This study proposes a multi-agent-based framework for Peer-to-Peer (P2P) power trading in a locality electricity market (LEM) for self-interested smart residential prosumers. In LEM, prosumers may sell (buy) their excess generation (demand) at a profitable market prices compared to utility prices to achieve a win–win outcome. In LEM, three agents namely locality electricity trading system (LETS), utility and prosumer act together to achieve P2P power trading in a day-ahead market. LETS computes the internal market prices employing any one of the market-clearing mechanisms and broadcasts it to the prosumers. Prosumers optimise the generation-demand schedule for the next day using residential energy management and trading system to achieve minimum electricity bill. The performance of the proposed framework is validated through different case studies on a residential locality with ten prosumers. The simulation is carried out using MATLAB parallel computation tool box and the load data is collected from the residential locality of National Institute of Technology Tiruchirappalli, India. It is evident from the simulation results that all the participants are economically benefited by P2P power trading. It is also found that the SDR mechanism in P2P outperforms and reduces the locality electricity bill by 27–68% under different operating conditions.
1 Introduction
The network of present electricity grid was refurbished more than 50 years ago, when the usage of electricity was limited. The legacy grid was designed for utilities to deliver electricity to consumer buildings with one-way interaction. This makes it difficult for the utility to respond for ever increasing energy demands of the 21st century with integrated distributed energy resources (DERs). The new born smart grid introduces a two-way dialog, where electricity and information can be exchanged between utility and customers. The smart grid enables newer technologies for easy integration of DERs and plug-in hybrid electric vehicles (PHEVs). The smart grid is expected to revive the ageing infrastructure of today's grid and utilities can communicate the time-varying tariff structure with the customers through smart meters to help them for better managing the electricity needs to reduce electricity bills. The smart grid is designed to enable utilities to manage and moderate electricity usage especially during peak demand with the co-operation of customers. Nowadays, the customers are adapting themselves to use smart thermostats, smart plugs, smart bulbs, smart sensors, smart door locks, smart security cameras, smart audio speakers integrated with voice assistance and so on. Residential energy management system (REMS) helps to manage the operation of smart loads based on the time-varying tariff to support the reliable operation of the grid during stressed conditions. More advanced REMS products include algorithms and optimisation approaches that schedule the operation of household appliances and thereby reduce energy and peak demand charges while maintaining an acceptable level of comfort and convenience for the customers, and enable customer participation in demand response, and demand-side management programs. With the smart grid technologies and customer participation, utilities can more easily handle the increased demand due to PHEV charging and variable in-feed from DERs like roof-top solar photovoltaics (PVs), wind turbines and energy storage devices. The customers with in-house DERs are known as prosumers. A prosumer can either act as seller or buyer depending upon the electricity price and excess generation/demand available at different time periods of a day. The power trading between the prosumers and utility can be done by the market operators [1].
In conventional peer-to-grid (P2G) paradigm prosumers trade their net power (demand minus generation) independently with the utility. If the demand of the prosumer is more than generation, then the deficit power must be bought from the utility with the utility decided price. If the generation is more than demand, then the excess power must be sold to the utility according to the feed-in-tariff (FIT) scheme. In many countries, the FIT is less than the utility price, which results in a lengthy investment pay-back period for DER installations to the prosumers. In P2G paradigm, the prosumer is not allowed to sell excess power more than power injection limit to the grid because of power system network operational and stability constraints. Hence, prosumers are in search of new business models for electricity market to obtain high returns, which lift-offs a new peer-to-peer (P2P) power trading paradigm. P2P power trading inspires the prosumers to trade the excess generated power with the neighbourhood. In P2P, if the prosumer has excess power (demand is less than generation), then the first priority is to trade with the neighbourhood and further if there is any redundant power that is used to charge the battery. If the prosumer has deficit power (demand is greater than generation), then the first priority is to meet the unmet demand by buying power from the neighbourhood and further unmet demand if any can be met from battery backup [2]. In late 70s at Massachusetts Institute of Technology (MIT), Fred Schweppe anticipated the usage of dynamic pricing in the electricity markets, which is an essential element of most of the transactive energy (TE) techniques today, to resolve imbalance between supply and demand [3]. TE is associated with the democratisation of electricity and internet of things (IoT) with information and communication technology. According to Grid Wise Architecture Council, TE is a mechanism that sustains a real-time balance between demand and supply in the entire power system network using dynamic pricing models [4]. According to IEEE Power and Energy Society, multi-agent systems (MASs) are suitable for modelling and simulating the architectures in the power system engineering and MAS is found to be the best suitable technique for simulating the behaviour of prosumers under P2P paradigm [5].
