Detached Eddy Simulation of Unsteady Stall Flows of a Full Annulus Transonic Rotor

2010 ◽  
Author(s):  
Hong-Sik Im ◽  
Xiangying Chen ◽  
Ge-Cheng Zha
Keyword(s):  
Pressure Ratio ◽  
Axial Flow ◽  
Leading Edge ◽  
Rotating Stall ◽  
Tip Clearance ◽  
Tip Vortex ◽  
Detached Eddy Simulation ◽  
Stall Inception ◽  
Tip Clearance Flow ◽  
Eddy Simulation

This paper uses the advanced Delayed-Detached Eddy Simulation (DDES) of turbulence to simulate rotating stall inception of NASA Rotor 67. The rotor is a low-aspect-ratio transonic axial-flow fan with a tip speed of 429 m/s and a pressure ratio of 1.63. A full annulus simulation was employed with the time accurate compressible Navier-Stokes code in order to accurately capture the the formation of long-length disturbance and a short-length inception (spike). The validation for all numerical methods used in this study was accomplished by the comparisons of the CFD solutions with the test data in advance of unsteady simulations. Self-induced rotating stall development is simulated holding the same back pressure at the near stall experiment without any throttling. Spike type rotating stall occurs and rotates at roughly 50% of rotor speed counter to the rotation. After spike onset, rotating stall fully develops approximately within 2 rotor revolutions. Two distinct characteristics that can advance the mechanism of spike type rotating stall are observed. First, the passage shock is fully detached from rotor and decays during the spike inception. Consequently the shifted sonic line at the upstream of rotor allows stalling flow to propagate to the neighboring passage. Second, the trailing edge back flow contributes to the build up of a fully developed stall cell by pushing tip clearance flow toward blade leading edge and inducing tip spillage flow. Tip vortex originated from the leading edge dies out during spike inception as the swirl angle of incoming tip flow decreases, while in the unstalled passages it develops without breakdown. DDES challenge for the complete blade row reflects well the sequence of rotating stall and its unsteady behavior.

Large-Scale Detached-Eddy Simulation Analysis of Stall Inception Process in a Multistage Axial Flow Compressor

2017 ◽  
Vol 139 (7) ◽  
Author(s):  
Kazutoyo Yamada ◽  
Masato Furukawa ◽  
Yuki Tamura ◽  
Seishiro Saito ◽  
Akinori Matsuoka ◽  
...  
Keyword(s):  
Gas Turbine ◽  
Large Scale ◽  
Axial Flow ◽  
Leading Edge ◽  
Rotating Stall ◽  
Detached Eddy Simulation ◽  
Stall Inception ◽  
Eddy Simulation ◽  
Flow Compressor ◽  
Edge Separation

This paper describes the flow mechanisms of rotating stall inception in a multistage axial flow compressor of an actual gas turbine. Large-scale numerical simulations of the unsteady have been conducted. The compressor investigated is a test rig compressor that was used in the development of the Kawasaki L30A industrial gas turbine. While the compressor consists of a total of 14 stages, only the front stages of the compressor were analyzed in the present study. The test data show that the fifth or sixth stages of the machine are most likely the ones leading to stall. To model the precise flow physics leading to stall inception, the flow was modeled using a very dense computational mesh, with several million cells in each passage. A total of 2 × 109 cells were used for the first seven stages (3 × 108 cells in each stage). Since the mesh was still not fine enough for large-eddy simulation (LES), a detached-eddy simulation (DES) was used. Using DES, a flow field is calculated using LES except in the near-wall where the turbulent eddies are modeled by Reynolds-averaged Navier–Stokes. The computational resources required for such large-scale simulations were still quite large, so the computations were conducted on the K computer (RIKEN AICS in Japan). Unsteady flow phenomena at the stall inception were analyzed using data mining techniques such as vortex identification and limiting streamline drawing with line integral convolution (LIC) techniques. In the compressor studied, stall started from a separation on the hub side rather than the commonly observed leading-edge separation near the tip. The flow phenomenon first observed in the stalling process is the hub corner separation, which appears in a passage of the sixth stator when approaching the stall point. This hub corner separation grows with time, and eventually leads to a leading-edge separation on the hub side of the stator. Once the leading-edge separation occurs, it rapidly develops into a rotating stall, causing another leading-edge separation of the neighboring blade. Finally, the rotating stall spreads to the upstream and downstream blade rows due to its large blockage effect.

