Turbomachinery Flows Research Articles

Leakage and Cavity Pressures in an Interlocking Labyrinth Gas Seal: Measurements Versus Predictions

Gas labyrinth seals (LS) restrict secondary flows (leakage) in turbomachinery and their impact on the efficiency and rotordynamic stability of high-pressure compressors and steam turbines can hardly be overstated. Among seal types, the interlocking labyrinth seal (ILS), having teeth on both the rotor and the stator, is able to reduce leakage up to 30% compared to other LSs with either all teeth on the rotor (TOR) or all teeth on the stator. This paper introduces a revamped facility to test gas seals for their rotordynamic performance and presents measurements of the leakage and cavity pressures in a five teeth ILS. The seal with overall length/diameter L/D = 0.3 and small tip clearance Cr/D = 0.00133 is supplied with air at T = 298 K and increasing inlet pressure Pin = 0.3–1.3 MPa, while the exit pressure/inlet pressure ratio PR = Pout/Pin is set to range from 0.3 to 0.8. The rotor speed varies from null to 10 krpm (79 m/s max. surface speed). During the tests, instrumentation records the seal mass flow (m˙) and static pressure in each cavity. In parallel, a bulk-flow model (BFM) and a computational fluid dynamics (CFD) analysis predict the flow field and deliver the same performance characteristics, namely leakage and cavity pressures. Both measurements and predictions agree closely (within 5%) and demonstrate that the seal mass flow rate is independent of rotor speed. A modified flow factor Φ¯=m˙T/(PinD1−PR2) characterizes best the seal mass flow with a unique magnitude for all pressure conditions, Pin and PR.

Subgrid-scale stress parameterization for anisotropic turbomachinery flow as deduced from stereoscopic particle image velocimetry measurements

The present study searches for parametric subgrid-scale models for the turbomachinery flow based on single-plane stereoscopic particle image velocimetry measurements. A group of deformation tensors are used as the predictors to parameterize the subgrid-scale stress tensor as the response. Missing terms in the rotation rate tensor are completed with a novel approach which is based on geometrical relations between deformation and subgrid-scale tensors. The optimal relations for parametric subgrid-scale models are shaped after minimizing the square of error between modeled and real SGS stresses. The most superior and inferior one-predictor models are based on strain rate and rotation rate tensors, respectively. The two-predictor models show the most efficient prediction of subgrid-scale stress where the correlation coefficient is up to 43% larger than that of the classical Smagorinsky model. The three- to five-predictor models are inefficient since they give minor enhancement of correlation coefficient with a considerably larger computational cost.

Picard–Newton iterative algorithm to solve the potential flow equation for different turbomachinery flow regimes

Computing the velocity field magnitude with prescribed velocities on the pressure and suction sides represents the most important step in the inverse design of transonic axial turbomachinery blade profiles. The objective is to develop a numerical method to solve the isentropic velocity field with prescribed velocities at the boundaries. We will present the computational methods developed to solve the potential flow equation leading to the isentropic velocity field, overcoming the problems linked to the presence of singularities at the leading and trailing edges. A combination of Picard and Newton iterative algorithms is used. The numerical algorithm is validated on subsonic and supersonic planar source flow test cases then used on real flow cases of subsonic compressors and transonic turbines. We are able to obtain fully converged isentropic velocity fields.

Leveraging LDV techniques for the investigation of unsteady turbomachinery flows

ABSTRACTSteps required for proper acquisition and processing of laser Doppler velocimetry data for turbomachinery research applications are addressed. Turbomachinery applications are difficult due to the small internal passages, high-frequency fluctuations, large turbulence intensities, and strong secondary flows resulting in low overall signal-to-noise ratios and narrowband noise sources that cannot be removed by simple band-pass filters. Special aspects that must be considered for successful and accurate laser Doppler velocimetry studies to be conducted in turbomachinery are discussed. Specifically, the design of the measurement volume size, reflection mitigation, engineering of seed particle size and injection schema, and alignment of the traverse mechanism are addressed in terms of their importance (from literature sources) and the solutions implemented by the authors. These techniques have been applied to successfully obtain three-component, unsteady velocity data in a high-speed centrifugal compressor for aeroengine application. Processing techniques are also presented including a novel mixture-model-based statistical method for narrowband noise isolation developed by the authors. The method, validation steps, and example results are presented, showing the successful rejection of noise with high accuracy, a low failure rate, and a significant reduction in required manual inspection. This newly developed method elucidated flow features that were not clear prior to the noise removal.

