control end users' energy-intensive appliances in PP demand through the so-called "interruptible service."To activate DGs (ancillary services), an aggregator provides DR programs to modify the customer group's demands and supply backup power during PP times to reduce interruptible loads.Onsite generation in a smart grid such as (1) renewable energy and stored energy and (2) microgenerators, small-scale co-generation systems, and diesel generators [6][7][8] can be temporarily employed to support net changes in main-grid supplied power.Under this environment, advanced metering infrastructure (AMI) or advanced grid infrastructure (AGI) provides a bidirectional monitor and control communication network between main and smart grids.Hence, a smart grid platform can perform the consumptions, and DGs are dispatched with smart meters, communication, intelligent computations, and control algorithms to achieve the DR and ancillary service tasks.
For an innovative electricity market in smart grids, green policies need to (1) adjust dynamic DR programs using active-duty DGs, (2) promote an aggregator business model for energy management, load management, and ancillary services, and (3) encourage green financial loans for constructing DGs and storage devices in Taiwan.If customers have their onsite generations, virtual net metering (VNM) network [9][10][11] is a metering arrangement to assign or transfer exported electricity generation to other customers, as shown in Figure 1.In addition, an aggregator can also integrate the exported electricity generation within its common load area and resell it to nearby customers.This metering infrastructure allows customers in a residential, commercial, and industrial multitenant group to participate in a common renewable source, such as solar energy or wind energy system.AMI and VNM environments also allow an aggregator to dispatch the ancillary power for participation customers and sell to the power utility at a renewable energy rate.Therefore, this study proposes the dynamic game model (DGM) [12,13] to dispatch the onsite DGs for meeting the customers' extra demands, with the following advantages: (1) to reduce PP and partial peak period (PPP) demands, (2) to avoid exceeding the contract demands, (3) to reduce high-price demand from the main grid, and (4) to activate the active-duty DGs.
The remainder of this study is organized as follows: Section 2 describes the demand side aggregator in the smart grid, Section 3 addresses the DGM, and Sections 4 and 5 present simulation results and conclusions to demonstrate the efficiency of the proposed model.

Demand Side Aggregator
In general, power utilities provide incentives to customers for minimizing the operation, transmission, and generation costs (fossil fuels and nuclear energy) in the upper level of the wholesale market.Demand side aggregators then provide DR services to modify customers' consumption patterns to meet the demands by the day-ahead market and demand profiles.Based on DR strategies and ancillary services, demand aggregators and generation aggregators (virtual power plant, VPP) offer the following tasks: (1) to obtain the lowest electricity prices, (2) to plane the uses of DGs, and (3) to shift and control the loads with incentives [14].These two aggregators can be combined in a smart grid, as shown in Figure 1.DR seeks to reduce the peak demand or enable dynamic changes in consumption, responding to market price and incentive payments.The flowchart of DRs and ancillary services in an electricity market is shown in Figure 2.
Given the day-ahead forecast consumptions, a power utility operator announces the incentives to reduce the consumption for each aggregator.DLC is the most appropriate method to control demand with contractual controls or incentive controls that allow the aggregators to shed their customers' appliances [14,15].In addition, some customers, such as residential users or commercial end users, can also use the automatic control system to shed the interruptible appliances with the scheduling DLC.For large-scale commercial and industrial customers, renewable energy (solar and wind generation), energy storage systems, and small-scale power plants are used to support the demand as ancillary services.Therefore, the use of expensive generation costs during PP hours is decreased.In addition, this measure will flatten the consumption load pattern (20%-30%) to avoid high electricity prices, as shown in the modified consumption in Figure 3.The DR program requires exchange data between the end users and aggregators.The AMI or AGI can provide available information for aggregators and electricity consumers, including forecast data, operation conditions, daily/real-time usages, and reserve/storage capacity.To achieve DR, the meter data management system provides two-way communication to and from power utilities and aggregators; available information, such as electricity prices and DR capacities, can be sent to consumers, or an amount of DG capacity can be provided for end users or sold back to the main grid.In addition, VNM allows for less expensive renewable energy installations without needing separate bidirectional meters, inverters, and wiring.It constructs flexibility because the renewable sources can share from one unit to another, which credits customers for onsite renewable energy against their onsite electricity use at full retail electricity rates, including generation, transmission, and distribution fees, and taxes paid based on usage [11].Hence, within the main grid, a smart grid can be operated as a controlled entity integrating a single aggregate and a generator.DGs as ancillary service can support the reserve power pool for peak demands.The aim of the proposed DR model is a program to provide an increase of 20%-30% flexibility and DG ancillary service during PP hours.

