Economic Pricing in Peer-to-Peer Electrical Trading for a Sustainable Electricity Supply Chain Industry in Thailand
Abstract
:1. Introduction
2. Related Work/Literature Review
2.1. P2P Energy Trading
2.2. Demand and Supply Theory
Researchers | Highlights | Methods | Results | Limitations |
---|---|---|---|---|
Lin et al. [34] | Bidding strategies | Demand and supply | k-double auction | Benefit contribution |
Zheng et al. [35] | Residential sharing | Content filtering | Reduced costs | Limited energy supply |
Lohachab et al. [37] | Smart cyber–physical systems | Hyperledger caliper | Scalability | Capability of CPS |
An et al. [38] | An appropriate trading price | Optimal trading | Model application | No auction mechanism |
Wu et al. [39] | Multi-scale flexibility | Multi-level market | Reduced costs | Need to develop platform |
Zhang et al. [40] | Hybrid random walk | Double auction | Reducing peak | Confidential trading |
Zhang et al. [41] | Interative double auction | Demand and supply | Trading model | Twenty procumers |
Wu et al. [42] | Sharing economy | Literature review | Pricing mechanism | No pricing optimization |
2.3. Bipartite Graph Theory
2.4. Contents and Contributions
- (1)
- In this study, the P2P energy trading model, based on the mix of the double-auction technique, the pricing optimization, the demand-and-supply law, and bipartite graph theory, is presented to minimize the previous research gaps. Several N-house case studies that take the responsive demand and the varied numbers of participating prosumers into account are carried out in order to demonstrate the efficacy of the studied model.
- (2)
- After the developed model is conducted and verified, the benefits of the practical implications to the economic pricing in P2P electrical trading for the SESCI in Thailand are explored in terms of the social environment and the economic area. In addition, the SESCI consists of five main processes in the electrical industry, i.e., fuel procurement, electricity production, electrical transmission system, electricity transportation in the distribution system, and electrical retail.
3. Research Methodology
3.1. The 4-House P2P Trading Model
3.1.1. Random Price Numbers in Double-Auction Bidding for 4 Houses
random.uniform (0.2, 5.0)
random.randint (10, 100)
3.1.2. Price at the Equilibrium State
3.2. The N-House P2P Trading Model
- Random numbers for use of double-auction bids for more than 4 houses.
- Determination of purchasing sentiment and selling using the bipartite graph principle.
- Determining the probability of winning an auction compared to the price.
- Finding the equilibrium price using the principle of supply and demand.
4. Tested Results
4.1. The Four-House P2P Trading Model
= |(15.00–20.00)/(4.00–5.00)| = 5
= |(7.00–3.00)/(4.00–5.00)| = 4
23 − 4 Px = −5 + 5 Px
4.2. The N-House P2P Trading Model
4.2.1. Relationship between Buying Success Confidence and Bid Price
4.2.2. Relationship between Sales Success Confidence and the Offer Price
4.2.3. The Trend of the Relationship between Price and Sentiment for N Houses
5. Discussion
6. Beneficially Practical Implications of Economic Pricing in P2P Electrical Trading for a Sustainable Electricity Supply Chain Industry (SESCI) in Thailand: Social, Environment, and Economic Aspects
7. Conclusions, Limitations, and Future Work
7.1. Explaining the Price Mechanism
7.2. The Appropriate Price at the Equilibrium Point
Author Contributions
Funding
Data Availability Statement
Conflicts of Interest
References
- ERC. Establishment of the Energy Regulatory Commission. Available online: https://www.erc.or.th/en/history-page (accessed on 15 July 2023).
- Leelasantitham, A. A Business Model Guideline of Electricity Utility Systems Based on Blockchain Technology in Thailand: A Case Study of Consumers, Prosumers and SMEs. Wirel. Pers. Commun. 2020, 115, 3123–3136. [Google Scholar] [CrossRef]
- ERC. Establishment of the Energy Regulatory Commission. Available online: https://www.erc.or.th/en/power-purchasing3/2697 (accessed on 30 September 2022).
- MEA. Type 1 Residential Service. Available online: https://www.mea.or.th/en/our-services/tariff-calculation/other/-yosbxMGAjzp0 (accessed on 2 August 2023).
