Numerical and experimental investigation of the shock-vane interaction in a transonic compressor

A numerical and experimental investigation of the shock-IGV (inlet guide vane) interaction in a transonic compressor is conducted at the Laboratoire de Thermique Appliquée et de Turbomachines (LTT) at the Ecole Polytechnique Fédérale de Lausanne (EPFL). The objective of the presented research work is to determine the impact of the upstream propagating shock pattern of an axial transonic compressor rotor on an aerodynamically loaded IGV and the influence of this interacting process on the compressor flow field for different operating points and aerodynamic setups. For the approach to the subject, a literature review shed light on the different research efforts over the past years. A large numerical study is conducted including the work with two different flow solvers. A numerical 3D transient analysis (commercial code) was conducted in order to provide information about the influence of the axial spacing on the overall mass flow through the stage. A numerical Q-3D transient analysis (industry owned code) on one streamline at a constant rotor inlet Mach number is performed in order to compare the shock-IGV interaction for different vane types similar to what is known in literature. A semi-analytical model was developed in order to evaluate in a fast approach the impact of a rotor bow shock on an upstream situated vane. The results of the numerical investigation can be summarized as following: The overall results of this investigation show that the upstream vane of a transonic compressor rotor transforms the rotor forcing function into an excitation with vane shape depending orientation and force. The time averaged flow field of the IGV does not show a measurable change of the exit flow angle due to the shock-IGV interaction for the studied operating points and axial distances. The 3D transient calculations do not show any influence of the examined axial spacing between vane and blade on the mass flow rate through the stage. The results of the Q3D analysis are compared with the semi-analytic approach and the measurements, both showing with some restrictions good accordance to the model. The results of the numerical investigation give an accurate prediction of the transient aerodynamic load of the upstream vane. An analytic tool was developed which uses three input parameters, the rotor relative inlet Mach number, the stagger angle of the rotor and the axial spacing between vane and rotor in order to provide an estimation of the amplitude of the pressure variation on the upstream situated vane. An extensive modification of the existing test facility of the LTT precedes the experimental study. The main steps of the approach are: A subsonic axial compressor test facility was modified to allow test runs at transonic rotor inlet conditions. A perfect leak tightness of the closed circuit was accomplished to enable its operation in heavy gas. Heavy gases like refrigerants have been used in many studies for compressor testing. While respecting certain restrictions during the use of heavy gas, a reduction of operation and manufacturing costs of the facility is beneficial. A high molecular weight of the gas reduces the sonic speed to the half of that of air thus reducing the dimension and the rotational speed of the rotor significantly and consequently reduces the demand in the material quality of the blading. A 1.5 stage transonic compressor had been designed and manufactured for the heavy gas (R134a) in a first step. Preliminary measurements showed that the influence of the downstream stator makes it difficult to separate the influence of shock-IGV interaction on the compressor flow field from loss generation mechanisms downstream of the rotor. Thus, the stator was removed and measurements were conducted in a one stage transonic compressor. Modifications on the construction of the testing setup are done in a way that the distance between IGV and rotor can be varied by means of spacers. Thus the flow can be determined for up to four different axial distances at varying rotational speed for different operating points. For transonic operation the time-dependent load of the IGV (inlet guide vane) due to the forcing function of the rotor bow shock was recorded with a high speed data acquisition system. The results of the experimental investigation can be summarized as following: No influence of the axial spacing on the efficiency and the mass flow was observed in the investigation. The operation characteristics remained equal between the minimum and the maximum vane-blade spacing. The measured global values of the stage are close to the values the design was made for, although the overall efficiency was generally lower than predicted. The experimental results of the pressure variation on the upstream vane agree well with the predicted values. The study in general showed a good matching between the numerical and the experimental investigation taking into account a rotor blade shape which is not optimized to the extend. The shock-vane interaction is responsible for an excitation of the entire vane blade. With decreasing distance between the rotor leading edge and vane trailing edge the excitation increases. The strength and orientation of the reflected shock/pressure wave part depends upon the vane shape. A more sophisticated design of the rotor blade shape should be in the focus of further investigations using this test facility at transonic rotor inlet conditions. Within this investigation a tool was developed in order to predict the load variation on the vane driven by the transient rotor forcing function. The tool maps the fact that the pressure on the vane surface oriented against rotational direction of the rotor exceeds the pressure of the passage shock due to the amplification effect of the reflection of the shock on a surface. Its possible application not only for compressors but also for turbine related topics should be an objective for similar investigations.

