Computational Fluid Dynamics for Centrifugal Compressors

In today’s world where “time is money,” each and every industry involving turbomachinery wants to deliver their high performance products in the quickest time possible. Computational fluid dynamics (CFD) replaces the huge number of testing requirements thus not only shortening the design cycle time, but also reducing development costs.

Today with advancements in computational resources, numerical methods, and the availability of commercial tools, CFD has become a major tool for the design phase of a project. With a large number of validations and bench markings available on the applicability of CFD for centrifugal compressors, it has become an indispensable tool for the aerodynamic designer to verify the design and understand the flow physics inside a compressor’s flow path. However, CFD is still computationally expensive and requires a high level of user-knowledge and experience to get meaningful results. CFD analysis can be performed with and without considering viscous effects of the flow. The inclusion of viscosity into the flow introduces additional complexities for choosing the most appropriate turbulence closure model. CFD however, has some limitations due to:

  • – Errors created during modeling where the true physics are not well-known and are very complex to model.
  • – Multiple approximation and model errors created during the calculation process (such as mesh resolution, steady flow assumption, turbulence closure, geometric approximation, unknown boundary profile etc.). These approximations impact the calculations of local values of vital parameters.

In CFD for example, if the 1D design is not accurate, (stage loading and blade diffusion factors etc.), then CFD cannot turn out a good design. It is critical to use a design tool such as AxSTREAM®  which can generate optimized designs with less time and effort starting from the specification.

The preliminary design modules of AxSTREAM® uses inverse design tasks to generate the initial flow path for the centrifugal compressor. By choosing the right combination of geometrical and design parameters from the start, AxSTREAM® reduces the number of design cycle iterations required in generating an accurate design.

This initial design obtained is further analyzed and optimized using throughflow solvers in AxSTREAM® which considers various operating conditions. The throughflow solvers in AxSTREAM®  predict the performance parameters at different sections and stations, and presents the blade loading, flow distribution along the flow path, etc.

The generation of 3D geometry for the impeller and diffuser is another complex activity which is greatly simplified by using the radial profiler and 3D blade design module in AxSTREAM®. The geometry generated in AxSTREAM® is fully parameterized with complete control for the user to modify as and when required. Figure 1 shows a parameterized impeller geometry generated using seven spanwise sections with contours of the curvature.

Parameterized Impeller Geometry
Figure 1. Parameterized Impeller Geometry

In CFD analysis of turbomachines, grid generation becomes a very challenging task due to the geometries of complicated, twisted blades. To achieve reliable CFD results, the grid must resolve the topology accurately to preserve this geometric information. The quality of the grid should be in an acceptable range especially the angle, aspect ratio, and skewness of the grid elements. Automatic mesh generation tools are employed to reduce the turbomachines meshing complications. The AxSTREAM® platform uses AxCFD™ to generate a high quality mesh in considerably short time which captures the accurate flow features.

Solver settings also have a great impact on the CFD analysis. Improper solver settings lead to needing more time to obtain a converged solution while also affecting the accuracy of the results. Sometimes improper settings even lead to the failure of the CFD analysis. CFD analysis is an iterative process, which progresses as time marches by. By using a higher time step, the solution convergence will be faster, but it may lead to the solution failure in certain circumstances.

Partial differential equation (PDE) order can also affect the solution accuracy. Generally, CFD analysis should start with first order PDEs and after getting the stable solution, it should change to second order PDEs to achieve a more accurate solution. Understanding CFD results requires great skill and experience. Interpretation of CFD results for unsteady flow is as difficult to decipher as experimental results and therefore need improved post processing techniques.

AxCFD™, in the AxSTREAM® platform, provides user an opportunity to perform CFD analysis by applying standard methods of full three-dimensional CFD, axisymmetric CFD (meridional) and blade-to-blade analysis. Figure 2 shows the results of CFD analysis from AxCFD™.

CFD Analysis Results using AxSTREAM
Figure 2. CFD Analysis Results using AxSTREAM

CFD has the potential to utilize the 3-dimensional nature of the flow to restrict undesirable features like strong secondary flows or corner separations in compressors, and can save time and cost during the early phases of the design process. The integration of CFD with the design tools like in AxSTREAM® provides users with an opportunity to develop optimized designs in the shortest possible time.

Read more about how AxSTREAM® can be used in your centrifugal compressor design project here or contact us for more information at 

Leave a Reply