- Wongsamerchue, T.; Leelasantitham, A. An Electronic Double Auction Prepaid Electricity Trading Using Blockchain Technology. J. Mob. Multimed. 2022, 18, 1829–1850. [Google Scholar] [CrossRef]
- Baig, M.J.A.; Iqbal, M.T.; Jamil, M.; Khan, J. Design and implementation of an open-Source IoT and blockchain-based peer-to-peer energy trading platform using ESP32-S2, Node-Red and, MQTT protocol. Energy Rep. 2021, 7, 5733–5746. [Google Scholar] [CrossRef]
- Bandara, K.Y.; Thakur, S.; Breslin, J. Flocking-based decentralised double auction for P2P energy trading within neighbourhoods. Int. J. Electr. Power Energy Syst. 2021, 129, 106766. [Google Scholar] [CrossRef]
- Khorasany, M.; Dorri, A.; Razzaghi, R.; Jurdak, R. Lightweight blockchain framework for location-aware peer-to-peer energy trading. Int. J. Electr. Power Energy Syst. 2021, 127, 106610. [Google Scholar] [CrossRef]
- Park, B.R.; Chung, M.H.; Moon, J.W. Becoming a building suitable for participation in peer-to-peer energy trading. Sustain. Cities Soc. 2022, 76, 103436. [Google Scholar] [CrossRef]
- Huang, T.; Sun, Y.; Jiao, M.; Liu, Z.; Hao, J. Bilateral energy-trading model with hierarchical personalized pricing in a prosumer community. Int. J. Electr. Power Energy Syst. 2022, 141, 108179. [Google Scholar] [CrossRef]
- Wongthongtham, P.; Marrable, D.; Abu-Salih, B. Blockchain-enabled Peer-to-Peer energy trading. Comput. Electr. Eng. 2021, 94, 107299. [Google Scholar] [CrossRef]
- Dorahaki, S.; Rashidienjad, M.; Ardestani, S.F.F.; Abdollahi, A.; Salehizadeh, M.R. A Peer-to-Peer energy trading market model based on time-driven prospect theory in a smart and sustainable energy community. Sustain. Energy Grids Netw. 2021, 28, 100542. [Google Scholar] [CrossRef]
- Hu, Q.; Zhu, Z.; Bu, S.; Li, F. A multi-market nanogrid P2P energy and ancillary service trading paradigm: Mechanisms and implementations. Appl. Energy 2021, 293, 116938. [Google Scholar] [CrossRef]
- Leong, C.H.; Gu, C.; Li, F. Auction Mechanism for P2P Local Energy Trading considering Physical Constraints. Energy Procedia 2019, 158, 6613–6618. [Google Scholar] [CrossRef]
- Esmat, A.; de Vos, M.; Ghiassi-Farrokhfal, Y.; Palensky, P.; Epema, D. A novel decentralized platform for peer-to-peer energy trading market with blockchain technology. Appl. Energy 2021, 282, 116123. [Google Scholar] [CrossRef]
- Mengelkamp, E.; Garttner, J.; Rock, K.; Kessler, S.; Orsini, L. Designing microgrid energy markets: A case study: The Brooklyn Microgrid. Appl. Energy 2018, 210, 870–880. [Google Scholar] [CrossRef]
- Umar, A.; Kumar, D.; Ghose, T. Blockchain-based decentralized energy intra-trading with battery storage flexibility in a community microgrid system. Appl. Energy 2022, 322, 119544. [Google Scholar] [CrossRef]
- Zeng, X.; Liu, L.; Leung, S.; Du, J.; Wang, X.; Li, T. A decision support model for investment on P2P lending platform. PLoS ONE 2017, 12, e0184242. [Google Scholar] [CrossRef] [PubMed]
- Khorasany, M.; Gazafroudi, A.S.; Razzaghi, R.; Morstyn, T.; Shafie-khah, M. A framework for participation of prosumers in peer-to-peer energy trading and flexibility markets. Appl. Energy 2022, 314, 118907. [Google Scholar] [CrossRef]
- Azim, M.I.; Tushar, W.; Saha, T.K. Cooperative negawatt P2P energy trading for low-voltage distribution networks. Appl. Energy 2021, 299, 117300. [Google Scholar] [CrossRef]
- PankiRaj, J.S.; Yassine, A.; Choudhury, S. An Auction Mechanism for Profit Maximization of Peer-to-Peer Energy Trading in Smart Grids. Procedia Comput. Sci. 2019, 151, 361–368. [Google Scholar] [CrossRef]
- López-García, D.A.; Torreglosa, J.P.; Vera, D. A decentralized P2P control scheme for trading accurate energy fragments in the power grid. Int. J. Electr. Power Energy Syst. 2019, 110, 271–282. [Google Scholar] [CrossRef]
