This paper describes the work done on the understanding of infrastructure interdependencies and impact on the overall system. The work on the model described in this report started in September 2021. Access to the data was given at the end of October 2021 and the technical work ran until mid-January 2022.
The work was led by Lars Schewe and primarily carried out by Mariel Reyes Salazar. The integration of the multiple different networks was carried out by Maksims Abalenkovs. We achieved to demonstrate that we can integrate the data from a digital twin into component networks models and could connect these with an overarching coordinating algorithm. This allows us to propagate failures in the networks and then analyse the impacts on the different networks. The observed runtimes for the test networks indicate that the implemented methods will work on realistic networks and that implementing more complex models is feasible in follow-up projects.
The technical work planned in the work package was to model each of the component networks, build models that allow to propagate failures through each of them, and propose methods to propagate the failures between them.
To structure the work, the team proposed three levels of detail for the network models and two levels for the integration. In addition, the objective functions for the underlying optimization problems were to be developed. Due to unavailability of data and the short timescale, it was decided to focus on the first levels for all networks and the integration. As no data was available that could guide the definition of an objective function, this work was not undertaken.
The basic models were implemented in Python and tested on a small-scale model of part of a UK town. This allowed to demonstrate that the overall methodology is sound and that data from a digital twin can be transferred to more detail network models and the results can be played back to the digital twin.