HeX Lab

(This page was last updated on 2022 and is under construction. Please refer to Publication for the latest research.)

Contents

• Turbulence Modeling
1. A Turbo-Oriented Data-Driven Modification to the SA Turbulence Model
2. Delayed-Detached Eddy Simulation for Separated Compressor Flows
3. Parametric Uncertainty Quantification of RANS Turbulence Model
• Turbomachinery Aerothermodynamics
1. Flow Instabilities in Centrifugal Compressors
2. Transonic Flows in Centrifugal Compressors

---

Turbulence Modeling

A Turbo-Oriented Data-Driven Modification to the SA Turbulence Model

Spalart-Allmaras (SA) turbulence model is one of the most popular models for compressor aerodynamics. However, it often underpredicts the compressor stall margin. Such a deficiency mainly routes from the empirical assumptions made in the constitutive relation (e.g., linear eddy viscosity relation) and the transport equation (e.g., equilibrium turbulence). Previous researchers have proposed different flavors of the SA model to improve the model accuracy, including SA-QCR (quadratic constitutive relation), SA-RC (rotation and curvature correction), SA-H (helicity correction) and SA-PG (pressure gradient correction). From an evaluation study on both compressor flows and canonical flows, however, none of the flavors showed significant improment on compressor flows without deteriorating canonical flows.

In this research, we proposed a new flavor of the SA model namely SA-PG _ω . The modification is based on a dimensionless vortical pressure gradient term (i.e., PG _ω ). It effectively identified both the tip leakage blockage and the hub corner blockage; whereas in 2D flows and rotor-stator interface flows, the modification is automatically switched off. To calibrate the new model coefficients, Bayesian inference was adopted to give a robust coefficient set. The calibrated SA-PG _ω showed improved prediction on NASA Rotor 67, TUDa-GLR-OpenStage, BUAA Stage B Rotor and Lyon linear compressor cascade.




Related publication:

• He, X.*, Zhao, F., and Vahdati, M., “A Turbo-Oriented Data-Driven Modification to the Spalart-Allmaras Turbulence Model,” ASME Journal of Turbomachinery, 2022, 144(12), 121007. [preprint] [doi]
• He, X.*, Zhao, F., and Vahdati, M., “Evaluation of Spalart-Allmaras Turbulence Model Forms for a Transonic Axial Compressor,” 2020, GPPS Paper No. GPPS-CH-2020-0013. [preprint] [doi]

Advisor: Prof. Mehdi Vahdati

*Winner of the Katopodis Prize of Imperial College London (2023); finalist of ASME Turbo Expo Turbomachinery Committee Best Paper Award (2022).

Detached Eddy Simulation for Separated Compressor Flows

High-fidelity turbulence data are indispensable for training a data-drive RANS turbulence model. Hybrid RANS/LES methods (e.g., DES-type methods) are the most cost-effective for turbomachinery flows featured by high Reynolds numbers and Mach numbers. In this research, we reviewed the state-of-the-art upgrades of DES methods and identified a combination of upgrades for compressor tip leakage flow. The upgraded DES method is implemented in the in-house flow solver HADES, and then successfully validated against PIV measurement data of BUAA Stage B rotor.

Based on the database generated by the upgraded DES method, spectral proper orthogonal decomposition (SPOD) is then performed to analyze the modal component of the tip leakage flow. Large-scale coherent structures are identified in the tip blockage zone downstream of the tip leakage vortex breakdown. These structures play a crucial role in turbulence modeling of tip leakage flow and non-synchronous vibration of compressor blades.




Related publication:

• He, X.*, Zhao, F., and Vahdati, M., “Detached Eddy Simulation: Recent Development and Application to Compressor Tip Leakage Flow,” ASME Journal of Turbomachinery, 2022, 144(1), 011009. [preprint] [doi]
• He, X.*, Fang, Z., Rigas, G., and Vahdati, M., “Spectral Proper Orthogonal Decomposition of Compressor Tip Leakage Flow,” Physics of Fluids, 2021, 33(10), 105105. [preprint] [doi]

Advisor: Prof. Mehdi Vahdati, Dr. Georgios Rigas

Parametric Uncertainty Quantification of RANS Turbulence Model

RANS turbulence models are deficient in predicting compressor near-stall flows due to the uncertainties involved in the model development procedures. In this research, the parametric uncertainty of the Spalart-Allmaras turbulence model is investigated in canonical flows and NASA Rotor 67. An efficient metamodel-based UQ framework is established using the artificial neural network.