- Han, D.; Zhang, C.; Ping, J.; Yen, Z. Smart contract architecture for decentralized energy trading and management based on blockchains. Energy 2020, 199, 117417. [Google Scholar] [CrossRef]
- Görgülü, H.; Topcuogla, Y.; Yaldiz, A.; Gökcek, T.; Erding, O. Peer-to-peer energy trading among smart homes considering responsive demand and interactive visual interface for monitoring. Sustain. Energy Grids Netw. 2022, 29, 100584. [Google Scholar] [CrossRef]
- Liu, Y.; Ma, H.; Jiang, Y.; Li, Z. Learning to recommend via random walk with profile of loan and lender in P2P lending. Expert Syst. Appl. 2021, 174, 114763. [Google Scholar] [CrossRef]
- Taleizadeh, A.A.; Safaei, A.Z.; Bhattacharya, A.; Amjadian, A. Online peer-to-peer lending platform and supply chain finance decisions and strategies. Ann. Oper. Res. 2022, 315, 397–427. [Google Scholar] [CrossRef]
- Zhou, Y.; Wu, J.; Song, G.; Long, C. Framework design and optimal bidding strategy for ancillary service provision from a peer-to-peer energy trading community. Appl. Energy 2020, 278, 115671. [Google Scholar] [CrossRef]
- Kong, K.G.H.; Lim, J.Y.; Leong, W.D.; Ng, W.P.Q.; Teng, S.Y.; Sunarso, J.; How, B.S. Fuzzy optimization for peer-to-peer (P2P) multi-period renewable energy trading planning. J. Clean. Prod. 2022, 368, 133122. [Google Scholar] [CrossRef]
- Zhou, W.; Wang, Y.; Peng, F.; Liu, Y.; Sun, H. Distribution network congestion management considering time sequence of peer-to-peer energy trading. Int. J. Electr. Power Energy Syst. 2022, 136, 107646. [Google Scholar] [CrossRef]
- Chen, Y.; Lei, X.; Yang, J.; Zhong, H.; Huang, T. Decentralized P2P power trading mechanism for dynamic multi-energy microgrid groups based on priority matching. Energy Rep. 2022, 8, 388–397. [Google Scholar] [CrossRef]
- Kanakadhurga, D.; Prabaharan, N. Demand response-based peer-to-peer energy trading among the prosumers and consumers. Energy Rep. 2021, 7, 7825–7834. [Google Scholar] [CrossRef]
- Suryono, R.R.; Purwandari, B.; Budi, I. Peer to Peer (P2P) Lending Problems and Potential Solutions: A Systematic Literature Review. Procedia Comput. Sci. 2019, 161, 204–214. [Google Scholar] [CrossRef]
- Xu, S.; Zhao, Y.; Li, Y.; Zhou, Y. An iterative uniform-price auction mechanism for peer-to-peer energy trading in a community microgrid. Appl. Energy 2021, 298, 117088. [Google Scholar] [CrossRef]
- Lin, J.; Pipattanasomporn, M.; Rahman, S. Comparative analysis of auction mechanisms and bidding strategies for P2P solar transactive energy markets. Appl. Energy 2019, 255, 113687. [Google Scholar] [CrossRef]
- Zheng, B.; Wie, W.; Chen, Y.; Wu, Q.; Mei, S. A peer-to-peer energy trading market embedded with residential shared energy storage units. Appl. Energy 2022, 308, 118400. [Google Scholar] [CrossRef]
- Yahaya, A.S.; Javaid, N.; Alzahrani, F.A.; Rehman, A.; Ullah, I.; Shahid, A.; Shafiq, M. Blockchain Based Sustainable Local Energy Trading Considering Home Energy Management and Demurrage Mechanism. Sustainability 2020, 12, 3385. [Google Scholar] [CrossRef]
- Lohachab, A.; Garg, S.; Kang, B.H.; Amin, M.B. Performance evaluation of Hyperledger Fabric-enabled framework for pervasive peer-to-peer energy trading in smart Cyber–Physical Systems. Future Gener. Comput. Syst. 2021, 118, 392–416. [Google Scholar] [CrossRef]
- An, J.; Hong, T.; Lee, M. Development of the business feasibility evaluation model for a profitable P2P electricity trading by estimating the optimal trading price. J. Clean. Prod. 2021, 295, 126–138. [Google Scholar] [CrossRef]