Recently, several P2P energy sharing/trading mechanisms and pricing concepts have been proposed under smart grid paradigm using multi-agent-based systems. Herman et al. [6] proposed P2P energy sharing between electric vehicles. This proposed method significantly maintains the dynamic balance between demand and supply during peak and off-peak hours using game-theoretic approach. Celik et al. [7] proposed multi-agent-based co-ordinated neighbourhood energy sharing using game theory for P2P energy sharing. This study explains the day-ahead energy sharing between prosumers in the smart community. Cintuglu et al. [8] proposed a MAS with prosumers and utility using game theory based reverse auction model. This study did not model the prosumers as individual agents, and schedulable loads (SLs) are not considered. Shamsi et al. [9] proposed agent-based simulation framework for community energy market for energy sharing between participants and recursive least square learning mechanism is used to find a dynamic price for decision making but this study also did not consider the modelling of SLs. Long et al. [10] presented three different market clearing mechanisms, namely, bill sharing (BS), mid-market rate (MMR) and auction-based pricing strategy to clear the market in P2P energy sharing market under TE paradigm. This study also did not consider the modelling of battery and SLs. Liu et al. [11] presented a new internal pricing mechanism for P2P energy trading for DER owners in a community microgrid, and the internal prices are calculated as a function of supply-to-demand ratio (SDR). The drawback of this study is the convergence problem in an iterative pricing process. In the P2P paradigm, the participants have to decide on the amount of power and the time duration at which they wish to sell/buy. The decision-making process is complex and requires high computational power for smart consumers, and it plays a vital role in enabling P2P energy sharing. Suyang Zhou et al. [12] proposed a reinforcement learning (fuzzy Q-learning) algorithm to improve the decision-making process for participants to participate in the P2P market paradigm. The proposed Q-learning algorithm solves the decision problems without any requirement of the model and hence the time for decision-making process will significantly reduce. Liu et al. [13] proposed energy storage-equipped energy service provider to enable energy trading among group of PV prosumers using stochastic programming and Stackelberg game theory. Several P2P energy sharing/trading concepts using game theory are detailed in [14–17].
The approaches presented in the aforementioned studies did not consider proper modelling of household appliances (schedulable and non-schedulable). Prosumer agent has been modelled with many assumptions and willingness of prosumer to shift the time of operation of SLs was also not included. Demand response was not considered, while implementing the P2P power trading under TE paradigm. However, proper modelling of household appliances and prosumer agent are carried out in this paper. This paper also deals with two different kinds (willing and unwilling to shift the operating time of SLs) of prosumers for P2P power trading.
The main contributions of this paper are summarised as follows:
(i) Presenting a locality electricity trading system (LETS) agent for a residential community of a smart city to enable P2P power trading between the participants.
(ii) Presenting a residential energy management and trading system (REMTS) architecture at the participant premises for day-ahead power trading under smart grid environment.
(iii) Presenting a TE-based demand response program to minimise the participant's electricity bill using genetic algorithm (GA) as an optimisation tool for re-scheduling the SLs.
2 Architecture of the proposed system
The architecture of proposed multi-agent-based TE system is shown in Fig. 1, which consists of several autonomous agents namely prosumer, a LETS and utility (retailer agent).
Fig. 1Architecture of proposed multi-agent-based TE system
These agents interrelate, negotiate and collaborate with each other to achieve individual goals and a common goal. Prosumer agents are the DER owners, who can sell (buy) their excess generation (demand) based on the power availability from the in-house DERs and internal market prices. The locality electricity market (LEM) is a virtual market built through LETS agent for the mutual benefit of utility and locality participants. LETS is an intermediate supervisory and non-profited distribution market operator agent for managing P2P power trading activities. The objectives of LETS are not only encouraging the prosumers to participate in the LEM but also to maintain the dynamic balance between generation and demand of the locality and hence ensuring its self-sustainability. The utility is a passive agent, who sells and buys electricity at a pre-defined price and responsible for the reliability and security of the locality power system. Modelling of these agents is described in the following sections.
2.1 Prosumer agent
In a smart residential locality, every participant having different residential appliances and in-house DERs is equipped with a REMTS. REMTS is intended to manage the power consumption of the household appliances to obtain minimum electricity bill. All the participants in the locality have an objective to reduce their electricity bill payable to the utility by trading their excess generation (demand) with other participants of the LEM. The architecture of REMTS is shown in Fig. 2, which mainly consists of an energy management system and a trading module.
Fig. 2Architecture of REMTS
The energy management system communicates with the residential appliances through a wireless home area network (HAN) [18]. The residential appliances are categorised into three categories such as non-interruptible non-schedulable loads (NINSLs), interruptible non-schedulable loads (INSLs), and SLs [19]. Essential and entertainment loads are examples of NINSLs. Temperature control loads like heating, ventilation and air conditioning are examples of INSLs. PHEV, dishwasher, washing machine are the examples of SLs. The energy management system receives the dynamic price information from the utility through smart meter interface and load parameters (status and setpoint temperature of INSLs, starting time and ending time of SLs) from the household appliances. The control signals are generated by the energy management algorithm and transmitted to control the INSLs and SLs through HAN. The operational parameters of NINSLs, INSLs and SLs such as load initialisation time, load dead time and pre-emptive status of SLs can be directly given through the user interface. The expected day-ahead power exchange pattern is predicted by the trading module considering the comfort and desire of the residents, history data of demand variation and forecasted atmospheric weather conditions. The smart meter interface module helps the REMTS to communicate with a smart meter in HAN, which in turn communicates with the utility through LETS via a neighbourhood area network (NAN) established with WiMAX or cellular technology. Interconnection of the smart meters of all the residents of the locality through NAN enables every prosumer to participate in LEM for P2P power trading.
