The efficiency of modern axial turbomachinery is strongly driven by secondary flows within the vane or blade passages. The secondary flows are characterized by a complex pattern of vortical structures that origin, interact and dissipate along the turbine gas path. A reliable vortex identification technique is required to predict and analyse the impact of secondary flows on the turbine performance. However, literature shows a remarkable lack of general methods to detect vortices, determine the location of their cores and to quantify their strength. This paper presents a novel technique for vortex identification in a general 3D flow field. The method operates on the local flow field and it is based on a triple decomposition of motion proposed by Kolár. In contrast to a decomposition of velocity gradient into the strain and vorticity tensors, this method considers a third, pure shear component to allow distinction between rigid rotation and shear. The triple decomposition of motion serves to obtain a 3D field of residual vorticity whose magnitude is used to define vortex regions. The present technique locates automatically the core of each vortex, quantifies its strength and determines the bounding surface. The vortex identification method is applied to a high-pressure turbine with three optimized blade tip geometries. The 3D flowfield is obtained by CFD computations performed with Numeca FINE/Open, using steady-state RANS equations with the Spalart-Allmaras turbulence model. Although developed for turbomachinery applications, the proposed vortex detection technique is of general applicability to any three-dimensional flow-field.

The present study deals with the application of the transient thermochromic liquid crystal (TLC) technique in a flow network of intersecting circular passages as a potential internal turbine component cooling geometry. The investigated network consists of six circular passages with a diameter d = 20 mm that intersect coplanar at an angle θ = 40 deg, the innermost in three, the outermost in one intersection level. Two additional nonintersecting passages serve as references. Such a flow network entails specific characteristics associated with the transient TLC method that have to be accounted for in the evaluation process: the strongly curved surfaces, the mixing and mass flow redistribution at each intersection point, and the resulting gradients between the wall and passage centerline temperatures. All this impedes the choice of a representative fluid reference temperature, which results in deviations using established evaluation methods. An alternative evaluation approach is introduced, which is supported by computational results obtained from steady-state three-dimensional (3D) Reynolds-averaged Navier–Stokes equations (RANS) simulations using the shear-stress transport (SST) turbulence model. The presented analysis uncouples local heat transfer (HT) coefficients from actually measured local temperatures but uses the time information of the thermocouples (TC) instead that represents the fluid temperature step change and evolution along the passages. This experimental time information is transferred to the steady-state numerical bulk temperatures, which are finally used as local references to evaluate the transient TLC experiments. As effective local mass flow rates in the passage sections are considered, the approach eventually allows for a conclusion whether HT is locally enhanced due to higher mass flow rates or the intersection effects.

The continuous drive for ever higher turbine entry temperatures is leading to considerable interest in high performance cooling systems which offer high cooling effectiveness with low coolant utilization. The double-wall system is an optimized amalgamation of more conventional cooling methods including impingement cooling, pedestals, and film cooling holes in closely packed arrays characteristic of effusion cooling. The system comprises two walls, one with impingement holes, and the other with film holes. These are mechanically connected via pedestals allowing conduction between the walls while increasing coolant-wetted area and turbulent flow. However, in the open literature, experimental data on such systems are sparse. This study presents a new experimental heat transfer facility designed for investigating double-wall systems. Key features of the facility are discussed, including the use of infrared thermography to obtain overall cooling effectiveness measurements. The facility is designed to achieve Reynolds and Biot (to within 10%) number similarity to those seen at engine conditions. The facility is used to obtain overall cooling effectiveness measurements for a circular pedestal, double-wall test piece at three coolant mass-flows. A conjugate computational fluid dynamics (CFD) model of the facility was developed providing insight into the internal flow features. Additionally, a computationally efficient, decoupled conjugate method developed by the authors for analyzing double-wall systems is run at the experimental conditions. The results of the simulations are encouraging, particularly given how computationally efficient the method is, with area-weighted, averaged overall effectiveness within a small margin of those obtained from the experimental facility.