- Wu, Y.; Cimen, H.; Vasquez, J.C.; Guerrero, J.M. P2P energy trading: Blockchain-enabled P2P energy society with multi-scale flexibility services. Energy Rep. 2022, 8, 3614–3628. [Google Scholar] [CrossRef]
- Zhang, M.; Eliassen, F.; Taherkordi, A. Demand–Response Games for Peer-to-Peer Energy Trading With the Hyperledger Blockchain. IEEE Trans. Syst. Man Cybern. Syst. 2021, 52, 19–31. [Google Scholar] [CrossRef]
- Zhang, C.; Yang, T.; Wang, Y. Peer-to-Peer energy trading in a microgrid based on iterative double auction and blockchain. Sustain. Energy Grids Netw. 2021, 27, 100524. [Google Scholar] [CrossRef]
- Zhang, H.; Zhao, H.; Liu, Q.; Chen, E.; Huang, X. Finding potential lenders in P2P lending: A Hybrid Random Walk Approach. Inf. Sci. 2018, 432, 376–391. [Google Scholar] [CrossRef]
- Suryono, R.R.; Budi, I.; Purwandari, B. Detection of fintech P2P lending issues in Indonesia. Heliyon 2021, 7, e06782. [Google Scholar] [CrossRef] [PubMed]
- W3Schools. Python Random Module. Available online: https://www.w3schools.com/python/module_random.asp (accessed on 15 March 2022).
- Sukma, N.; Leelasantitham, A. A community sustainability ecosystem modeling for water supply business in Thailand. Front. Environ. Sci. 2022, 10, 940955. [Google Scholar] [CrossRef]
- Sukma, N.; Leelasantitham, A. From conceptual model to conceptual framework: A sustainable business framework for community water supply businesses. Front. Environ. Sci. 2022, 10, 1013153. [Google Scholar] [CrossRef]
- Wu, Z.; Wang, J.; Zhong, H.; Gao, F.; Pu, T.; Tan, C.; Chen, X.; Li, G.; Zhou, M.; Xia, Q. Sharing Economy in Local Energy Markets. J. Mod. Power Syst. Clean Energy 2023, 11, 714–726. [Google Scholar] [CrossRef]
Energy Source | Unit Price (Baht/kWh) |
---|---|
Solar Energy | |
Rooftop | 2.20 |
On ground | 2.16 |
Wind Energy | 3.10 |
Waste | |
VSPP | 5.08 |
SPP | 3.66 |
Biomass | 2.79 |
Biogas | 3.57 |
Consumer Type | No. | Unit Range | Unit Price (THB/kWh) | Monthly Fee (THB) | Average Unit Price with Monthly Fee (THB) |
---|---|---|---|---|---|
Group 1 <150 kWh/Month | 1.1 | 1–15 | 2.3488 | 8.19 | 2.5417 |
1.2 | 16–25 | 2.9882 | 8.19 | 3.3689 | |
1.3 | 26–35 | 3.2405 | 8.19 | 3.6464 | |
1.4 | 36–100 | 3.6237 | 8.19 | 3.6813 | |
1.5 | 101–150 | 3.7171 | 8.19 | 3.7947 | |
1.6 | 151–400 | 4.2218 | 8.19 | 4.2388 | |
1.7 | More than 400 | 4.4217 | 8.19 | 4.4661 | |
Group 2 >150 kWh/Month | 2.1 | 1–150 | 3.2484 | 38.22 | 3.2717 |
2.2 | 151–400 | 4.2218 | 38.22 | 4.2392 | |
2.3 | More than 400 | 4.4217 | 38.22 | 4.5172 |
Researchers | Highlights | Methods | Results | Limitations |
---|---|---|---|---|
Wongsamerchue et al. [5] | Double auction prepaid-trading | Double auction | Case study | Four-peer case study |
Baig et al. [6] | Microgrid energy markets | Implementation | Trading platform | Remote area; no pricing optimization |
Bandara et al. [7] | Neighborhood energy trading | Double auction | Trading algorithm | No pricing optimization |
Khorasany et al. [8] | Participation of prosumers | Proof of location | Cost reduction by 17.09% | Needs smart meter and pricing model |
Park et al. [9] | Building suitability | Building capability | guidelines | Energy management |
Huang et al. [10] | Energy trading model | ADMM | Decrease 5.11% | No pricing optimization o confidential trade |
Wongthongtham et al. [11] | Increasing scalability | Scalability | Scalability | No technical details Implementation and pricing mechanism |
Dorahaki et al. [12] | Energy trading model | Discount impact | Satisfaction of user | Only win–win situation |
Hu et al. [13] | High level of efficiency | Maximize profits | Power sharing | Pricing optimization |
Leong et al. [14] | Considering physical constraints | Game theory | Bidding strategy | Not Include Economic Pricing |