In MASs for implementing P2P power trading, each autonomous prosumer agent receives an input set (internal price set) from the LETS agent and performs an optimisation process using binary genetic algorithm (BGA) to determine the optimal intervals for the operation of SLs to minimise electricity cost, and produces an output set (optimal intervals for SLs and updated demand pattern). The input set and output set of a prosumer agent are calculated as follows. The input set and output set of prosumer agent can be derived as follows:
Let be the set of prosumer agents willing to participate in the LEM for P2P power trading. N is the total number of prosumer agents in the locality. Let be the set of operating time intervals and T is total number of time intervals. In this paper, 1 h is considered as the time duration of an operating interval. The input set for the prosumer agent n is defined as
(1)
where , and are the set of internal prices (buying and selling prices) which are calculated and broadcasted by the LETS, SL data obtained through user interface and initial day-ahead participant electricity trading schedule (PTES), respectively. The day-ahead PETS of prosumer agent n is defined as
(2)
where and are the set of day-ahead total expected power demand and total expected power generation, respectively, of prosumer agent n throughout the time horizon T, which are defined as
The day-ahead ted and teg of prosumer during the interval are calculated by
(3)
(4)
where , , are the day-ahead total expected demand of NINSLs, INSLs and SLs, respectively, of prosumer n during interval t. The , are the day-ahead total expected power generation from solar and wind, respectively, of prosumer n during the interval t. The total expected demand for different appliances are calculated as follows.
The total expected demand of NINSLs for prosumer n during the interval t is calculated by
(5)
where is prosumers’ user-defined factor and is average of demand during interval t of similar days over a user-defined number of weeks. The total expected demand of INSLs for prosumer n during the interval t is calculated by
(6)
where is the expected status (ON = 1, OFF = 0) of INSL a during the interval t where is INSLs set, A is the total number of INSLs in the home, and is the rated power of INSL a. The expected status of INSLs is influenced by weather predictions and consumer comfort with respect to time. The total expected demand of SLs for prosumer n during the interval t is calculated by
(7)
where is the expected operational status (ON = 1, OFF = 0) of SL b during the interval t, where is SLs set, B is the total number of SLs in the home, and the is the rated power of SL b. The expected operational status of SLs is influenced by consumer preference with respect to time. Scheduling of SLs plays a key role in the P2P power trading. The day-ahead total expected generation from different in-house DERs are calculated as follows:
The solar power generation is depending on the available amount of solar irradiation and ambient temperature with respect to time t, and it is calculated using (8) [20, 21]
(8)
(9)
where is output power delivered by solar PV panels in kW, is the de-rating factor of solar PV panel, is the PV power () at standard test condition (STC), is the averaged solar irradiation in during interval t. is the solar irradiation at STC (), is the PV cell temperature during interval t in , is the temperature at STC in , is the PV cell temperature coefficient, NOCT is the normal operating cell temperature in and is the averaged ambient temperature during interval t in . The total expected wind power generation is highly depending on wind speed, and it is calculated using
(10)
where is the power generated (kW) by wind turbine during interval t, air density () is , swept area () is , is the averaged wind velocity () during the interval t and is the power coefficient. The accuracy of total expected renewable power generation completely relies on the precision of weather prediction in the residential locality. The weather prediction can be done either by the individual prosumer agent or by the LETS agent and communicated to the individual prosumer through smart meter interface. In this study, the day-ahead expected power generation is predicted using artificial neural network (ANN) in the prosumer agent by REMTS. The computation of exact renewable power supply is very crucial and plays a major role for the successful operation of P2P power trading.
Based on the information received from the LETS agent, the REMTS of every prosumer agent executes an optimisation algorithm to minimise day-ahead electricity bill by re-scheduling the SLs. The objective function for prosumer n is described as
(11)
where
where , , are net expected power of prosumer n, internal buying price and internal selling price, respectively, during the interval t, and the functions and are the equality and inequality constraints describing the limits on load operating intervals and prosumer satisfaction, respectively [22]. In this study, BGA is used to optimise the objective function. The output set for the prosumer agent n is defined as
(12)
where and are the updated operational status of SLs and updated PETS, respectively, of prosumer agent n over the time horizon T. After the optimisation process, the updated PETS is transmitted to the LETS agent through the smart meter interface.