The current industrial standard for numerical simulations of axial compressors is the steady Reynolds-averaged Navier–Stokes (RANS) approach. Besides the well-known limitations of mixing planes, namely their inherent inability to capture the potential interaction and the wakes from the upstream blades, there is another flow feature which is lost, and which is a major accountable for the radial mixing: the transport of streamwise vorticity. Streamwise vorticity is generated for various reasons, mainly associated with secondary and tip-clearance flows. A strong link exists between the strain field associated with the vortices and the mixing augmentation: the strain field increases both the area available for mixing and the local gradients in fluid properties, which provide the driving potential for the mixing. In the rear compressor stages, due to high clearances and low aspect ratios, only accounting for the development of secondary and clearance flow structures, it is possible to properly predict the spanwise mixing. In this work, the results of steady and unsteady simulations on a heavy-duty axial compressor are compared with experimental data. Adopting an unsteady framework, the enhanced mixing in the rear stages is properly captured, in remarkable agreement with experimental distributions. On the contrary, steady analyses strongly underestimate the radial transport. It is inferred that the streamwise vorticity associated with clearance flows is a major driver of radial mixing, and restraining it by pitch-averaging the flow at mixing planes is the reason why the steady approach cannot predict the radial transport in the rear part of the compressor.

Objective of this paper is to analyze the consequences of borescope blending repairs on the aeroelastic behavior of a modern high pressure compressor (HPC) blisk. To investigate the blending consequences in terms of aerodynamic damping and forcing changes, a generic blending of a rotor blade is modeled. Steady-state flow parameters like total pressure ratio, polytropic efficiency, and the loss coefficient are compared. Furthermore, aerodynamic damping is computed utilizing the aerodynamic influence coefficient (AIC) approach for both geometries. Results are confirmed by single passage flutter (SPF) simulations for specific interblade phase angles (IBPA) of interest. Finally, a unidirectional forced response analysis for the nominal and the blended rotor is conducted to determine the aerodynamic force exciting the blade motion. The frequency content as well as the forcing amplitudes is obtained from Fourier transformation of the forcing signal. As a result of the present analysis, the change of the blade vibration amplitude is computed.

Time-accurate transient blade row (TBR) simulation approaches are required when there is a close flow coupling between the blade rows, and for fundamentally transient flow phenomena such as aeromechanical analysis. Transient blade row simulations can be computationally impractical when all of the blade passages must be modeled to account for the unequal pitch between the blade rows. In order to reduce the computational cost, time-accurate pitch-change methods are utilized so that only a sector of the turbomachine is modeled. The extension of the time-transformation (TT) pitch-change method to multistage machines has recently shown good promise in predicting both aerodynamic performance and resolving dominant blade passing frequencies for a subsonic compressor, while keeping the computational cost affordable. In this work, a modified 1.5 stage Purdue transonic compressor is examined. The goal is to assess the ability of the multistage time-transformation method to accurately predict the aerodynamic performance and transient flow details in the presence of transonic blade row interactions. The results from the multistage time-transformation simulation were compared with a transient full-wheel simulation. The aerodynamic performance and detailed flow features from the time-transformation solution closely matched the full-wheel simulation at fractional of the computation cost.

When the operating condition of a gas turbine engine changes from one steady-state to another, the cooling must ensure that the solid's temperatures never exceed the maximum allowable throughout the transient process. Exceeding the maximum allowable temperature is possible even though cooling is increased to compensate for the increase in heating because there is a time lag in how the solid responds to changes in its convective heating and cooling environments. In this paper, a closed-form solution (referred to as the 1D model) is derived to estimate the over temperature and its duration in a flat plate subjected to sudden changes in heating and cooling rates. For a given change in heating rate, the 1D model can also be used to estimate the minimum cooling needed to ensure that the new steady-state temperature will not exceed the maximum allowable. In addition, this model can estimate the temperature the material must be cooled to before imposing a sudden increase in heat load to ensure no over temperature throughout the transient process. Comparisons with the exact solutions show the 1D model to be accurate within 0.1%. This 1D model was generalized for application to problems in multidimensions. The generalized model was used to estimate the duration of over temperature in a two-dimensional problem involving variable heat transfer coefficient (HTC) on the cooled side of a flat plate and provided results that match the exact solution within 5%.

