Data assimilation for energy storage to improve operational control

Project description

With the increasing integration of renewable energy into the global power mix as part of the energy transition, the guarantee of a stable power supply is becoming increasingly difficult. The reason for this is the strong fluctuations and the difficult predictability of the production output of renewable energies, such as photovoltaics, wind energy or hydropower. 

Partial remedies promise energy storage facilities, such as hydroelectric plants at reservoirs, compressed air storage or power-to-gas plants. But that will not be enough. Therefore, decentralized storage technologies are necessary, such as battery systems or thermal storage systems that can buffer the process heat demand in industrial plants or the heat demand in housing estates. 

One promising new technology is currently under investigation: the storage of heat from hot air in a technical tank filled with a porous bed of slaked lime (calcium hydroxide – Ca(OH)2). When the system is charged with hot air, the chemical balance of lime is shifted, absorbing large amounts of energy. When charged with colder, damp air, the balance shifts back, the energy is released in the form of heat again.

 In order to make this technique manageable, simulations of the operation are necessary, among other things. Corresponding models and numerical methods are currently being developed at LH2, supported by experiments on an industrial scale at DLR. However, this modeling is an extremely great challenge: it is a non-isothermal flow of a multicomponent gas mixture, chemical reaction between gas and solid phase with energy turnover, coupled with a structural change of the porous bed by the chemical reaction - and in a three-dimensional dynamic and heterogeneous system. Therefore, forecasts of the internal state (charge status, temperature, health status) of the technical system are either highly error-prone or, in the case of extremely complex modeling and high numerical resolution, not real-time capable. For a meaningful control (model-predictive control) of the system, however, reliable simulation-based forecasts are necessary.

To the top of the page