Esmat et al. [15] | Ant colony optimization | Ant colony | Efficient market solution | Uncertain prosumer commitment |
Mengelkamp et al. [16] | Balancing supply and demand | Framework | Case study | No pricing optimizatoin or confidential trade |
Umar et al. [17] | Energy trading with battery storage | Framework | Self-sustainability | Hourly based trading; no auction mechanism |
Zeng et al. [18] | Model for investment | Logistic classification | Higher efficiency | No auction or pricing optimization |
Khorasany et al. [19] | Anonymous proof of location | Distribution system | Lightweight FW | No pricing mechanism |
Azim et al. [20] | Voltage regulation | Coalition game | Feasibility | Small-sized prosume; no pricing optimizatoin |
Researchers | Highlights | Methods | Results | Limitations |
---|---|---|---|---|
PankiRaj et al. [21] | Profit maximization | Sealed bid auction | High profit return | Single side auction with no confidential condition |
López-García et al. [22] | Power balance in grid | Splitting the energy | Scalability | Accuracy and efficiency of devices |
Han et al. [23] | Energy trading and management | Balancing profits | Efficiency | Need to develop platform |
Görgülü et al. [24] | Energy management system | Priority matching | Domestic models | Only smart home applied |
Liu et al. [25] | Profile of loan and lender | P2P lending | Effectiveness | No upper and lower limit price |
Taleizadeh et al. [26] | Finance decisions and strategies | P2P lending | Optimal strategies | Deterministic demand and financial SC |
Zhou et al. [27] | Optimal bidding strategy | Residual balancing | Critically reviewed | No auctionor confidential mechanism |
Kong et al. [28] | Fuzzy optimization | Fuzzy sets | Carbon emission reducion of 61% | Single-step trading mechanism |
Zhou et al. [29] | Congestion management | Cost allocation | Profit increase | No auction or confidential mechanism |
Chen et al. [30] | Mechanism for dynamic multi-energy | IMMGS | Effectiveness | No auction and confidential mechanism |
Kanakadhurga et al. [31] | Demand response-based P2P | Particle swarm | Cost reduction | Confidential trading needed |
Suryono et al. [32] | P2P lending issues in Indonesia | None | Solutions | Literature review |
Xu et al. [33] | Auction mechanism | AMSA | Saving costs | Uniform clearing price |
Research | P2P | Energy | Trading Platform | Auction Method | Pricing Optimization | Demand and Supply Economic Law | Confidential Trading | Benefits and Contribution to SESCI |
---|---|---|---|---|---|---|---|---|
[6,8,12,13,17,18,19,20] | √ | √ | √ | - | - | - | - | - |
[5,7,9,10,11,14,16,27] | √ | √ | √ | √ | - | - | - | - |
[25,26,28,29,30] | √ | √ | √ | - | √ | √ | - | - |
[22,23,24,31] | √ | √ | - | √ | √ | - | - | - |
[33,34,38,41] | √ | √ | √ | - | √ | √ | - | - |
[37,44] | √ | - | - | - | √ | √ | √ | - |
This work | √ | √ | √ | √ | √ | √ | √ | √ |
Demand | Supply | |||||
---|---|---|---|---|---|---|
THB/kWh | House A | House B | Total | House C | House D | Total |
5.00 | 1.50 | 0.50 | 3.00 | 8.50 | 14.00 | 20.00 |
4.00 | 3.00 | 3.00 | 7.00 | 6.50 | 11.00 | 15.00 |
3.00 | 4.50 | 7.50 | 11.00 | 5.00 | 7.50 | 10.00 |
2.00 | 7.00 | 9.00 | 15.00 | 4.50 | 3.00 | 5.00 |
1.00 | 10.00 | 10.00 | 19.00 | 2.00 | 0.50 | 0.00 |
Buying confidence | 0.86 | 0.87 | 0.88 | 0.89 | 0.90 | 0.91 | 0.92 | 0.93 | 0.94 | 0.95 |
Buying success | 322 | 326 | 390 | 325 | 432 | 497 | 671 | 568 | 699 | 910 |
All bidding | 410 | 420 | 490 | 400 | 530 | 610 | 800 | 670 | 810 | 1030 |
Unsuccessful | 21.46 | 22.38 | 20.41 | 18.75 | 18.49 | 18.52 | 16.13 | 15.22 | 13.70 | 11.65 |
AR | 78.54 | 77.62 | 79.59 | 81.25 | 81.51 | 81.48 | 83.88 | 84.78 | 86.30 | 88.35 |
Sales confidence | 0.86 | 0.87 | 0.88 | 0.89 | 0.90 | 0.91 | 0.92 | 0.93 | 0.94 | 0.95 |
