Multi-Agent Based Simulation Framework for Evaluating Digital Energy Solutions and Adoption Strategies

Kristoffer Christensen*

*Corresponding author for this work

Research output: ThesisPh.D. thesis

291 Downloads (Pure)


The increasing digitalization of the electricity grid into a Smart Grid implies theintroduction of an increasing number of new digital energy solutions to the energyecosystems. Digital energy solutions are expected to solve problems occurring, especiallydue to the transition to carbon neutrality. This is especially the case for sector couplingbetween the electricity grids and the transportation sector. However, the electrification ofthe transportation sector also introduces challenges to the energy system. The challengesrelate to many ecosystem stakeholders, and the consequence of mass electric vehicleadoption is unclear. Furthermore, there is no sufficient method to identify and evaluatedifferent digital energy solutions, such as electric vehicle charging algorithms, and theirimpacts on the energy ecosystem.

To address these challenges, this research proposes and develops an integratedframework consisting of three main parts: ecosystem modeling and multi-dimensionalecosystem impacts analysis, multi-agent based modeling and simulation, and multi-criteriadecision-making for: 1) investigating the ecosystem dynamics with various scenarios ofelectric vehicle charging strategies and dynamic distribution tariffs, and 2) evaluating theimpacts of electric vehicle charging strategies and dynamic distribution tariffs on relatedstakeholders.

Four main methods are applied in this research to develop the framework: Businessecosystem modeling and CSTEP multi-dimensional ecosystem impact analysis; Multicriteria feasibility evaluation for assessing and identifying state-of-the-art technologies;Multi-agent based modeling and simulation; and Multi-criteria decision-making. A casestudy of a Danish radial distribution network with 126 residential consumers withincreasing electric vehicle adoption is chosen to verify and validate the developedframework.

Based on the developed multi-dimension, multi-criteria feasibility method, threedecentralized and four centralized electric vehicle charging strategies in the literature areidentified and selected for evaluation besides the traditional electric vehicle chargingstrategy. Furthermore, various state-of-the-art dynamic distribution tariffs are evaluated,and the Time-of-Use distribution tariffs are evaluated as the most feasible in Denmark.The next generation Danish tariff model, Tariff Model 3.0, is a type of Time-of-Usedistribution tariff.

From the multi-agent based simulation results, the main findings show that overloadoccurs in the grid in 2031 with the Traditional charging with a 67% electric vehicle adoption.Adopting other decentralized charging strategies results in faster and more frequentoverload, e.g., 2028 with the Real-Time Pricing strategy. Meanwhile, by applyingcentralized strategies, overloads can be avoided with a limited impact on electric vehicleusers’ charging experience (with a 71-78% share of electric vehicles in the grid) in 2032.Overloads are also avoided with a 100% electric vehicle adoption in 2039; with thecentralized strategies, electric vehicle users with electric vehicle models charging with 7.2kW or more can reach the desired charging level before departure.

The multi-criteria decision-making evaluation results show that cost-oriented andenvironment-oriented electric vehicle users will most likely adopt the Real-Time Pricingand Time-of-Use Pricing charging strategies with the Tariff Model 3.0 and hourly electricityprice scheme. The distribution system operator prefers the Traditional charging (under allprice structures) for the electric vehicle users and is most likely to adopt the Round Robinand First-Come-First-Serve centralized charging strategies.

The evaluation of the combined strategies shows that the best combination is with the Real-Time Pricing and Round Robin strategies. The Round Robin is the best centralized strategy to manage the Real-Time Pricing and Time-of-Use Pricing decentralized strategies. Combining Real-Time Pricing with Round Robin or First-Come-First-Serve significantly affects electric vehicle users. The users experience an increased frequency of not reaching their desired charging level before departure. The frequency increases by 42 times with Round Robin and 47 times with First-Come-First-Serve in 2032 (electric vehicle share of 71-78%) compared to Real-Time Pricing without centralized control. The combination of the Real-Time Pricing and Time-of-Use Pricing strategies results in overload in 2029 but with around 80% fewer overloads the year after (compared with the individual strategy experiments).

This research proposes a novel integrated framework that can capture the full ecosystem dynamics during a long-term period with a high resolution (hourly), therefore, a clear load profile of distribution grids in the future due to electric vehicle adoption can be captured. Meanwhile, this research introduces a systematic method that can be applied to investigating the impacts of increasing electric vehicle adoption or dynamic distribution tariffs in the energy system considering the ecosystem dynamics. Furthermore, the developed multi-agent based simulation can decode the complex behaviors and aggregated consequences due to the ecosystem dynamics. It is also applicable for investigating other future and what-if situations in similar energy ecosystems. Moreover, the multi-criteria decision-making model allows the evaluation of smart energy solutions and strategies considering individual stakeholders and ecosystem perspectives.
Original languageEnglish
Awarding Institution
  • University of Southern Denmark
  • Jørgensen, Bo Nørregaard, Principal supervisor
  • Ma, Zheng Grace, Co-supervisor
  • Demazeau, Yves, Co-supervisor
Date of defence21. Nov 2022
Publication statusPublished - 30. Sept 2022


Dive into the research topics of 'Multi-Agent Based Simulation Framework for Evaluating Digital Energy Solutions and Adoption Strategies'. Together they form a unique fingerprint.

Cite this