Dynamic Game Model (DGM)
In a smart grid, renewable energy and onsite generation as the reserve pool can provide dispatchable power across the microdistribution system that can be used during peak hours.These available auxiliary power sources are installed near electricity consumers and can operate to reduce power losses on transmission lines/feeders.Safety and reliable operations are important criteria, and the frequency and voltage (active/reactive powers) can also accommodate the consumers required.The advantage of renewable energy, such as solar energy and wind energy, can synchronously store into the storage devices and supply loads via the power converter (DC-DC or AC-DC) and inverter (DC-AC) [16,17].The auxiliary power capacity should be at least 50% of the maximum participant demands.The aggregators need to forecast loads and DG capacity to meet the demand with or without the bilateral contracts.However, renewable energy usually relies on local weather conditions, with a typical forecast error of 5%-15% of installed capacity [5].The AC generators, including a microgenerator and a diesel generator, have a stable auxiliary capacity to supply peak demands with the specific contract.
In this study, DGM is used to describe the DG dispatched with the desired demand, Sl,h, and interruptible load, Sil,h, at the h time slot during PP and PPP hours.The desired DG capacity, SDG,h, is estimated by where Spl,h is the demand at PP and PPP, Nl is the number of electricity customer, l = 1, 2, 3, …, Nl.PP is from 10:00 a.m. to 12:00 a.m. and 13:00 p.m. to 17:00 p.m. daily.PPP is from 07:30 a.m. to 10:00 a.m., 12:00 p.m. to 13:00 p.m., and 17:00 p.m. to 22:30 p.m. daily, as shown in Figure 3.The dispatches of DGs can be determined by Sl,h, and the total of Sil,h is determined by the interruptible service.Their quantities are parameterized with certainty factors (CFs), as shown in Figures 4(a) and (b), where the probability value is between "0" and "1" for presenting possible operation states as follows: • for the desired demand (70% -100%): where DR ' l,h is the probability for a customer without attending DR in an electricity market.• for desired capacity of renewable energy and AC generator: where SDG,l is the desired capacity of DGs during PP and PPP hours; the capacity of Srenew is forecasted and announced by the aggregators.
Four dispatch strategies are described in a strategic form [12,13], as shown in Table 1, with rows representing the capacity of selected DGs and columns representing consumer demand with or without attending DR in an electricity market.Four operational states have strategy combinations.Their states are independent if and only if their joint probabilities follow a multiplication rule.For each customer, the outcomes of joint probabilities are given by • desired capacity of ac generator, Sacgen: with attending DR: ) ( where Sacgen,max and Srenew,max are the maximum capacity of AC generator and renewable energy, respectively; Sstorage represents storage energy.AC generator should be within its lower and upper rated capacities as Equation ( 13), and it can rapidly deliver electrical power loads from within 10 s to a few minutes.The flowchart of DG dispatches with subject conditions is shown in Figure 5.