For the transonic axisymmetric bump and the backward-facing step cases, re-calibration of the model coefficients lead to contradictory results when considering multiple flow quantities of interest. For the NACA 0012 airfoil and the NASA Rotor 67 stall cases, the parametric uncertainty range envelopes the 2D airfoil stall point, but not for the 3D compressor stall point. These results highlight the need for a modified model form with respect to compressor flow features, i.e. 3D swirling motion and strong adverse pressure gradient.




Related publications:

• He, X.*, Zhao, F., and Vahdati, M., “Uncertainty Quantification of Spalart-Allmaras Turbulence Model Coefficients for Compressor Stall,” ASME Journal of Turbomachinery, 2021, 143(8), 081007. [preprint] [doi]
• He, X.*, Zhao, F., and Vahdati, M., “Uncertainty Quantification of Spalart-Allmaras Turbulence Model Coefficients for Simplified Compressor Flow Features,” ASME Journal of Fluids Engineering, 2020, 142(9), 091501. [preprint] [doi]

Advisor: Prof. Mehdi Vahdati

---

Turbomachinery Aerothermodynamics

Flow Instabilities in Centrifugal Compressors

Flow instabilities with various temporal and spatial scales narrow the stable operating range of centrifugal compressors. In this study, the pathology of a high pressure ratio centrifugal compressor and the flow control method by self-recirculation casing treatment are investigated based on compressor rig tests with fast-response pressure measurements.

By performing the Fast Fourier Transform, diverse instabilities have been identified namely rotating instability, rotating stall, and surge in ascending order of temporal and spatial scale. At low speeds, small-scale rotating instability develops and evolves into mid-scale impeller stall, which eventually triggers the compression system to large-scale surge. At middle speeds, the compressor successively experiences stable state, mild surge, rotating instability, and deep surge, where the mild surge region coincides with the inflection point of the compressor S-shape pressure rise characteristic. At high speeds, mild surge and deep surge abruptly occur without preceding rotating stall or rotating instability.

Self-recirculation casing treatment was developed to suppress the flow instabilities with various scales. For the small-scales and mid-scales, both the injection flow of the front slot and the suction flow of the rear slot will restrain the formation of radial vortex connecting the impeller suction surface and the casing wall, thus the elimination of rotating instability and impeller stall. For the large-scales, the flow recirculation process turns the pressure rise slope of the impeller and the compressor stage more negative, which stabilizes the compression system and constrains surge.




Related publications:

• He, X., and Zheng, X., “Roles and Mechanisms of Casing Treatment on Different Scales of Flow Instability in High Pressure Ratio Centrifugal Compressors,” Aerospace Science and Technology, 2019, 84, 734-746. [preprint] [doi]
• He, X., and Zheng, X., “Flow Instability Evolution in High Pressure Ratio Centrifugal Compressor with Vaned Diffuser,” Experimental Thermal and Fluid Science, 2018, 98, 719-730. [preprint] [doi]

Advisor: Prof. Xinqian Zheng

Transonic Flows in Centrifugal Compressors

Shock and shock-leakage interaction in transonic centrifugal compressors deteriorate the flow field and lower the compressor efficiency. In this study, the loss mechanism of transonic centrifugal compressors and the flow control method by 3D blade leading edge are investigated via RANS CFD simulations.

Based on the detailed analysis of the entropy field, it was found that the shock-leakage interaction near the blade leading edge enhances the wake region and thus increases the local loss. A parametric study of the blade leading edge design degrees of freedom was conducted, and phenomenological flow models were proposed to complement the understanding of their aerodynamic effects. To obtain the optimal shape of the blade leading edge, an optimization framework based on the artificial neural network and the genetic algorithm was established. The final design achieves a 2.2% increment in isentropic efficiency across a wide operating range.




Related publications:

• He, X., and Zheng, X., “Performance Improvement of Transonic Centrifugal Compressors by Optimization of Complex Three-Dimensional Features,” IMechE, Part G: Journal of Aerospace Engineering, 2017, 231(14), 2723-2738. [preprint] [doi]
• He, X., and Zheng, X., “Mechanisms of Sweep on the Performance of Transonic Centrifugal Compressor Impellers,” Applied Sciences, 2017, 7(10), 1081. [doi]
• He, X., and Zheng, X., “Mechanisms of Lean on the Performance of Transonic Centrifugal Compressor Impellers,” AIAA Journal of Propulsion and Power, 2016, 32(5), 1220-1229. [preprint] [doi]

Advisor: Prof. Xinqian Zheng

*Winner of the Excellent Bachelor Thesis Award of Tsinghua University (2015)

---