2.2 LETS agent
The functions of the LETS agent are to receive the PETS from prosumer agents and compute the internal market price set, and broadcast it to them. The input set to the LETS agent is defined as
(13)
where RP and EP are the retail price and export price, respectively, given by the utility and these prices are considered to be unvarying throughout the time horizon T. After receiving the information from the utility agent and prosumer agents, LETS agent calculates the internal market price sets. Different pricing mechanism, such as SDR, BS, MMR were proposed in the literature to determine the internal market price set. These techniques are used in this study to calculate internal price sets. The output set of the LETS agent is defined as
(14)
where and are the internal buying and the internal selling price sets, respectively, throughout time horizon T. In order to fascinate the prosumers into the LEM, the internal prices are modelled as the prices between utility prices, i.e.
(15)
The internal prices are calculated using market-clearing mechanisms, which are described in the following sub-sections.
2.2.1 Supply-to-demand ratio
In this mechanism, the internal prices of LEM are modelled as function of SDR, which is calculated by
(16)
The value of SDR varies in every time interval t because of the fluctuations in renewable power generation and consumer's demand. This results in time-varying internal market prices (both buying and selling). The internal price set computed by LETS agent is defined as
(17)
where , are the internal buying and internal selling price sets throughout time horizon T, respectively, which are expressed as
(18)
These internal price sets of LEM are computed by LETS using SDR value and utility prices. Let RP and EP represent the utility retail and export prices, respectively. Utility retail price is considered to be higher than the utility export price. Then the internal prices are calculated by
(19)
(20)
2.2.2 Bill sharing
In BS mechanism, the internal market prices are calculated based on amount of energy consumed/generated by every participant proportional to the whole demand/generation in the LEM. In P2G paradigm, residual power demand of prosumer n, i.e. the amount of importing power from the utility is denoted as and surplus power generation of prosumer n, i.e. exporting power to the utility is denoted as . The importing and exporting powers of prosumer n during interval t are calculated by
(21)
(22)
The total energy importing from the utility and total energy exporting to the utility in the community is denoted as and , respectively, throughout the time horizon T, which are calculated by
(23)
(24)
In P2P paradigm, the excess power demand of LEM, i.e. the amount of buying power from the utility is denoted as and excess power generation of LEM, i.e. selling power to the utility is denoted as , which are calculated by
(25)
(26)
where
(27)
(28)
Let the buying and selling prices of utility in LEM are RP and EP, respectively. The total energy buying from the utility and total energy selling to the utility by LEM are calculated by
(29)
(30)
The internal prices (buying and selling) under P2P paradigm are calculated by
(31)
(32)
Consider , the numerator term is the cost of total energy buying from the utility by the LEM, the denominator term is the sum of individual prosumers total energy demand consumption throughout time horizon T.
2.2.3 Mid-market rate
In the MMR technique, the internal prices are modelled as an average of retail and export prices of the utility. The average of retail and export prices is denoted as , which is calculated by
(33)
LETS calculates the total expected demand and generation of LEM during interval t, which are calculated by (27) and (28), respectively. Considering fluctuations in renewable power supply and consumer demand, the internal prices will vary with respect to time interval t.
•
When then the internal prices are calculated by
(34)
•
When then the internal prices are calculated by
(35)
•
When then the internal prices are calculated by
(36)
2.3 Utility agent
Utility is a passive agent in the P2P paradigm. The quality of supply, load shape, peak load are the crucial factors for utility agent. The utility agent sell power at the cost of retail price when the LEM has deficit power demand, utility agent buy power at the cost of export price according to FIT scheme when the LEM has excess power generation. In this paper, constant retail and export prices are considered because in most of the countries, the retail price and export price are constant throughout the day.
3 Participants day-ahead power trading strategy
To accomplish the power trading between participants in the LEM by minimising the objective function (cost function), every participant must send PETS to LETS agent simultaneously following a request from LETS agent. LETS agent calculates the internal buying and selling prices of LEM using any one of the market-clearing mechanism, and these internal prices are broadcasted to all the participants in the LEM through smart meter interface. After knowing the internal prices, every participant executes an optimisation for re-scheduling operating intervals of SLs considering the internal prices to minimise the electricity cost function using BGA. The PETS of every participant is re-calculated by the trading module of REMTS after solving the optimised cost function and then sends back to LETS agent through smart meter interface. This process is simultaneously performed by all the participants in LEM. The exchange of PETS and internal prices information between LETS agent and REMTS of every participant are repeated till the locality load demand is met by the locality generation or till the maximum number of iterations is reached, which is decided by the LETS. Once the iterative process is over, the participants are expected to strictly follow the trading schedule. Classification of participants as buyers and sellers in every interval is decided by the SDR of individual prosumer , which is defined as
Let be the set of sellers and be the set of buyers: , .
, then and
, then The steps followed for implementing the P2P power trading are presented in the flowchart of Fig. 3, and the algorithm for implementing P2P power trading is shown in Algorithm 1 (see Fig. 4).