Simulation Results
Within a smart grid, three customers are operated as a controlled entity integrating a single aggregate unit and a generator, as illustrated by the consumption demands (the maximum demand: 52.8 kW, voltage level of 220 V/380 V) in Figure 6(a).We consider the capacity of auxiliary power to be at least 50% of the maximum demand.The DGs include renewable energy (solar and wind energy) and AC generators (a 10 kW microgenerator and a 10 kW diesel generator) with 20-25 and 20 kW auxiliary capacities, respectively, and rated 10 kW of storage device.We assume that the renewable energy is obtained from a small-scale wind farm (five wind turbine generators) and a photovoltaic (PV) array (15 PV panels), in which each one has a rated power of 10 and 15 kW, respectively, as shown in Figure 6(c).However, the efficiency of renewable energy conversions are influenced by native environmental conditions such as solar radiation, wind speed, temperature, and weather conditions.Thus, an average rated capacity of 50% is estimated to be obtained from solar and wind energy that varies from 0.2 kW/m 2 to 1.0 kW/m 2 and from 25 °C to 45 °C in daytime, and wind speed ranges vary from 5 m/s to 10 m/s during the night or in the winter season.The capacity of renewable energy is forecasted by the aggregator.
In this study, 1 day can be divided into 24 time slots.The total of consumption demands in each time slot (1 h interval) is shown in Figure 6(b).Each customer is willing to attend the DR programs with auxiliary services from 7:30 a.m. to 22:30 p.m. (16 h).The following procedures are used: (1) to estimate the demand and DG capacities by day-ahead forecast, (2) to modify the demand load patterns by DLCs, incentive controls, or contractual controls, and (3) to dispatch the auxiliary power by the DGM.For two time slots, such as 10:00 a.m. and 17:00 p.m., the simulation results are shown in Table 2.For S1,10 = 0.6375 pu, S2,10 = 0.8910 pu, and S3,10 = 0.7920 pu, l = 1, 2, 3, and h = 10, the following procedures are  12) to (14).
In addition, the simulation results at 17:00 p.m. are also shown in Figure 6(b) and Table 2.In Figure 6(b), the blue and red lines represent consumption demands before and after DR programs, respectively.The capacities of renewable energy and AC generators can also be determined in each time slot by the DGM.The peak demand can avoid exceeding the contract quantity, which can reduce the demands from 52.800 kW to 36.825 kW (15.975 kW reduction quantity, 30.26% of peak demand) per day.In this study, renewable and storage energy has priority curtailment strategies to reduce the demand from the main grid and avoid expensive energy sources.The flexibility ranges for integrating 45%-100% of renewable energy are achieved.Based on day-ahead operation mode, the best reward, $766.80,can be paid back to the aggregator, and the best total of energy charge, $858.92,can be reduced within the 16 h DR program, as illustrated in various operation modes in Table 3.These charges can be reduced for decreasing the costs of electricity equipment installation, maintenance, transmission, and generation fuel at the power utility side.In addition, this model can also provide for a generation aggregator to bill charges for green energy and AC generation.Under VNM, the electricity produced onsite can be allocated to multiple onsite customers, and the generation is credited to benefiting accounts in this business model.

Conclusion
This study has presented the DR and ancillary services using DGM for an aggregator business model.Based on day-ahead mode, the aggregator acts as a mediator to purchase electricity energy from power utilities.DR is used to reduce the demands during short term or contractual hours.The aggregator uses the incentive controls, contractual controls, or DLC to modify the consumption demands.This measure will avoid the usages exceeding the contracted demand and decrease the use of expensive electricity base and energy rates.The desired demands and capacities of DG are parameterized with CFs, and they are represented strategies to dispatch the DGs in a strategic form.DGM can impartially dispatch ancillary power for short-term imbalances or contractual hours.The simulation results show the efficiency of the proposed model in smart grids.For this business model, the expected outcome provides a promising solution for aggregators to perform DR programs and dispatch the ancillary power.Therefore, this model can promote an innovative business into DR 2.0 in Taiwan.

Figure 1 .
Figure 1.An aggregator business model in smart grids

Figure 2 .Figure 3 .
Figure 2. The flowchart of DRs and ancillary services in an electricity market
Proceedings of the 4th IIAE International Conference on Intelligent Systems and Image Processing 2016 DOI: 10.12792/icisip2016.073

Table 1 .
Strategic form of dynamic game model for each customer

Table 2 .
The simulation results of DR programs and auxiliary services at 10:00a.m. and 17:00p.m.

Table 2 ,
Step 4) dispatch the auxiliary power, as shown in

Table 2 ,
Step 5) check the restricted conditions by equations (