An investigation to characterize the effect of entrainment in a confined jet impingement arrangement is presented. The investigated configuration shows an impingement-cooled turbine blade passage and holds two staggered rows of inclined impingement jets. In order to distinctly promote thermal entrainment phenomena, the jets were heated separately. A steady-state liquid crystal technique was used to obtain near-wall fluid temperature distributions for the impingement surfaces under adiabatic conditions. Additionally, flow field measurements were undertaken using particle image velocimetry (PIV). Furthermore, compressible Reynolds-averaged Navier–Stokes (RANS) simulations carried out with ansys cfx using Menter's shear stress transport (SST) turbulence model accompany the experiments. Distributions of effectiveness, velocity, and turbulent kinetic energy detail the complexity of the aerothermal situation. The study was conducted for a jet Reynolds number range from 10,000 to 45,000. The experimental and numerical results are generally in good agreement. Nevertheless, the simulations predict flow features in particular regions of the geometry that are not as prominent in the experiments. These affect the effectiveness distributions, locally. The investigations reveal that the effectiveness is independent of the temperature difference between the heated and cold jet as well as the jet Reynolds number.

This paper presents the design and experimental results of a new micro gas turbine architecture exploiting counterflow within a single supersonic rotor. This new architecture, called the supersonic rim-rotor gas turbine (SRGT), uses a single rotating assembly incorporating a central hub, a supersonic turbine rotor, a supersonic compressor rotor, and a rim-rotor. This SRGT architecture can potentially increase engine power density while significantly reducing manufacturing costs. The paper presents the preliminary design of a 5 kW SRGT prototype having an external diameter of 72.5 mm and rotational speed of 125,000 rpm. The proposed aerodynamic design comprises a single stage supersonic axial compressor, with a normal shock in the stator, and a supersonic impulse turbine. A pressure ratio of 2.75 with a mass flow rate of 130 g/s is predicted using a 1D aerodynamic model in steady state. The proposed combustion chamber uses an annular reverse-flow configuration, using hydrogen as fuel. The analytical design of the combustion chamber is based on a 0D model with three zones (primary, secondary, and dilution), and computational fluid dynamics (CFD) simulations are used to validate the analytical model. The proposed structural design incorporates a unidirectional carbon-fiber-reinforced polymer rim-rotor, and titanium alloy is used for the other rotating components. An analytical structural model and numerical validation predict structural integrity of the engine at steady-state operation up to 1000 K for the turbine blades. Experimentation has resulted in the overall engine performance evaluation. Experimentation also demonstrated a stable hydrogen flame in the combustion chamber and structural integrity of the engine for at least 30 s of steady-state operation at 1000 K.

The present work aims at evaluating the effect of the impeller–diffuser interaction on the control of a hub corner separation, which develops in the radial vaned diffuser of a centrifugal compressor designed and built by Turbomeca, Safran group. Unsteady numerical simulations of the flow in the aspirated centrifugal compressor are then performed. Numerical results are validated by comparison with the available experimental results. The analysis of the numerical flow field shows that the hub-corner separation is not completely removed by the suction, on the contrary to the steady-state results that were obtained in previous work. The boundary layer separation is only translated downstream. Its location is explained by the scrolling of the pressure waves generated by the impeller–diffuser interaction, which strengthen when crossing the diffuser throat. This result highlights the major role played by the impeller–diffuser interaction, which should be taken into account for developing control strategies in radial vaned diffusers, and stresses the shortcoming of the steady-state numerical model when suction is applied.

This paper presents a computational fluid dynamics (CFD) study performed to assess the prediction of the minimum stable volume flow for a high Mach number, high head, and high volume flow compressor stage. CFD was run on a “pie slice” or sector stage model in steady-state condition and on a full 360 deg stage model under both steady and unsteady state conditions. The predictions of the minimum stable flow were compared to experimental data. Results showed the CFD performed on the “pie slice” stage model over-predicted the minimum stable flow by 9% compared to the test results, while the transient CFD predicted the minimum stable flow within 5.8%. Flow field comparisons of the impeller between unsteady and steady state CFD revealed that the steady state CFD accurately predicted the flow phenomena until the onset of surge. However, the unsteady flow features could not propagate through the diffuser because of the limitations of the impeller-diffuser interface modeling in the steady state analysis.

