Project
Robust 3D Rotor Design of Future Centrifugal Compressors for Modern Energy Conversion Processes
The political demand for a massive expansion of renewable energy conversion has a major impact on future power plant capacities, but also on the existing power plant fleet, as it has to be adapted to the volatile nature of renewable energy sources. Traditional thermal power plants will remain indispensable in the future to compensate for fluctuations in the volatile energy sources. In addition to operational and fuel flexibility, the plants and their components have to be highly efficient in order to meet the high price of electricity generated from renewable fuels. The investigations in this project serve to advance the development of new compressor technologies that contribute to the decarbonization of the industrial sector. In the future, high-pressure centrifugal compressor stages will play a key role in decarbonized energy conversion processes involving renewable sources. These technologies will support next-generation turbocharged piston engines – ideally powered by greenhouse gas-neutral synthetic fuels or, alternatively, by green hydrogen – as well as turbocharged fuel cell systems. Accordingly, this project focuses on optimizing robust high-pressure centrifugal compressors tailored for use in hydrogen-powered piston engines.
For this purpose, a compressor stage with a new 3D impeller is being developed, where both the blade geometry and the hub base are designed using 3D freeform surfaces. The downstream diffuser is designed and optimized in close coordination with the impeller design. Additionally, map-stabilizing measures, such as an inlet recirculation cavitiy (IRC), are planned for the design. The development follows state-of-the-art numerical optimization methods. Furthermore, a resonance-safe design is to be ensured to prevent high-cycle fatigue (HCF). This requires knowledge of the dynamic loads induced in the impeller blades and their supporting disk, primarily due to aerodynamic interactions between the rotating impeller and the stationary bladed diffuser through fluid-structure interaction (FSI) simulations. A detailed description of the individual sub-projects is provided in the following sections.
Project Details
Project term
May 1, 2025–April 30, 2026
Affiliations
RWTH Aachen University
Institute
Institute of Jet Propulsion and Turbomachinery
Principal Investigator
Methods
For the aerodynamic optimization workflow, the steady-state Reynolds-averaged Navier-Stokes (RANS) equations are solved for each investigated design at different operating points (OP). Furthermore, it is ensured that each design meets sufficient safety requirements regarding static structural loads. Therefore, the calculation of deformations and the resulting stresses is performed. The overarching gradient-free optimization method is a Genetic Algorithm (GA). Since these methods commonly require the evaluation of a large number of designs to find the global optimum, the optimizer utilizes meta-models for the RANS simulations. These surrogates are trained on a dataset and progressively refined during the optimization process, improving their prediction accuracy. The developed FSI toolchain initially requires a structural modal analysis using a Finite Element Method (FEM) solver to determine the natural mode shapes and the corresponding frequencies. To identify aerodynamic excitations, the RANS equations are then solved in a frequency-based manner using the Nonlinear Harmonic (NLH) method. This approach enables the resolution of interactions between the rotating impeller and the stationary diffuser without solving the computationally very expensive unsteady RANS equations (URANS). Finally, aerodynamic damping is also determined through frequency-based CFD simulations, where a mesh deformation algorithm allows the structure to oscillate in its corresponding natural mode shape.
Results
The number of free parameters in the aerodynamic optimization was increased progressively. To the end, two separate optimizations with different parameterizations were performed, compared, and subsequently combined. One optimization was carried out using 23 meridional geometry parameters and three blade parameters. The resulting design achieved a maximum efficiency increase of 0.25% while satisfying the constraints on pressure ratio, map width, and structural integrity. Under the same constraints, an optimization using only 29 free 3D blade parameters achieved an efficiency improvement of 0.56%. This confirms the significant efficiency potential of full 3D blade shaping. A coupled optimization with a total of 72 free parameters is currently in progress. In addition to combining the two parameterizations, the number of free 3D blade parameters was increased to 49, providing the optimizer with maximum freedom for 3D blade shape optimization.
Contrary to the original project plan, the aerodynamic optimization workflow was extended in consultation with the industrial partner to enable the inclusion of an IRC directly within the optimization process. This extension was considered necessary by the industrial partner to meet the stringent requirements for sufficient compressor map width at higher rotational speeds. Initial optimization test runs with integrated IRC resulted in a high failure rate (>50%) of evaluated designs due to numerical stability issues. These problems have been significantly reduced through the mitigation of parameter interactions, investigations into the selection of numerical solver settings, and the implementation of a modified mesh topology. Therefore, an optimization including an IRC with a total of 72 free parameters is being carried out in parallel with the optimization without an IRC. The configuration without an IRC is intended to serve as a fallback solution should the numerical stability issues not be resolved satisfactorily.
Discussion
In the next project phase, the focus will be on completing and extending the optimization studies. First, the partially completed optimizations with and without IRC will be finalized. To account for the project goals, a machine with reduced mass flow (flow cut) will be computed for the same parameterization, both without and with IRC. Finally, an extended parameterization including additional degrees of freedom for the diffuser blades, will be optimized in order to further investigate and demonstrate the advantages of a fully 3D blade design. The best-performing optimized designs of each configuration will be assessed for HCF using the developed FSI toolchain to ensure stable and robust operation.
Additional Project Information
DFG classification: 404-04 Hydraulic and Turbo Engines and Piston Engines
Software: Cadence Fine Turbo, ANSYS
Cluster: CLAIX