#### Automated 1D Analysis of Stall Inception in Multistage Compressors

During the design process of a multistage compressor, engineers need to consider a wide variety of geometric and performance parameters. If a particular compressor design exhibits poor performance, the engineer can make geometric changes to compensate. One crucial parameter in this regard is the stall inception point and the corresponding surge margin of a specific design. The stall points for several speedlines make up the Stall Line of a compressor performance map (as in Figure 1).  This line designates the lowest flow rates a compressor can stably operate at. If there is an issue with the stall point of a given design, wouldn’t it be advantageous for the engineer to identify it as early as possible in the design process? SoftInWay has developed a methodology that can quickly predict the onset of stall using a 1D-solver approach.

While there are several definitions of stall that one could use, we employed a definition based on the total-static pressure ratio. According to this criterion, a compressor stage is said to be stalling if its total-static pressure ratio decreases as the mass flow rate through the compressor decreases. This type of definition of stall has been used previously in other studies [1]. Another way to state this is when the slope of the total-static pressure ratio speedline reaches zero, the stall point is reached.

To test for this stall criterion, an automated workflow was developed using AxSTREAM ION. Several operating points can be analyzed in a single workflow, each time checking if the stall criterion is met. Since the 1D analysis for a single operating point only takes a few seconds at the most, this allows all the speedlines (including the stall points) to be calculated within a few minutes. This offers significant time savings when compared to CFD analyses of the same compressor geometry. The conceptual workflow algorithm is illustrated below.

At the start of the workflow, two high-flow operating points can be analyzed and checked for stall. If stall is not reached, a lower flow rate is analyzed and the lowest flow rates are again checked for stall. This process continues until the stall criterion is met for at least one stage. For a more accurate prediction of stall, the step size between the analyzed flow rates can be reduced, and the previous process repeated. Once the stall point is found with the desired accuracy, the workflow can be executed again at different speedlines. The output of all these analyses is both the stall line of the given compressor, as well as the stage at which stall is occurring. Since the stage where stall is occurring is identified, design engineers can make more targeted design modifications.

To validate the 1D prediction of stall, a two-stage transonic axial compressor from literature was analyzed [2]. The stall predictions were compared between the 1D method, two separate CFD solvers (STAR CCM+ and ANSYS CFX), and the experimental results from literature. The compressor geometry and results of this analysis are indicated below.

Each analysis indicated that the second stage experienced stall first. The 1D solution had a lower stall mass flow rate prediction compared to the other methods, resulting in the largest discrepancy with the experimental results. If the loss models are tuned for a specific compressor type, this could help improve the 1D stall prediction. While the accuracy of the prediction may be limited compared to CFD approaches, the 1D approach still offers significant design benefit. This stall prediction can guide an engineer’s design modifications, as it can be generated much more quickly compared to CFD predictions. While it is still good design practice to verify the compressor’s performance using CFD, this verification can take place later on in the design process.

The 1D methodology allows for quick prediction of stall in multistage compressors. While the accuracy of this prediction is limited compared to CFD predictions, the 1D prediction can be made in a matter of minutes, allowing for more rapid design iterations earlier on in the design process.

To learn more about using AxSTREAM for multistage compressor design, schedule a meeting with our team at Info@SoftInWay.com.

References:

[1] S. M. Hipple, H. Bonilla-Alvarado, P. Pezzini, L. Shadle and K. M. Bryden, “Using Machine Learning Tools to Predict Compressor Stall,” Journal of Energy Resources Technology, vol. 142, July 2020.

[2] C. L. Ball, L. Reid and J. F. Schmidt, “End-Wall Boundary Layer Measurements in a Two-Stage Fan (NASA-TM-83409),” NASA, 1983.