Sales success | 325 | 392 | 354 | 388 | 259 | 160 | 251 | 278 | 224 | 162 |
All bidding | 370 | 450 | 400 | 440 | 280 | 180 | 280 | 310 | 250 | 180 |
Unsuccessful | 12.16 | 12.89 | 11.50 | 11.82 | 10.71 | 11.11 | 10.36 | 10.32 | 10.40 | 10.00 |
AR | 87.84 | 87.11 | 88.50 | 88.18 | 89.29 | 88.89 | 89.64 | 89.68 | 89.60 | 90.00 |
Existing Processes of the Electrical Supply Chain | Beneficially Practical Implications of Economic Pricing in P2P Electrical Trading for Sustainable Electricity Supply Chain Industry in Thailand | ||
---|---|---|---|
Social | Environment | Economics | |
(1) Fuel procurement | Participation in regulating electrical energy prices. | Lowering fuel transportation or restricting gas pipeline installations helps diminish the overall environmental impact on the surrounding areas. | Reducing expenses on importing fuel and natural gas. |
(2) Electricity production | The electrical power generation from PV system is simple and user-friendly, empowering communities to self-educate and install it themselves. | Using clean energy for power generation decreases pollutant emissions. | Possible to develop a business that supplies equipment for PV system installations at the community level. |
(3) Electrical transmission system | Flexibility in the installation and utilization of local microgrids helps diminish reliance on centralized electricity transmission. | Implementing community microgrid systems diminishes the demand for nationwide transmission system installations, leading to less intrusion into forested areas and mitigated environmental impacts. | Microgrids improve the electrical system’s stability and dependability, resulting in less compensations due to power outages and blackouts in the local area. |
(4) Electricity transportation in the distribution system | There is adaptability in handling energy management at the community level, contributing to a decrease in reliance on central energy management. | Clean energy-produced electricity, devoid of pollutants, contributes to establishing a Green Community identity. | By deploying microgrids, electricity losses in the transmission system are reduced, allowing for the highest possible revenue from electricity sales. |
(5) Electrical retail | Utilizing peer-to-peer (P2P) platforms can enhance the reputation and support the sustainability of community-based electricity sales business. | It involves creating a sustainable awareness that encourages community members to understand the importance of environmental conservation, pollution reduction, and mitigating the impact of the greenhouse effec on both the local populace and the entire nation. | Establishing electricity prices that are fair and appropriate for stakeholder based on the supply and demand economic law. Income from selling electricity will be directed back into the community to enhance local economic growth. |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Leelasantitham, A.; Wongsamerchue, T.; Sukamongkol, Y. Economic Pricing in Peer-to-Peer Electrical Trading for a Sustainable Electricity Supply Chain Industry in Thailand. Energies 2024, 17, 1220. https://doi.org/10.3390/en17051220
Leelasantitham A, Wongsamerchue T, Sukamongkol Y. Economic Pricing in Peer-to-Peer Electrical Trading for a Sustainable Electricity Supply Chain Industry in Thailand. Energies. 2024; 17(5):1220. https://doi.org/10.3390/en17051220
Chicago/Turabian StyleLeelasantitham, Adisorn, Thammavich Wongsamerchue, and Yod Sukamongkol. 2024. "Economic Pricing in Peer-to-Peer Electrical Trading for a Sustainable Electricity Supply Chain Industry in Thailand" Energies 17, no. 5: 1220. https://doi.org/10.3390/en17051220