Rotating Stall Observations in a High Speed Compressor—Part II: Numerical Study

2015 ◽  
Vol 137 (5) ◽  
Author(s):  
J. Dodds ◽  
M. Vahdati
Keyword(s):  
High Speed ◽  
Numerical Study ◽  
Axial Flow ◽  
Leading Edge ◽  
Rotating Stall ◽  
Tip Clearance ◽  
Tip Clearance Flow ◽  
Numerical Predictions ◽  
Stage 1 ◽  
Variable Stator

In this two-part paper the phenomenon of part span rotating stall is studied. The objective is to improve understanding of the physics by which stable and persistent rotating stall occurs within high speed axial flow compressors. This phenomenon is studied both experimentally (Part I) and numerically (Part II). The experimental observations reported in Part I are now explored through the use of 3D unsteady Reynolds-averaged Navier–Stokes (RANS) simulation. The objective is to both validate the computational model and, where possible, explore some physical aspects of the phenomena. Unsteady simulations are presented, performed at a fixed speed with the three rows of variable stator vanes adjusted to deliberately mismatch the front stages and provoke stall. Two families of rotating stall are identified by the model, consistent with experimental observations from Part I. The first family of rotating stall originates from hub corner separations developing on the stage 1 stator vanes. These gradually coalesce into a multicell rotating stall pattern confined to the hub region of the stator and its downstream rotor. The second family originates from regions of blockage associated with tip clearance flow over the stage 1 rotor blade. These also coalesce into a multicell rotating stall pattern of shorter length scale confined to the leading edge tip region. Some features of each of these two patterns are then explored as the variable stator vanes (VSVs) are mismatched further, pushing each region deeper into stall. The numerical predictions show a credible match with the experimental findings of Part I. This suggests that a RANS modeling approach is sufficient to capture some important aspects of part span rotating stall behavior.

Criteria for Spike Initiated Rotating Stall

2008 ◽  
Vol 130 (1) ◽  
Author(s):  
Huu Duc Vo ◽  
Choon S. Tan ◽  
Edward M. Greitzer
Keyword(s):  
Computational Study ◽  
Axial Compressor ◽  
Leading Edge ◽  
Rotating Stall ◽  
Short Length ◽  
Tip Clearance ◽  
Length Scale ◽  
Stall Inception ◽  
Tip Clearance Flow ◽  
Clearance Flow

A computational study to define the phenomena that lead to the onset of short length-scale (spike) rotating stall disturbances has been carried out. Based on unsteady simulations, we hypothesize there are two conditions necessary for the formation of spike disturbances, both of which are linked to the tip clearance flow. One is that the interface between the tip clearance and oncoming flows becomes parallel to the leading-edge plane. The second is the initiation of backflow, stemming from the fluid in adjacent passages, at the trailing-edge plane. The two criteria also imply a circumferential length scale for spike disturbances. The hypothesis and scenario developed are consistent with numerical simulations and experimental observations of axial compressor stall inception. A comparison of calculations for multiple blades with those for single passages also allows statements to be made about the utility of single passage computations as a descriptor of compressor stall.

An Experimental and Computational Investigation of Tip Clearance Flow and Its Impact on Stall Inception

2010 ◽  
Author(s):  
Matthew A. Bennington ◽  
Mark H. Ross ◽  
Joshua D. Cameron ◽  
Scott C. Morris ◽  
Juan Du ◽  
...  
Keyword(s):  
Axial Compressor ◽  
Axial Flow ◽  
Leading Edge ◽  
Tip Clearance ◽  
Flow Coefficient ◽  
Momentum Balance ◽  
Tip Leakage Flow ◽  
Stall Inception ◽  
Tip Clearance Flow ◽  
Clearance Flow

A numerical and experimental study was conducted to investigate the tip clearance flow and its relationship to stall in a transonic axial compressor. The CFD results were used to identify the existence of an interface between incoming axial flow and the reverse tip clearance flow. A surface streaking method was used to experimentally identify this interface as a line of zero axial shear stress at the casing. The position of this line, denoted xzs, moved upstream with decreasing flow coefficient in both the experiments and computations. The line was found to be at the rotor leading edge plane when the compressor stalled. Further measurements using rotor offset and inlet distortion further corroborated these results, and demonstrated that the movement of the interface upstream of the leading edge leads to the generation of rotating (“spike”) disturbances. Stall was therefore interpreted to occur as a result of a critical momentum balance between the approach fluid and the tip-leakage flow.

Short Length-Scale Rotating Stall Inception in a Transonic Axial Compressor: Criteria and Mechanisms

2006 ◽  
Author(s):  
Chunill Hah ◽  
Jo¨rg Bergner ◽  
Heinz-Peter Schiffer
Keyword(s):  
Axial Compressor ◽  
Rotating Stall ◽  
Tip Clearance ◽  
Stall Inception ◽  
Tip Clearance Flow ◽  
Transonic Compressor ◽  
Eddy Simulation ◽  
Clearance Flow ◽  
Flow Mechanisms ◽  
Tip Clearance Vortex