Aeroelastic effects of mistuning and coupling in turbomachinery bladings

Gerhard Kahl

A numerical method has been developed to study the effects of structural mistuning on the aeroelastic behaviour of turbomachinery cascades. The specific objectives of the work presented here were the following: A method was to be developed that can assess the major effects of structural mistuning in a cascade in sub- and transonic flow situations. The influence of mechanical and aerodynamic coupling between the blades as well as coupling between multiple modes for each blade was to be included. Accurate representations of the aerodynamic forces, taken from a modern three-dimensional unsteady aerodynamic solver were to be used. The method should be applicable for design use, meaning that it has to supply results quickly for a large number of configurations. The applicability and accuracy of the method was to be demonstrated by comparison of numerical results to available data from recent experiments. The influence of major parameters on the aeroelastic stability and on the resonant amplitude of representative test cases should be assessed. The approach used to achieve these goals is the combination of a linearised Euler method for the aerodynamic calculations with a modal reduction technique, where the structural properties of each blade are represented by only a few eigenmodes. The method is validated and applied to two test cases, comprising of a transonic compressor rotor and a high pressure turbine rotor. Both are representative of modern turbomachinery designs. The final conclusions of this work are: The newly developed method is capable to assess the dominant effects of structural mistuning in turbomachinery cascades, including the mechanical and aerodynamic coupling of adjacent blades and the aerodynamic coupling of multiple modes with arbitrarily complex modeshapes. In this method, the aerodynamic characteristics of the cascade are accurately represented using the generalised unsteady aerodynamic coefficients derived from a modern three-dimensional flow solver, applicable to and validated for both sub- and transonic configurations. The simplifications employed significantly contribute to the computational efficiency of the method, making it applicable for design purposes as well as for the assessment of large parametric variations or for statistical studies of stochastically mistuned configurations. The current method is successfully validated by a comparison of numerical to experimental data. The applicability and accuracy is demonstrated by the favourable agreement between measured and computed results. Based on these validations, the method is applied to study the influence of major parameters on the aeroelastic behaviour of the selected test cases. The results show a wide range of phenomena, dependent on the type and strength of mistuning, frequency, modeshape and interblade phase angle. The results highlight the close inter-dependence of the aeroelastic stability derived from the eigenvalue analysis and the resonant amplitudes derived from the forced response analysis.

EPFL2002

Highly loaded compressor with boundary layer suction

Klaus Hubrich

An experimental and numerical investigation of boundary layer suction (BLS) on a transonic compressor blade has been conducted. The objective of the present work is to identify the possible benefit of boundary layer suction via a slot in the region of the shock/boundary layer interaction (SBLI). The study has two major parts: First, an investigation of the flow on an isolated airfoil without and with BLS on the suction side (SS). This investigation is intended to optimise the suction slot geometry in terms of location and width. Second, a study on a transonic cascade with supersonic inlet flow has been conducted with the goal to show the influence of the BLS on the pressure loss coefficient and the surface Mach number distribution. An initial literature study revealed that only few work has been carried out in the field of BLS in transonic compressors and no data close to the suction slot was available. The main steps of the present study are: The design of an isolated airfoil for testing in a Laval nozzle, representative for the flow at a compressor root section, with enough space inside to evacuate the aspirated air. A numerical study of the flow on the isolated airfoil without and with BLS on the SS . An experimental investigation of the isolated airfoil without and with BLS on the SS using static surface pressure taps, PSP and the Schlieren method for data acquisition. Design and numerical investigation of a suitable transonic compressor cascade with enough space inside to evacuate the suction mass flow. An experimental investigation of the designed transonic cascade with an inlet Mach number of 1.23 with two sets of blades in the non-rotating annular cascade test-rig; one without and one with BLS at 40% chord. The main results are: The numerical study of the flow on the isolated airfoil without and with BLS on the SS shows, that the onset of shock-induced separation could be shifted to an increased inlet Mach number. Steady 2D-NS simulations show that the reference configuration exhibits an attached flow behind the shock up to an inlet Mach number of up to 0.70 for an inlet flow angle of +4°. The configuration with BLS shows attached flow behind the shock for an an inlet Mach number of 0.725. Unsteady 2D-NS simulations show for an inlet flow angle of +4° at an inlet Mach number of 0.725 for the reference configuration a shock, which moves periodically on the SS due to a separated boundary layer created by the shock/boundary layer interaction. The same configuration with BLS shows a stable shock position. The experimental investigation of the isolated airfoil without and with BLS on the SS shows that BLS at 15% chord upstream of the initial shock location for an inlet Mach number of 0.725 at a flow angle of +4° leads to a shock location 10% chord downstream of the initial position. Boundary layer suction suppresses flow separation and stabilises the shock. Due to the suction a stagnation point is created at the downstream edge of the suction slot. Downstream of this stagnation point, the flow is re-accelerated to the isentropic Mach number level of the reference configuration without BLS. The design and 3D-numerical investigation of a suitable transonic compressor cascade shows, that BLS with a suction mass flow of 2% can increase the maximum pressure ratio and diffusion factor of approximately 10%. This is the consequence of the stabilisation of the shock in the blade passage due to the suction. The experimental investigation of the designed transonic cascade with an inlet Mach number of 1.23 with 2% suction mass flow shows, that BLS acts locally and that suction leads to an increased isentropic surface Mach number in front of the suction slot and due to the suction related stagnation point, to a locally decreased velocity immediately downstream of the suction slot. Then the flow re-accelerates to the same level as the non-suction configuration attains. The pressure loss coefficient is decreased by up to 3.6% for the most positive incidences. The results of this investigation confirm that boundary layer suction on the suction side of an airfoil is a suitable way to increase the working range of a compressor as the shock position can be stabilised (shock-trapping) and the increase of the pressure loss coefficient can be shifted to higher incidences.