Fig. 3Flowchart for the P2P power trading process
Fig. 4Algorithm 1: algorithm for implementing P2P power trading
4 Case study
The performance of proposed multi-agent-based TE system in the LEM for enabling P2P power trading between households in the residential community of a smart city is validated by carrying out few case studies with ten residential prosumers as self-interested participants. The residential load data has been collected from the residential locality of NIT Tiruchirappalli, one of the smart city projects of Govt. of India. The functionalities of REMTS and LETS have been simulated in MATLAB. The renewable power generation at the prosumers buildings is estimated based on the relevant mathematical expressions using the weather data predicted by ANN. The historical weather data is collected from the weather station installed at NIT Tiruchirappalli. The case studies are done in a computer simulation using the collected data from the field. The utility retail price and export price are $0.15 and $0.05, respectively [10]. Since the reduction in electricity price in LEM depends upon power generation from renewable sources, the participants can size their renewable sources with due consideration to installation cost, and utility regulations. The duration of trading interval is taken as 1 h (t = 1 h and T = 24 h), and in this interval, the internal prices are assumed to be constant. The details of participants’ DERs and household appliances along with their ratings are shown in Tables 1–4, respectively.
Table 1 Participants’ DER sizes
Participant index
Solar PV panel (each rated as 0.1 kW)
Wind turbine (each rated as 1 kW)
P-1
80
2
P-2
80
2
P-3
65
3
P-4
75
3
P-5
50
3
P-6
40
2
P-7
30
2
P-8
30
2
P-9
25
2
P-10
30
2
Table 2 Non-interruptible and non-schedulable loads
S. No.
Load
Power, kW
Durationa, h
Qtya
1
fan
0.06
00.00–05.00
3
05.00–08.00
2
17.00–21.00
2
21.00–24.00
4
2
fluorescent lamp
0.06
05.00–07.00
6
18.00–22.00
6
3
CFL
0.04
00.00–05.00
4
05.00–07.00
4
18.00–22.00
6
22.00–24.00
4
4
television
0.08
06.00–08.00
1
17.00–22.00
1
5
mobile/
0.06
06.00–08.00
2
laptop charging
17.00–19.00
2
aDuration; quantity varies from participant to participant.
Table 3 Interruptible and non-schedulable loads
S. No.
Load
Power, kW
Duration, h
1
air conditioner
1.0
00.00–06.00
21.00–24.00
2
refrigerator
0.5
00.00–24.00
3
water heater
2.0
05.00–08.00
17.00–20.00
Table 4 Schedulable loads
S. No.
Load
Rating, kW
Pre-emptive status
Durationa, h
Run-time, h
1
grinder
0.8
1
03–08
1
2
washing machine
0.5
1
04–08
1
3
dish washer
2
0
05–09
1
11–14
1
20–22
1
4
well pump
1.5
0
02–05
1
15–19
1
5
PHEV charging
2.5
0
02–05
3
20–24
3
aDuration – duration of SLs varies from participant to participant.
The consumption pattern of household appliances (SLs) varies from participant to participant in a day.
4.1 Case I: all participants are prosumers and all the SLs are having operating time constraints
Consider there are ten participants in the smart locality and all the participants are prosumers and they are more sensitive to change the SL operating intervals. Fig. 5 illustrates the locality net consumption (total expected demand minus total expected generation) and total DERs generation pattern under P2G paradigm in this case study.
Fig. 5Net consumption and total generation of LEM under P2G paradigm in case-I
Under the P2G paradigm, surplus generation is available from interval 08 to 18 (08.00 AM to 06.00 PM) to meet the demand of the SLs in the whole locality. Due to the rigid operating time limits of SLs, the surplus generation is fed into the utility. Fig. 6 illustrates the net consumption pattern comparison between P2G and P2P paradigms with different internal pricing mechanisms.
Fig. 6Net consumption of the locality under P2G and P2P paradigms in case-I
The day-ahead electricity bills of the participants in the locality under P2G and P2P paradigms are illustrated in Fig. 7.
Fig. 7Electricity bill for prosumers under P2G and P2P paradigms in case-I
It is noticeable that the participants gain economic benefits in P2P paradigm compared to P2G paradigm. The total savings of the locality is 34% under P2P compared to P2G paradigm in this case study. The percentage savings of all the participants under P2P compared to P2G are furnished in Table 5.
Table 5 Percentage savings of participants under P2P compared to P2G in case-I
Prosumer
SDR
BS
MMR
IS, %
SS, %
IS, %
SS, %
IS, %
SS, %
P-1
31.53
8.18
30.94
8.14
32.71
8.50
P-2
36.03
12.33
33.92
11.77
36.03
12.35
P-3
27.70
8.77
24.26
7.79
27.70
8.78
P-4
30.86
10.20
32.19
10.80
31.79
10.53
P-5
37.80
9.68
38.40
9.98
37.80
9.70
P-6
36.08
11.67
35.68
11.71
36.08
11.69
P-7
43.44
10.50
42.78
10.49
43.44
10.52
P-8
33.89
9.60
33.15
9.53
33.90
9.63
P-9
30.75
9.57
31.44
9.92
29.79
9.28
P-10
32.68
9.46
33.41
9.81
31.02
8.99
IS – percentage saving of individual participant.
