Multi-Disciplinary Optimization of a Radial Compressor using Cadence CFD and Concepts NREC

Radial compressors, also known as radial fans or blowers, are primarily used for compression purposes. Radial blades attached to a rotating impeller draw air into the unit's center. They are well-suited for high-pressure applications, where their efficient design can save energy. Isentropic efficiency and blade loading are critical factors in the design of these compressors. A multiphysics approach, including both aerodynamic and structural objectives will ensure optimal results. Streamlining the design process and minimizing iterations can lead to practical, resource-efficient designs. This reduces time-to-market and enhances the overall efficiency of the design process. In this short case study, the optimization of a shrouded radial compressor considering both the aerodynamic stage performance and the structural integrity of the impeller, i.e., to maximize isentropic efficiency and reduce von Mises stress, is performed using Cadence Fidelity Turbomachinery suite and Concepts NREC.

Description

For this study, a single-stage compressor with a shrouded impeller, a vaneless diffuser, and no flow collector has been selected, as illustrated in the figure below. Here, the objective is to attain the following design conditions.

• Minimize the von Mises stresses or blade loading due to centrifugal forces.
• Increase the total isentropic efficiency.

3D Stage Geometry (left) and Meridional View of the Stage Geometry (right).

Optimization Methodology

Optimization Workflow

In this study, a surrogate-based optimization methodology is adopted whereby the workflow is initiated by creating a parametric geometry model, i.e., the geometry is defined by a set of parameters. Subsequently, a subset of free parameters is defined. As a next step in this workflow, the design of experiments (DOE) is conducted to explore the design space, and the results from these experiments are evaluated and summarized in a database. Further, a surrogate model is created using this database that can predict the efficiency and total pressure ratio based on the input parameters.

Optimization Workflow

Single or multiple candidates are generated by optimizing this surrogate model. These candidates further undergo computational fluid dynamics (CFD) and computational structural mechanism (CSM) analysis. Results from these analyses are fed into the database. This loop continues until the objective or convergence criteria are met and a new optimized design is generated.

Combination of Different Optimization Methods

von Mises Stress vs. Isentropic Efficiency – Database, Multi-Objective Optimisation, Single-Objective-Optimisation.

To attain the intended results, this study effectively utilizes a variety of optimization techniques. After the DOE step in the workflow, a database for the surrogate model is created, which is cross-validated using the leave-one-out analysis. Using this analysis, the accuracy of the surrogate model is confirmed before optimization is initiated. At first, a multi-objective optimization focusing on objectives such as efficiency and von Mises stresses in feasible ranges is explored. As a second step in the optimization workflow, single objective optimization fixated on stage efficiency as the objective and von Mises stress (set to maximum) as the constraint.

Tools Used in the Design Process

Tools Used in the Optimization Workflow

Geometric Design and Meshing

The generation of the parametric model is carried out in AxCent. It is a powerful tool for detailed turbomachinery 3D geometric design. This tool enables the user to conduct a preliminary flow and stress analysis. The parametric model for this optimization has 27 free parameters and 50 mathematical equations that can link different parameters or define complex quantities.

AxCent is also used for CSM meshing and simulation using a module called push-button FEA, a fully integrated stress analysis tool allowing users to do both structural and aerodynamic analysis simultaneously.

For CFD analysis, a structured mesh was generated using Fidelity Autogrid. It works with predefined topologies for different turbomachinery configurations. It encompasses advanced smoothing algorithms and scripting capabilities.

Solver

As mentioned in the geometric design and meshing section, the CSM simulation is carried out in AxCent, and the CSM setup is as follows:

• Single passage model – fillets included
• Unstructured grid with relative cell size
• One operating point
  • Material used - Austenitic Stainless Steel (300 series)
  • Rotation Speed: = 1.15
  • Material Properties at = 100°

For CFD simulation, the mesh obtained from Fidelity Automesh Autogrid is plugged into Fidelity Flow (Fine/Turbo), the fastest structured solver on the market, offering advanced options for rotor-stator interfaces, convergence acceleration, and with full and batch scripting capabilities. The CFD setup is as follows:

• Single passage model
• Structured grid
• Spalart–Allmaras turbulence model
• Wall function
• Fluid: 2, ideal gas
• Boundary conditions: inlet total pressure, inlet temperature, and outlet mass flow
• Three operating points along a one-speed line

Results and Conclusion

Maximum von Mises Stress and Total Isentropic Efficiency for five designs.

Based on the plotted data, it is clear that the v3 design is the most optimized, striking a good balance between objectives. This design boasts minimal von Mises stresses, high efficiency, and a satisfactory safety limit. To summarize, the total efficiency is increased by 0.6 %, while the stresses from centrifugal forces are decreased by more than 50%.

This case study clearly demonstrates that integrating different disciplines into one optimization process is essential for achieving practical designs. Cadence CFD and Concepts NREC offer effective tools to streamline this process. When facing conflicting objectives, implementing multi-objective optimization can help identify the most optimal solution.

---

Watch the webinar on Navier & Stokes vs. von Mises: Optimisation of a Radial Compressor to learn more about optimizing a radial compressor using Cadence CFD and Concepts NREC.