Modelling spatial and temporal agent travel patterns for optimal charging of electric vehicles in low carbon networks

Abstract

Abstract--The ability to determine optimal charging profiles of electric vehicles (EVs) is paramount in developing an efficient and reliable smart-grid. However, so far the level of analysis proposed to address this issue lacks combined spatial and temporal elements, thus making mobility a key challenge to address for a proper representation of this problem. This paper details the principles applied to represent optimal charging of EVs by employing an agent-based model that simulates the travelling patterns of vehicles on a road network. The output data is used as a reliable forecast so an optimal power flow model can devise optimal charging scenarios of EVs in a local electrical network. The effectiveness of the model is illustrated by presenting a multi-day case study in an urban area. Results show a high level of detail and variability in EV charging when a present-day carbon fuel mix is compared to one with lower carbon intensity.Index Terms-Agent based transport modelling, coordination of distributed energy resources, distribution network operation, electric vehicle charging, local smart-grids, optimal power flow.

Spatial and temporal electric vehicle demand forecasting in central London

“…Having performed the multi-faceted analysis detailed above for a virtual city [1,2], assessing an area of a real city using real network layouts (both road and electricity networks) combined with simulated data of a synthetic population was deemed as the next step in this research.…”

“…In previous work [1,2] an agent-based model developed in the Repast Simphony toolkit to generate trips in a (made-up) urban environment has been used. A GIS shapefile was read containing the location of different types of buildings including homes, offices, shops and schools, as well as a road network connecting the neighbourhoods of the city.…”

“…To achieve this goal, it is elemental to accurately portray vehicle usage patterns. Once the detailed temporal and spatial energy demands of EVs are estimated it is possible to assess various charging scenarios with the overreaching goal of identifying what charging strategies are the most cost-effective to stakeholders [2]. EV load forecasting is mostly an unknown topic for power engineers since these are not very familiar with transport system studies.…”

A novel stochastic modelling approach for electric vehicle charging power and energy requirements

“…Excluding the possibility of charging throughout the day, as presented in [2], [3] is expected to result in overestimation of the charging power requirements for the late evening hours. An agent based EV model has been presented in [5] which uses random distributions to determine individual departure times as well as type and detailed location of the next activity in the city. For every individual agent, the model generates the total distances driven and energy consumed, however no detail about the statistical method used for the computation of these variables has been provided.…”

Estimating plug-in electric vehicle demand flexibility through an agent-based simulation model

“…In [8] the charging infrastructure in not included explicitly in the analysis, assuming all the vehicles will have access to a charging point. In this paper, the model developed in [9] is enhanced, including explicitly the interaction between driving and charging behaviour, and access to charging infrastructure. The temporal and spatial electricity demand resulting from this interaction is then used to estimate the charging flexibility potential which drivers can offer to the system operator.…”