SS – percentage share in total savings of locality.
4.2 Case II: all participants are prosumers and all the SLs are having no operating time constraints
In case I, since the participants are reluctant to change the operating time of SLs, the locality generation is not properly utilised for locality loads, and the surplus power from DERs is fed to the utility. In this case, all the participants are willing to change the SL consumption intervals. Fig. 8 illustrates locality net consumption and total DERs generation patterns under P2G paradigm.
Fig. 8Net consumption and generation of LEM under P2G paradigm in case-II
Fig. 9 illustrates comparison between net consumption pattern under P2G paradigm and with different internal pricing mechanisms under P2P paradigm.
Fig. 9Net consumption of the locality under P2G and P2P paradigms in case-II
It is observed that the amount of power sold to the utility from interval 08 to 18 (08.00 AM to 06.00 PM) is less in P2P compared to P2G paradigm, which means that under P2P paradigm the locality demand is met by the locality generation, if the participants are flexible in operating SLs. The day-ahead expected electricity bill of the participants in the locality under P2G and P2P paradigms are illustrated in Fig. 10.
Fig. 10Electricity bill for prosumers under P2G and P2P paradigms in case-II
The total savings of the locality is 68% under P2P compared to P2G paradigm in this case study. The percentage savings of all the participants under P2P compared to P2G are furnished in Table 6.
Table 6 Percentage savings of participants under P2P compared to P2G in case-II
Prosumer
SDR
BS
MMR
IS, %
SS, %
IS, %
SS, %
IS, %
SS, %
P-1
65.96
8.56
61.19
8.71
71.71
9.13
P-2
63.92
10.93
59.04
11.08
63.06
10.59
P-3
62.98
9.97
53.10
9.22
56.96
8.85
P-4
57.88
9.57
55.59
10.09
59.29
9.62
P-5
78.57
10.06
71.37
10.03
85.07
10.70
P-6
67.44
10.90
62.92
11.17
71.45
11.34
P-7
79.14
9.57
75.79
10.06
86.80
10.30
P-8
68.61
9.72
61.22
9.52
69.07
9.61
P-9
66.66
10.37
58.60
10.00
62.66
9.57
P-10
71.28
10.31
63.31
10.06
72.16
10.25
IS – percentage saving of individual participant.
SS – percentage share in total savings of locality.
The simulation results shown in Fig. 10 confirm that all the participants in the LEM are economically benefited by the proposed multi-agent-based simulation framework under TE paradigm, and it is confirmed that the trading reduces the expected electricity bill of every participant in LEM in P2P compared to the conventional P2G paradigm. The amount of savings depends on many factors, which include precise weather prediction, deviation in real-time demand consumption and intermittent nature of DERs. The deviation in real-time consumption in LEM would penalise the participants, which depends on the magnitude of deviation from day-ahead demand and generation. Consider a case where a participant suddenly needs to use additional loads, which was not included in PETS. In this case, the LETS charges that participant with utility price for the additional energy usage. Consider another case, in which a participant's renewable power generation increases more than what was submitted in earlier PETS because of decreased demand or due to weather conditions. In this case, LETS gives a message to that particular participant to increase demand or he/she will be charged with the utility export price instead of the internal selling price during that period.
4.3 Case III: five participants are prosumers and five participants are consumers and all the SLs are having operating time constraints
The installation of renewable energy sources in the residential buildings merely depends upon the consumer interest and financial ability. All the consumers in a residential locality may not afford to do it. However, the proposed methodology extends the opportunity for such consumers to reduce their electricity bills by participating in P2P market instead of the P2G market. The study is extended by considering five participants (1, 2, 3, 4 and 5) as prosumers and remaining five participants (6, 7, 8, 9 and10) as consumers. Further, it is assumed in this case that, all the participants are more sensitive and reluctant to change the SL operating intervals. Fig. 11 illustrates locality net consumption and total DER generation patterns under P2G paradigm.
Fig. 11Net consumption and generation of LEM under P2G paradigm in case-III
Fig. 12 illustrates the comparison between net consumption pattern under P2G paradigm and with different internal pricing mechanisms in P2P paradigm.
Fig. 12Net consumption of the locality under P2G and P2P paradigms in case-III
It can be observed that the total excess power sold to the utility from interval 08 to 18 (08.00 AM to 06.00 PM) is less in P2P paradigm compared to P2G paradigm. The day-ahead expected electricity bill of the participants under P2G and P2P is shown in Fig. 13.
