Project area C: Components of Carnot Batteries: Machines (2nd funding period)
Inverse Design and Physics‑Based Transient Modelling of Rotary Positive Displacement Machines under Dynamic Operating Conditions
The proposed project aims to develop a systematic methodology for the inverse design of energetically optimal rotary positive displacement machines (RPDMs) for prescribed load profiles of Carnot Batteries. For these compressors and expanders, an enhanced chamber‑model simulation will be established to enable physically consistent prediction of transient operating behaviour during start‑up, shut‑down, and varying operating conditions. Furthermore, a physics‑based reduced‑order model describing the transient behaviour of RPDMs will be developed, calibrated, and validated using results from the advanced chamber‑model simulations. The development of these methods and models builds upon the open-source tool created in the first funding phase for inverse design at a single operating point and on the established simulation loop consisting of a pre‑processor for automatic chamber‑model generation, a solver for fluid‑dynamic and thermodynamic calculations, a post‑processor for evaluation, and an optimiser for targeted modification of machine geometry. The open-source tool will be extended by a methodology for decomposing the load profile into segments and synthesising the machine geometries optimised for each segment. The simulation loop will be extended to include the analysis of RPDMs with respect to time‑dependent thermal storages (e.g. rotors and housing) as well as models for heat transfer between the working fluid and the structure. Based on this enhanced chamber‑model simulation, a reduced‑order model will be derived that incorporates the essential physical mechanisms relevant for RPDMs and describes not only the steady‑state operating behaviour but also, in particular, the transient behaviour of this kind of machine. The implemented methods and models will be validated through measurements on existing RPDM test rigs at the applicant’s chair and by partners within the Priority Programme. Both the methodology for load‑profile‑dependent inverse design and the validated physics‑based reduced models representing transient RPDM operation will be made available to partners within the Priority Programme as a software package (tool) for use in their transient thermodynamic cycle simulations.
Professor Dr.-Ing. Andreas Brümmer
Technische Universität Dortmund
Fachgebiet Fluidtechnik
Inverse design of turbomachines using transfer functions
Carnot batteries as energy storage systems are subject to dynamic operating conditions such as start-up, shut-down, and load-changes. For an economically viable design, these processes are systematically analyzed and adapted to the requirements of the electricity market. The transient behavior is influenced significantly by the turbomachinery within the process cycles. Within SPP 2403, transfer functions are developed in the first funding period that enable the design of turbomachinery and the assessment of their steady-state operation. Building on this, the transfer functions are extended to include start-up, shut-down, and load-change transients. To this end, simplified models for unsteady processes are derived from detailed models created for SPP 2403. Based on these models, parameter studies are conducted to quantify the influence of key design parameters on transient behavior and to derive robust design rules. The resulting rules are incorporated into the transfer function for start-up, shut-down, and load-changes. It complements the transfer functions developed in the first period and constitutes the outcome of the first work package. This enables both the prediction of the transient behavior of specific machine designs and the formulation of robust design requirements for market-ready Carnot batteries, which are then implemented in the designed machines. In the second work package, the application scope of turbomachinery in the context of Carnot batteries is expanded. Some of the working fluids under consideration exhibit overhanging dew lines in the T–s diagram. During compression, the working fluid can enter the two-phase region, leading to condensation in the compressor. While such operating conditions are often dismissed in the literature as not technically feasible, they are systematically investigated here for centrifugal compressors. To this end, numerical simulations are performed that capture both the non-equilibrium condensation of real gases and flow with a dispersed liquid phase. The effect of condensation on the compressor performance map is quantified and incorporated into the transfer functions in the form of a simplified model. This enables adaptation of the performance maps and a reliable, application-oriented design of compressor geometry.
Professor Dr.-Ing. Dieter Brillert
Universität Duisburg-Essen
Fakultät für Ingenieurwissenschaften
Lehrstuhl für Strömungsmaschinen
Inverse aerodynamic design of turbo components for Carnot batteries by means of physics informed networks enhanced by generative learning
We develop AI based simulation tools for fast design space exploration and inverse design of Carnot battery specific turbo-machinery components. We produce a major data set on the instationary flow within a Carnot battery turbine stage with typical working fluids and boundary conditions. To accelerate the simulation, we train generative AI World Models on the produced data and develop curriculum learning for stable training physics informed neural networks. Based on these accelerated simulation tools, we enable the generation of performance maps and inverse design procedures for Carnot battery turbines. Our design procedure includes redundancy and uncertainty quantification for hallucination free generation of instationary solutions, explainability of AI by visual impressions of the obtained flow fields. Our data and code will be published according to FAIR principles.
Professor Dr. Hanno Gottschalk
Technische Universität Berlin
Fakultät II - Mathematik und Naturwissenschaften
Institut für Mathematik
Professorin Dr. Francesca di Mare
Ruhr-Universität Bochum
Fakultät für Maschinenbau
Lehrstuhl für Thermische Turbomaschinen und Flugtriebwerke