The current paper reports on investigations aimed at advancing the understanding of the flow mechanism that leads to the onset of short-length scale rotating stall in a transonic axial compressor. Experimental data show large oscillation of the tip clearance vortex as the rotor operates near the stall condition. Inception of spike-type rotating stall is also measured in the current transonic compressor with high response pressure transducers. Computational studies of a single passage and the full annulus were carried out to identify flow mechanisms behind the spike-type stall inception in the current transonic compressor rotor. Steady and unsteady single passage flow simulations were performed, first to get insight into the interaction between the tip clearance vortex and the passage shock. The conventional Reynolds-averaged Navier-Stokes method with a standard turbulence closure scheme does not accurately reproduce tip clearance vortex oscillation and the measured unsteady pressure field. Consequently, a Large Eddy Simulation (LES) was carried out to capture more relevant physics in the computational simulation of the rotating stall inception. The unsteady random behavior of the tip clearance vortex and it’s interaction with the passage shock seem to be critical ingredients in the development of spike-type rotating stall in a transonic compressor. The Large Eddy Simulation was further extended to the full annulus to identify flow mechanisms behind the measured spike-type rotating stall inception. The current study shows that the spike-type rotating stall develops after the passage shock is fully detached from the blade passages. Interaction between the tip clearance vortex and the passage shock creates a low momentum area near the pressure side of the blade. As the mass flow rate decreases, this low momentum area moves further upstream and reversed tip clearance flow is initiated at the trailing edge plane. Eventually, the low momentum area near the pressure side reaches the leading edge and forward spillage of the tip clearance flow occurs. The flows in the affected blade passage or passages then stall. As the stalled blade passages are formed behind the passage shock, the stalled area rotates counter to the blade rotation just like the classical Emmon’s type rotating stall. Both the measurements and the computations show that the rotating stall cell covers one to two blade passage lengths and rotates at roughly 50% of the rotor speed.

Experimental Study of Modal Disturbance and Rotating Stall Distribution

2013 ◽  
Vol 718-720 ◽  
pp. 1804-1810
Author(s):  
An Qing Lai ◽  
Jun Hu ◽  
Liang Li ◽  
Ju Luo
Keyword(s):  
Axial Compressor ◽  
Leading Edge ◽  
Propagation Speed ◽  
Axial Direction ◽  
Rotating Stall ◽  
Tip Clearance ◽  
Rotor Speed ◽  
Stall Inception ◽  
Tip Clearance Flow ◽  
Modal Type

To execute stall active control technology effectively and make clear of stall inception induced by modal disturbance, this paper carries out the correlative research on modal disturbance and rotating stall on the two-stage low-speed axial compressor. The results indicate that the stall inception of the compressor is modal style and the modal oscillation propagates at 38% rotor speed while the stall cell propagation speed is 42% rotor speed. The phase angles of modal oscillation and rotating stall along the axial direction are different, but their trajectories are both similar to the blade passage shape. The stall mechanisms of modal-type and spike-type inceptions are different. It doesnt appear that leading-edge tip clearance flow spillage blow the blade tip while the modal-type stall formation.

Delayed Detached Eddy Simulation of Rotating Stall for a Full Annulus Transonic Axial Compressor Stage

2016 ◽  
Author(s):  
Jiaye Gan ◽  
Hong-Sik Im ◽  
Ge-Cheng Zha
Keyword(s):  
Stokes Equations ◽  
Axial Compressor ◽  
Rotating Stall ◽  
Tip Clearance ◽  
Navier Stokes ◽  
Detached Eddy Simulation ◽  
Stall Inception ◽  
Eddy Simulation ◽  
Delayed Detached Eddy Simulation ◽  
Stall Cells

This paper solves the filtered Navier-Stokes equations to simulate stall inception of NASA compressor transonic Stage 35 with delayed detached eddy simulation (DDES). A low diffusion E-CUSP Riemann solver with a 3rd order MUSCL scheme for the inviscid fluxes and a 2nd order central differencing for the viscous terms are employed. A full annulus of the rotor-stator stage is simulated with an interpolation sliding boundary condition (BC) to resolve the rotor-stator interaction. The tip clearance is fully gridded to accurately resolve tip vortices and their effect on stall inception. The DDES results show that the stall inception of Stage 35 is initialized by a weak harmonic disturbance with the length scales of the full annulus and grows rapidly with two emerging spike like disturbance. The two spike disturbances propagate in counter rotational direction with about 42% of rotor speed. The spike stall cells cover about 6 blades. They lead to two stall cells grown circumferentially and inwardly.