Fig. 13Electricity bill for participants under P2G and P2P paradigms in case-III
From these results, it can be understood that the P2P power trading methodology provides substantial economic benefits to all the participants in a residential community. The total savings of the locality is 27% under P2P compared to P2G paradigm in this case study. The percentage savings of all the participants under P2P compared to P2G are furnished in Table 7.
Table 7 Percentage savings of participants under P2P compared to P2G in case-III
Prosumer
SDR
BS
MMR
IS, %
SS, %
IS, %
SS, %
IS, %
SS, %
P-1
42.10
9.69
42.72
10.58
68.91
14.38
P-2
43.39
14.07
46.76
16.32
60.52
17.80
P-3
30.87
9.11
37.42
11.89
52.01
13.93
P-4
38.53
12.00
42.37
14.20
56.59
15.98
P-5
48.88
11.06
47.22
11.50
81.05
16.63
P-6
20.13
10.10
14.67
7.926
11.04
5.025
P-7
19.35
7.82
14.67
6.39
10.62
3.89
P-8
21.14
9.20
14.67
6.88
10.08
3.98
P-9
19.12
8.72
14.67
7.20
10.55
4.36
P-10
18.23
8.17
14.67
7.07
9.77
3.97
IS – percentage saving of individual participant.
SS – percentage share in total savings of locality.
4.4 Case IV: five participants are prosumers and five participants are consumers and the SLs are having no operating time constraints
In this case, presence of prosumers and consumers in the locality is considered along with the assumption that all the participants are flexible to change the SL consumption pattern. Fig. 14 illustrates overall net consumption and total DERs generation patterns under P2G paradigm.
Fig. 14Net consumption and generation of LEM under P2G paradigm in case-IV
Fig. 15 illustrates comparison between net consumption pattern under P2G paradigm and with different internal pricing mechanisms under P2P paradigm.
Fig. 15Net consumption of the locality under P2G and P2P paradigms in case-IV
The day-ahead expected electricity bill of the participants in the locality under P2G and P2P paradigms are illustrated in Fig. 16.
Fig. 16Electricity bill for participants under P2G and P2P paradigms in case-IV
It can be clearly observed that P2P paradigm ensures better economic benefits to all the participants of LEM. The total savings of the locality is 60% under P2P compared to P2G paradigm in this case study. The percentage savings of all the participants under P2P compared to P2G are furnished in Table 8.
Table 8 Percentage savings of participants under P2P compared to P2G in case-IV
Prosumer
SDR
BS
MMR
IS, %
SS, %
IS, %
SS, %
IS, %
SS, %
P-1
84.96
9.08
73.44
10.92
100
17.83
P-2
76.98
11.59
70.67
14.81
93.93
19.69
P-3
77.13
10.57
71.37
13.61
89.92
17.15
P-4
70.20
10.15
67.50
13.58
91.00
18.31
P-5
100
10.68
85.07
12.44
100
19.99
P-6
43.91
10.23
23.84
7.73
20.35
6.60
P-7
44.58
8.37
23.84
6.23
20.97
5.48
P-8
46.15
9.33
23.84
6.71
21.39
6.02
P-9
47.08
9.97
23.84
7.03
11.14
3.28
P-10
47.93
9.97
23.84
6.90
21.81
6.31
IS – percentage saving of individual participant.
SS – percentage share in total savings of locality.
5 Conclusion
In recent days, significant transformation is taking place in power sector across the globe. Residential buildings are becoming smart homes, cities are expected to turn into smart cities, utility grid is being embedded with smartness using information and communication technology and IoT. With the growing trend in installation of DERs, availability of flexible loads and battery energy storage systems, P2P power trading is gaining paramount importance to improve the economy of the smart homes and smart cities. This paper presents a multi-agent-based framework for implementing P2P power trading in the LEM for the economic benefits of prosumers in the residential locality of smart cities. The prosumers may act as seller (buyer) of electricity depending upon the availability of excess (deficit) generation from in-house DERs. The independent market operator, LETS, decides the internal market price and broadcasts it to all the participants. The market participants optimise their generation-demand schedule to obtain minimised daily electricity bill. In the present work, the interaction between the participants and LETS is modelled using conceptual game theory and implemented in MATLAB parallel processing environment. An iterative distributed algorithm allows every participant to alter their demand pattern by rescheduling the SLs considering the internal market price. The simulation results demonstrate that the proposed approach effectively manages the utilisation of DERs in a residential locality. It is evident from the simulation results that all the participants are economically benefited by P2P power trading. It is also found that the SDR mechanism of determining market price outperforms and reduces the locality electricity bill by 27–68% under different operating conditions compared to conventional P2G trading. The advantages of implementing P2P power trading in residential communities of smart cities are availability of reliable and decentralised power supply, reduced electricity transportation cost, green power (carbon-free generation) from DERs, choice for consumers and transparency to buy power from oligopoly sources, reduced energy bills (in $), reduced kVA rating of distribution transformer, reduced power (kW) dependency from the utility. The proposed work will influence future energy for smart cities by promoting green buildings with integrated DERs. The option for rescheduling of loads may lead to modification in lifestyle/culture. Smart cities look for re-defining the urban energy distribution systems with interconnected neighbourhoods or smart-hoods. The P2P power trading will contribute a lot to achieve many objectives of the smart cities such as zero pollution, decrease in energy prices and self-sustainable cities.
