'This model is bottom-up and more realistic'
SparkCity is not a real city, but it is a realistic one. It forms the basis of a model that allows you to choose what developments can be expected for electric transport given certain parameters. This will enable authorities, grid operators and market parties to acquire a reliable picture of what they will need to fine-tune their policy and strategy to.
Look, there’s the family number 94: two working adults with one child in daycare. Mum has a job outside of the city. The family has noticed that the price of an electric car has fallen over the past couple of years and that you can now drive 400 km on a single charge. They head to the dealer to trade in their petrol car.
There are another couple of thousand households like family number 94, each with their own composition, needs and considerations. Together they form SparkCity – an imaginary city full of realistic households. It forms the basis of a model that Auke Hoekstra from Eindhoven University of Technology has been working on for two years. This model enables governments, businesses and other parties to look at the effect on people’s behaviour based on numerous parameters. A variety of factors is considered, such as income, energy price, presence of charging facilities, and so on. Thus enabling them to estimate what kind of charging infrastructure for electric vehicles is required in a neighbourhood or what the effect of electric charging will be on the grid.
It is a unique model: agent-based. In contrast to a mathematical model, which will endeavour to calculate a general optimum using the general data input, SparkCity comprises thousands of objects each performing their own calculations on the basis of their individual parameters. They weigh things up and make decisions almost like real people. ‘The advantage of this model is that it is less abstract than mathematical models. We work with recognisable objects: families, cars, charging stations, substation, et cetera’, says Hoekstra. ‘An agent-based model is bottom-up and more realistic. We work with real families. You can zoom in on a “family”, an object, enabling you to examine whether a follow-up step is logical. All follow-up steps collectively determine what is done overall.’ He continues: ‘Such models used to be far too laborious. However, today's fast computers will take care of all the calculation for you, meaning that reality can be approximated much more closely’.