A computational model has been developed to study the mechanisms responsible for hot gas ingestion into the wheel-space cavity of a stationary high pressure turbine (HPT) cascade rig. Simulations were undertaken for the stationary rig described by Bunker et al. (2009, “An Investigation of Turbine Wheelspace Cooling Flow Interactions With a Transonic Hot Gas Path—Part I: Experimental Measurements,” ASME Paper No. GT2009-59237) in a companion paper. The rig consists of five vanes, a wheel-space cavity, and five cylinders that represent the blockage due to the leading edge of the rotor airfoils. The experimental program investigated two cylinder diameters and three clocking positions for a nominal coolant flow rate. Comparisons are made between the computed and measured flow-fields for the smaller of the two cylinders. It is demonstrated that the circumferential variation of pressure established by the vane wake and leading edge bow wave results in an unstable shear layer over the rim seal axial gap (trench) that causes hot gases to ingest for a nominal coolant flow. Steady-state computational fluid dynamics (CFD) simulations did not capture this effect and it was determined that an unsteady analysis was required in order to match the experimental data. Favorable agreement is noted between the time-averaged computed and measured pressure distributions in the circumferential direction both upstream and downstream of the trench, as well as within the trench itself. Furthermore, it is noted that time-averaged buffer cavity effectiveness agrees to within 5% of the experimental data for the cases studied. The validated CFD model is then used to simulate the effect of rotation by rotating the cylinders and disk at rotational rate that scales with a typical engine. A sliding mesh interface is utilized to communicate data between the stator and rotor domains. The stationary cases tend to ingest past the first angel-wing for a nominal coolant flow condition, whereas the effect of rotation helps pressurize the cavity and is responsible for preventing hot gas from entering the buffer cavity.

A new experimental technique for the accurate measurement of steady-state metal temperature surface distributions of modern heavily film-cooled turbine vanes has been developed and is described in this paper. The technique is analogous to the thermal paint test but has been designed for fundamental research. The experimental facility consists of an annular sector cascade of high pressure (HP) turbine vanes from a current production engine. Flow conditioning is achieved by using an annular sector of deswirl vanes downstream of the test section, being both connected by a three-dimensionally contoured duct. As a result, a transonic and periodic flow, highly representative of the engine aerodynamic field, is established: Inlet turbulence levels, mainstream Mach and Reynolds numbers, and coolant-to-mainstream total pressure ratio are matched. Since the fully three-dimensional nozzle guide vane (NGV) geometry is used, the correct radial pressure gradient and secondary flow development are simulated and the cooling flow redistribution is engine-realistic. To allow heat transfer measurements to be performed, a mainstream-to-coolant temperature difference (up to 33.5 ° C ) is generated by using two steel-wire mesh heaters, operated in series. NGV surface metal temperatures are measured (between 20 ° C and 40 ° C ) by wide-band thermochromic liquid crystals. These are calibrated in situ and on a per-pixel basis against vane surface thermocouples, in a heating process that spans the entire color play and during which the turbine vanes can be assumed to slowly follow a succession of isothermal states. Experimental surface distributions of overall cooling effectiveness are presented in this paper. By employing resin vanes of the same geometry and cooling configuration (to implement adiabatic wall thermal boundary conditions) and the transient liquid crystal technique, surface distributions of external heat transfer coefficient and film cooling effectiveness can be acquired. By combining these measurements with those from the metal vanes, the results can be scaled to engine conditions with a good level of accuracy.

The experimental performance evaluation of a circumferentially divided, double-entry turbocharger turbine is presented in this paper with the aim of understanding the influence of pulsating flow. By maintaining a constant speed but varying the frequency of the pulses, the influence of frequency was shown to play an important role in the performance of the turbine. A trend of decreasing cycle-averaged efficiency at lower frequencies was measured. One of the principal objectives was to assess the degree to which the unsteady performance differs from the quasi-steady assumption. In order to make the steady-unsteady comparison for a multiple entry turbine, a wide set of steady equal and unequal admission flow conditions were tested. The steady-state data was then interpolated as a function of three, nondimensional parameters in order to allow a point-by-point comparison with the instantaneous unsteady operation. As an average, the quasi-steady assumption generally underpredicted the mass flow and efficiency loss through the turbine, albeit the differences were reduced as the frequency increased. Out-of-phase pulsations produced unsteady operating orbits that corresponded to a significant steady-state, partial admission loss, and this was reflected as a drop in the quasi-steady efficiency. However, these differences between quasi-steady in-phase and out-of-phase predictions were not replicated in the measured results, suggesting that the unequal admission loss is not as significant in pulsating flow as it is in steady flow.