Large-Scale DES Analysis of Stall Inception Process in a Multi-Stage Axial Flow Compressor

2016 ◽  
Author(s):  
Kazutoyo Yamada ◽  
Masato Furukawa ◽  
Yuki Tamura ◽  
Seishiro Saito ◽  
Akinori Matsuoka ◽  
...  
Keyword(s):  
Gas Turbine ◽  
Large Scale ◽  
Axial Flow ◽  
Leading Edge ◽  
Rotating Stall ◽  
Stall Inception ◽  
Eddy Simulation ◽  
Multi Stage ◽  
Flow Compressor ◽  
Edge Separation

The paper describes the flow mechanism of the rotating stall inception in a multi-stage axial flow compressor for an actual gas turbine. Large-scale numerical simulations have been conducted. The compressor investigated is a test rig compressor which was used for development of the industrial gas turbine, Kawasaki L30A. While the compressor consists of 14 stages, the front half stages of the compressor were analyzed in the present study. According to the test data, it is considered that the 5th or 6th stage is the one most suspected of leading to the stall. In order to capture precise flow physics that could happen at stall inception, a computational mesh was made dense, giving at least several million cells to each passage. It amounted to about two billion cells for the first 7 stages (three hundred million cells in each stage). Since the mesh was still not enough for the large-eddy simulation (LES), the detached-eddy simulation (DES) was employed. In the DES, a flow field is calculated by LES except near-wall and near-wall turbulent eddies are modeled by RANS. The computational resource required for such large-scale simulation was still quite large, so the computations were conducted on the K computer (RIKEN AICS in Japan). Unsteady flow phenomena at the stall inception were analyzed by using data mining techniques such as vortex identification and limiting streamline drawing with the LIC (line integral convolution) method. The present compressor has stall started from the separation on the hub side instead of the commonly observed leading-edge separation near the tip. The flow phenomenon first observed in the stalling process is the hub corner separation, which appears in some passage of the 6th stator when approaching the stall point. This hub corner separation expands with time, and eventually leads to the leading-edge separation on the hub side for the stator. Once the leading-edge separation happens, it rapidly develops into the rotating stall, causing another leading-edge separation for the neighboring blade in sequence. Finally, the rotating stall spreads to the upstream and downstream bladerows due to its large blockage effect.

Criteria for Spike Initiated Rotating Stall

2005 ◽  
Author(s):  
Huu Duc Vo ◽  
Choon S. Tan ◽  
Edward M. Greitzer
Keyword(s):  
Computational Study ◽  
Axial Compressor ◽  
Leading Edge ◽  
Rotating Stall ◽  
Tip Clearance ◽  
Length Scale ◽  
Stall Inception ◽  
Tip Clearance Flow ◽  
Clearance Flow ◽  
Circumferential Extent

A computational study to define the phenomena that lead to the onset of short length-scale (spike) rotating stall disturbances has been carried out. Unsteady simulations show there are two conditions necessary for the formation of spike disturbances, both of which are linked to the tip clearance flow. One is that the interface between the tip clearance and oncoming flows becomes parallel to the leading edge plane. The second is the initiation of backflow, stemming from the fluid in adjacent passages, at the trailing edge plane. The two criteria also imply a length scale circumferential extent of spike disturbances. The scenario developed is consistent with numerical simulations as well as with experimental observations of axial compressor stall inception. A comparison of calculations for multiple blades with those for single passages also allows statements to be made about the utility of single passage computations as a descriptor of compressor stall.

Rotating Stall Observations in a High Speed Compressor: Part 2 — Numerical Study

2014 ◽  
Author(s):  
J. Dodds ◽  
M. Vahdati
Keyword(s):  
High Speed ◽  
Numerical Study ◽  
Axial Flow ◽  
Leading Edge ◽  
Rotating Stall ◽  
Tip Clearance ◽  
Tip Clearance Flow ◽  
Numerical Predictions ◽  
Stage 1 ◽  
Variable Stator

In this two part paper the phenomenon of part span rotating stall is studied. The objective is to improve understanding of the physics by which stable and persistent rotating stall occurs within high speed axial flow compressors. This phenomenon is studied both experimentally (part 1) and numerically (part 2). The experimental observations reported in Part 1 are now explored through the use of 3D unsteady RANS simulation. The objective is to both to validate the computational model and, where possible, explore some physical aspects of the phenomena. Unsteady simulations are presented, performed at a fixed speed with the three rows of variable stator stagger vanes adjusted to deliberately mismatch the front stages and provoke stall. Two families of rotating stall are identified by the model, consistent with experimental observations from Part 1. The first family of rotating stall originates from hub corner separations developing on the stage 1 stator vanes. These gradually coalesce into a multi-cell rotating stall pattern confined to the hub region of the stator and its downstream rotor. The second family originates from regions of blockage associated with tip clearance flow over the stage 1 rotor blade. These also coalesce into a multi-cell rotating stall pattern of shorter length scale confined to the leading edge tip region. Some features of each of these two patterns are then explored as the variable stator vanes are mismatched further, pushing each region deeper into stall. The numerical predictions show a credible match with the experimental findings of Part 1. This suggests that a RANS modelling approach is sufficient to capture some important aspects of part span rotating stall behavior.

Sign in / Sign up

Export Citation Format

Share Document