EPFL2005

Experimental Investigation of the Aeroelastic Stability of an Annular Compressor Cascade at Reverse Flow Conditions

Virginie Anne Chenaux

Compressor surge events are unsafe operating regimes yielding highly unsteady flow fields in which complex aeroelastic phenomena occur. If the blade flutter and forced response behaviour (i.e. aeroelastic stability) can be predicted reliably for normal flow conditions, its assessment at severe-off design conditions remains a critical task for compressor development programs. During the flow reversal sequence of a surge cycle, combined aerodynamic phenomena occur which make the accurate prediction of the unsteady forces acting on the blades difficult to assess. The main objective of this investigation is to increase the physical understanding of the unsteady phenomena present during the reverse flow sequence of a typical deep surge cycle. The analysis of the blade surface unsteady pressure distribution enables the identification of the main physical mechanisms present during such extreme flow operating conditions, as well as the evaluation of their contribution on the blade global aerodynamic stability. The approach adopted consists in performing aeroelastic investigations on an annular compressor cascade at established reverse flow conditions. The investigations are carried out at EPFL, in the annular test facility for non-rotating cascades. The cascade is forced to vibrate in a torsional travelling wave mode (controlled vibration). With an upstream swirled flow corresponding to real axial turbomachine conditions, a constant flow can be set in the test section. The steady-state operating conditions are measured upstream and downstream of the test section, using 5-hole aerodynamic probes. Several cascade blades are equipped with pressure taps at 50% span in order to acquire the steady-state and unsteady blade surface pressure distributions. Static pressure taps are also inserted in the casing wall of the test section to assess the steady-state flow field characteristics in the blade tip area. The inlet flow operating conditions are varied in order to determine their influence on the blade unsteady aerodynamic forces. This study presents the measurement results and analyses in details the aerodynamic response of a cascade subjected to controlled vibrations at reverse flow conditions. The data analysis is oriented towards both physical and practical approaches. In particular, the following features are addressed: Identification of the main unsteady physical mechanisms influencing the unsteady aerodynamic forces acting on a blade oscillating and subjected to reverse flow conditions. Determination of the influence of the inlet flow condition variations on these unsteady mechanisms. Evaluation of the contribution of each unsteady phenomenon to the blade aerodynamic stability (in terms of stabilizing or destabilizing impact). Assessment of the key parameters to control in order to minimize flutter risks in case that reverse flow conditions should occur. The data analysis reveals that during the reverse flow sequence of the surge cycle, blade interaction mechanisms play a major role in the blade aerodynamic stability. The global aerodynamic damping coefficient highlights this feature and indicates that aerodynamic instability exists for some operating conditions. A second unsteady phenomenon was detected, generated by the uncommon steady-state flow field characteristics present at reverse flow operating conditions. Within this frame, the presence of a large recirculation zone on the blade suction side was identified, influencing the blade aerodynamic stability. From a more general point of view, this study constitutes a step forward to the understanding of the blade loading processes occurring during a typical deep surge sequence. Results highlight the impact of the steady-state and unsteady phenomena on the blade loading level at reverse flow conditions. For one test case, the measured data was compared with numerical results, performed in parallel to the measurements. The results indicate that even though the agreement is reasonable, the correct prediction of the aerodynamic damping curve re- quires the consideration of a complex blade interaction mechanism. Within this frame, since not many experiments exist at reverse flow conditions, these experimental results are also a precious data source for CFD validation. They enable the improvement of the prediction/simulation accuracy of the compressor performance at off-design flow conditions.

EPFL2012