This paper investigates the influence of stator-rotor interaction on the stage performance of three blade tip geometries. A reference flat tip is used to assess two different recess blade geometries. The study is made in the context of the realistic turbine stage configuration provided by the ETHZ 1.5-stage LISA turbine research facility. This numerical investigation describes the details of unsteady recess cavity flow structure and confirms the beneficial effects of the improved recess geometry over the flat tip and the nominal recess design both in terms of stage efficiency and tip heat load. The tip flow field obtained from the improved recess design combines the advantages of a nominal recess design (aerodynamic sealing) and the flat tip configuration. The turbine stage capacity is almost unchanged between the flat tip and the improved recess tip cases, which simplifies the design procedure when using the improved recess design. The overall heat load in the improved recess case is reduced by 26% compared with the flat tip and by 14% compared with the nominal recess. A key finding of this study is the difference in effects of the upstream stator wake on the recess cavity flow. Where cavity flow in the nominal design is only moderately influenced, the improved recess cavity flow shows enhanced flow unsteadiness. The tip Nusselt number from a purely steady-state prediction in the nominal recess case is nearly identical to the time-average prediction. The improved design shows a 6% difference between steady-state and time-average tip Nusselt number. This is due to the strong influence of the wake passing on the recess cavity flow. In fact, the wake enhances a small flow difference at the leading edge of the recess cavity between the nominal and improved recess cavities, which results in a completely different flow field further downstream in the recess cavity.

This paper will focus on two core-compressor instabilities, namely, rotating stall and surge. Using a 3D viscous time-accurate flow representation, the front bladerows of a core compressor were modeled in a whole-annulus fashion whereas the rest of bladerows were represented in single-passage fashion. The rotating stall behavior at two different compressor operating points was studied by considering two different variable-vane scheduling conditions for which experimental data were available. Using a model with nine whole bladerows, the unsteady flow calculations were conducted on 32 CPUs of a parallel cluster, typical run times being around 3–4 weeks for a grid with about 60 × 10 6 points. The simulations were conducted over several engine rotations. As observed on the actual development engine, there was no rotating stall for the first scheduling condition while malscheduling of the stator vanes created a 12-band rotating stall which excited the rotor blade first flap mode. In a separate set of calculations, the surge behavior was modeled using a time-accurate single-passage representation of the core compressor. It was possible to predict not only flow reversal into the low pressure compression domain but also the expected hysteresis pattern of the surge loop in terms of its mass flow versus pressure characteristic.

In the current paper we introduce an innovative new concept in turbochargers—that of using active control at the turbine inlet with the aim of harnessing the highly dynamic exhaust gas pulse energy emanating at high frequency from an internal combustion engine, in order to increase the engine power output and reduce its exhaust emissions. Driven by the need to comply to increasingly strict emissions regulations as well as continually striving for better overall performance, the active control turbocharger is intended to provide a significant improvement over the current state of the art in turbocharging: the Variable Geometry Turbocharger (VGT). The technology consists of a system and method of operation, which regulate the inlet area to a turbocharger inlet, according to each period of engine exhaust gas pulse pressure fluctuation, thereby actively adapting to the characteristics of the high frequency, highly dynamic flow, thus taking advantage of the highly dynamic energy levels existent through each pulse, which the current systems do not take advantage of. In the Active (Flow) Control Turbocharger (ACT) the nozzle is able to adjust the inlet area at the throat of the turbine inlet casing through optimum amplitudes, at variable out-of-phase conditions and at the same frequency as that of the incoming exhaust stream pulses. Thus, the ACT makes better use of the exhaust gas energy of the engine than a conventional VGT. The technology addresses, therefore, for the first time the fundamental problem of the poor generic engine-turbocharger match, since all current state of the art systems in turbocharging are still passive receivers of this highly dynamic flow without being able to provide optimum turbine inlet geometry through each exhaust gas pulse period. The numerical simulation and experimental work presented in this paper concentrates on the potential gain in turbine expansion ratio and eventual power output as well as the corresponding effects on efficiency as a result of operating the turbocharger in its active control mode compared to its operation as a standard VGT.