6 Acknowledgment
This research work was supported by the Ministry of Electronics and Information Technology, Government of India under the Visvesvaraya Young Faculty Research Fellowship (grant PhD-MLA-4(16)/2014 dated 7 April 2016).
7 References
1.
Wang C., Ekanayake J., and Wu J.: ‘Smart electricity distribution networks’ (Taylor & Francis; CRC Press, Boca Raton, Florida, USA, 2017, 1st edn.)
Paudel A., Chaudhari K., Long C., et al: ‘Peer-to-peer energy trading in a prosumer-based community microgrid: A game-theoretic model’, IEEE Trans. Ind. Electron., 2019, 66, (8), pp. 6087–6097
McArthur S.D.J., Davidson E.M., Catterson V.M., et al: ‘Multi-agent systems for power engineering applications part i: concepts, approaches, and technical challenges’, IEEE Trans. Power Syst., 2007, 22, (4), pp. 1743–1752
Alvaro-Hermana R., Fraile-Ardanuy J., Zufiria P.J., et al: ‘Peer to peer energy trading with electric vehicles’, IEEE Intell. Transp. Syst. Mag., 2016, 8, (3), pp. 33–44
Celik B., Roche R., Bouquain D., et al: ‘Coordinated neighborhood energy sharing using game theory and multi-agent systems’. 2017 IEEE Manchester PowerTech. Manchester, United Kingdom, 2017, pp. 1–6
Cintuglu M.H., Martin H., and Mohammed O.A.: ‘Real-time implementation of multiagent-based game theory reverse auction model for microgrid market operation’, IEEE Trans. Smart Grid, 2015, 6, (2), pp. 1064–1072
Long C., Wu J., Zhang C., et al: ‘Peer-to-peer energy trading in a community microgrid’. IEEE Power & Energy Society General Meeting, 2017, Chicago, IL, USA, 2017, pp. 1–5
Liu N., Yu X., Wang C., et al: ‘Energy-sharing model with price-based demand response for microgrids of peer-to-peer prosumers’, IEEE Trans. Power Syst., 2017, 32, (5), pp. 3569–3583
Zhou S., Hu Z., Gu W., et al: ‘Artificial intelligence based smart energy community management: A reinforcement learning approach’, CSEE J. Power Energy Syst., 2019, 5, (1), pp. 1–10
Liu N., Cheng M., Yu X., et al: ‘Energy-sharing provider for pv prosumer clusters: a hybrid approach using stochastic programming and stackelberg game’, IEEE Trans. Ind. Electron., 2018, 65, (8), pp. 6740–6750
Long C., Wu J., Zhang C., et al: ‘Feasibility of peer-to-peer energy trading in low voltage electrical distribution networks’, Energy Procedia, 2017, 105, pp. 2227–2232
Zhang C., Wu J., Cheng M., et al: ‘A bidding system for peer-to-peer energy trading in a grid-connected microgrid’, Energy Procedia, 2016, 103, pp. 147–152
Morstyn T., Darby S.J., Farrell N., et al: ‘Using peer-to-peer energy-trading platforms to incentivize prosumers to form federated power plants’, Nature Energy, 2018, 3, pp. 94–101
Singh S., Roy A., and Selvan M.P.: ‘Smart load node for nonsmart load under smart grid paradigm: a new home energy management system’, IEEE Consum. Electron. Mag., 2019, 8, (2), pp. 22–27
Arun S.L. and Selvan M.P.: ‘Intelligent residential energy management system for dynamic demand response in smart buildings’, IEEE Syst. J., 2018, 12, (2), pp. 1329–1340
Katsigiannis Y.A., Georgilakis P.S., and Karapidakis E.S.: ‘Hybrid simulated annealing–tabu search method for optimal sizing of autonomous power systems with renewables’, IEEE Trans. Sustain. Energy, 2012, 3, (3), pp. 330–338
Arun S.L. and Selvan M.P.: ‘Prosumer based demand response for profitable power exchange between end-user and utility’. 2018 20th National Power Systems Conf. (NPSC), Tiruchirappalli, India, 2018, pp. 1–6
Arun S.L. and Selvan M.P.: ‘Dynamic demand response in smart buildings using an intelligent residential load management system’, IET Gener. Transm. Distrib., 2017, 11, (17), pp. 4348–4357
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.
Citing Literature
Tayfur Gokcek, Ibrahim Sengor, Barry P. Hayes, Ozan Erdinc, A hierarchical approach for P2P energy trading considering community energy storage and PV‐enriched system operator, IET Generation, Transmission & Distribution, 10.1049/gtd2.12636, 16, 23, (4738-4749), (2022).
