The construction and dimensioning of new power lines is a science in itself. As a result of the development of renewable energy, for example, a large number of decentralized energy sources are now connected to the grid. To stabilize operation of the grid, the output of power stations, wind turbines and photovoltaic plants must repeatedly be restricted, reports the Jülich Research Center . Nevertheless, around 5% of the total wind energy generated was lost in 2017 alone as a result of problems with feed-in management. The development of the networks is an opportunity to optimize the system.
TenneT is planning to simulate and optimize the load flows over a whole year so that it can then adapt and design new lines to meet demand. Conventional servers would not be suitable for this task, however, because of the data volumes and complexity. The experts at the J ülich Supercomputing Centre have therefore adjusted the TenneT applications for parallel processing by means of a multitude of processors and adapted a parallel computer with smart resource management to them. And with some success: the simulation of load flows in the power grid has been accelerated by a factor of over 30, according to the project partners.
According to the German Federal Network Agency, the total costs for grid stabilization in 2017 were €1.4 billion – well above the high of 2015, when €1.1 billion had to be invested. The costs for feed-in management have increased in particular. In the context of the energy transition, this illustrates that there “is still an urgent need to develop the grid”.