The paper describes the theory and the numerical implementation of a three-dimensional finite volume scheme for the solution of the linearized, unsteady Favre-averaged Navier–Stokes equations for turbomachinery applications. A further feature is the use of mixed element grids, consisting of triangles and quadrilaterals in two dimensions, and of tetrahedra, triangular prisms, and hexahedra in three dimensions. The linearized unsteady viscous flow equations are derived by assuming small harmonic perturbations from a steady-state flow and the resulting equations are solved using a pseudo-time marching technique. Such an approach enables the same numerical algorithm to be used for both the nonlinear steady and the linearized unsteady flow computations. The important features of the work are the discretization of the flow domain via a single, unified edge-data structure for mixed element meshes, the use of a Laplacian operator, which results in a nearest neighbor stencil, and the full linearization of the Spalart–Allmaras turbulence model. Four different test cases are presented for the validation of the proposed method. The first one is a comparison against the classical subsonic flat plate cascade theory, the so-called LINSUB benchmark. The aim of the second test case is to check the computational results against the asymptotic analytical solution derived by Lighthill for an unsteady laminar flow. The third test case examines the implications of using inviscid, frozen-turbulence, and fully turbulent models when linearizing the unsteady flow over a transonic turbine blade, the so-called 11th International Standard Configuration. The final test case is a rotor/stator interaction, which not only checks the validity of the formulation for a three-dimensional example, but also highlights other issues, such as the need to linearize the wall functions. Detailed comparisons were carried out against measured steady and unsteady flow data for the last two cases and good overall agreement was obtained.

This paper presents results of unsteady viscous flow calculations and corresponding cold flow experiments of a three-stage low-pressure turbine. The investigation emphasizes the study of unsteady flow interaction. A time-accurate Reynolds-averaged Navier–Stokes solver is applied for the computations. Turbulence is modeled using the Spalart–Allmaras one-equation turbulence model and the influence of modern transition models on the unsteady flow predictions is investigated. The integration of the governing equations in time is performed with a four-stage Runge–Kutta scheme, which is accelerated by a two-grid method in the viscous boundary layer around the blades. At the inlet and outlet, nonreflecting boundary conditions are used. The quasi-three-dimensional calculations are conducted on a stream surface around midspan, allowing a varying stream tube thickness. In order to study the unsteady flow interaction, a three-stage low-pressure turbine rig of a modern commercial jet engine is built up. In addition to the design point, the Reynolds number, the wheel speed, and the pressure ratio are also varied in the tests. The numerical method is able to capture important unsteady effects found in the experiments, i.e., unsteady transition as well as the blade row interaction. In particular, the flow field with respect to time-averaged and unsteady quantities such as surface pressure, entropy, and skin friction is compared with the experiments conducted in the cold air flow test rig. [S0889-504X(00)02004-3]

Numerical stall flutter prediction methods are much needed, as modern jet engines require blade designs close to the stability boundaries of the performance map. A Quasi-3D Navier–Stokes code is used to analyze the flow over the oscillating cascade designed and manufactured by Pratt & Whitney, and studied at the NASA Glenn Research Center by Buffum et al. The numerical method solves for the governing equations with a fully implicit time-marching technique in a single passage by making use of a direct-store, periodic boundary condition. For turbulence modeling, the Baldwin–Lomax model is used. To account for transition, the criterion to predict the onset location suggested by Baldwin and Lomax is incorporated. Buffum et al. investigated two incidence cases for three different Mach numbers. The low-incidence case at a Mach number of 0.5 exhibited the formation of small separation bubbles at reduced oscillation frequencies of 0.8 and 1.2. For this case the present approach yielded good agreement with the steady and oscillatory measurements. At high incidence at the same Mach number of 0.5 the measured steady-state pressure distribution and the separation bubble on the upper surface was also found in good agreement with the experiment. But computations for oscillations at high incidence failed to predict the negative damping contribution caused by the leading edge separation. [S0889-504X(00)01304-0]