Experimental study of the unsteady vibration signature for a Sirocco fan unit

The small forward-curved blades known as Sirocco fan units are very common and widespread solution for air conditioning used in public transportation applications, as buses or trains. The users quietness and comfort have become a main concerns in the automotive industry. For such kind of turbomachinery flow , the patterns becomes always highly 3D and unsteady, compromising the referred comfort, and setting the focus on the working flow variables. A mathematically exact solution for that flow, which would provide any required information on pressure or forces, is out of scope at the current engineering design processes. Nevertheless, some flow features and mechanical data are needed to progress in the frame of a modern industrial environment, involving maintenance protocols with important temporal and economic constraints for different design procedures. The correctness of a given maintenance protocol relies on its feasibility to handle a set of machine working parameters or variables, including a number of them as wide as possible. Doing so, a set of not-dangerous ranges for them can be established. Such ranges are often defined promoting a series of failures similar to real ones, when the machine is in its operative lifetime. In this paper and in order to establish proper working ranges for maintenance purposes, a series of failures have been experimentally tested for a Sirocco fan unit. Initially, real data from industry have been required and a list of main failures was made, including (1) impeller or rotor unbalance, (2) impeller channel obstruction and (3) blocked inlet. The failures are studied using a purified orbit diagram (POD) technique and a symmetrized dot pattern (SDP) technique. All four working conditions are studied for at least three different flow rates and, therefore, a deeper insight into the fan working parameters and options are made feasible. In the frame of the maintenance protocol, a full set of ranges for the considered failures has been obtained. Therefore, the present paper shows a novel possibility to enhance existing maintenance protocol using two advanced frequency-based techniques.

Editorial

Editorial

A semi-empirical model for streamwise vortex intensification

Vortex intensification plays an important role in a wide range of flows of engineering interest. One scenario of interest is when a streamwise vortex passes through the contracting streamtube of an aircraft intake. There is, however, limited experimental data of flows of this type to reveal the dominant flow physics and to guide the development of vortex models. To this end, the evolution of wing-tip vortices inside a range of streamtube contractions has been measured using stereoscopic particle image velocimetry. A semi-empirical model has been applied to provide new insight on the role of vorticity diffusion during the intensification process. The analysis demonstrates that for mild flow contractions, vorticity diffusion has a negligible influence due to the low rates of diffusion in the vortex flow prior to intensification and the short convective times associated with the streamtube contraction. As the contraction levels increase, there is a substantial increase in the rates of diffusion which is driven by the greater levels of vorticity in the vortex core. A new semi-empirical relationship, as a function of the local streamtube contraction levels and vortex Reynolds number, has been developed. The model comprises a simple correction to vortex filament theory and provides a significant improvement in the estimation of vortex characteristics in contracting flows. For the range of contractions investigated, errors in the estimation of vortex core radius, peak tangential velocity and vorticity are reduced by an order of magnitude. The model can be applied to estimate the change in vortex characteristics for a range of flows with intense axial strain, such as contracting intake streamtubes and swirling flows in turbomachinery.

Numerical simulation using the bubble size distribution model of cryogenic cavitation around an axial helical inducer

A mathematical model using a bubble size distribution was developed by the author for a cryogenic high-speed cavitating flow in turbomachinery. In this model, bubble growth/decay is solved for each class of the bubble mass (as an accurate expression of bubble size) when bubbles with various masses mix in the same spatially discretized calculation region. The bubble growth/decay calculations employ a combination of two rigorous methods: a Rayleigh–Plesset equation for the bubbles’ oscillation, and a heat conduction equation in a thermal boundary layer around a bubble to evaluate the mass rate of evaporation or condensation. This mathematical model was coded using the OpenFOAM C++ toolbox, and a numerical simulation was conducted for a 3D rotational inducer in a cryogenic fluid, the results of which showed that bubbles of a certain specified size were distributed at certain regions on the inducer. This study demonstrated that the mathematical model of cavitation exhibits potential to contribute to fidelity in simulation for turbomachinery considering a fluid with strong thermodynamic effect.

Challenges in Modeling of Turbulence in the Tip Region of Axial Turbomachines

The flow and turbulence in turbomachines are characterized by spatially and temporally nonuniform strain rate fields and tendency to develop partially understood large-scale instabilities. Consequently, the applicability of popular Reynolds stress models in Reynolds-averaged Navier-Stokes simulations of turbomachinery flows is questionable and requires validation under relevant conditions. Experimental studies performed in the Johns Hopkins University optical refractive index-matched facility have investigated the complex flow and turbulence in the blade tip region of several axial turbomachines, including two waterjet pumps and an aviation compressor. The comprehensive databases obtained for different flow conditions and tip geometry using high-resolution particle image velocimetry are used in this article to examine the characteristics of turbulence in the vicinity of the tip leakage vortex (TLV). The results reveal several common features in the spatial distribution of the extremely anisotropic Reynolds stress tensor, suggesting that there are similarities in the mechanisms involved among the different machines. Of the quantities that could be measured accurately, both the local production and mean-flow advection rates appear to govern the evolution of turbulence in the passage. The dissipation rate also appears to be a major factor, but it cannot be measured accurately because of its dependence on motions at small scales that are not resolved by the measurements. The turbulent transport terms are important only in limited cases and locations. The strain rate tensor components, which appear to play primary roles in the production of turbulent kinetic energy (TKE), display common features in the spatial distributions of regions with high contraction, extension, and shear under different conditions. Some but not all of the inhomogeneity and anisotropy in the distributions of Reynolds stresses can be related to the corresponding local production rates. There is disagreement near the TLV core, where the persistent presence of multiple interacting vortices contributes significantly to the elevated TKE there. Lack of a clear functional correlation between the Reynolds stresses and mean strain rates is ubiquitous in all cases, resulting in widely spatially varying positive and negative values for the eddy viscosity. However, surprisingly, results from different machines and conditions have similar spatial distributions of individual stress and strain rate components and, accordingly, the eddy viscosity.

Advanced application of a sublayer fence probe in highly instationary turbomachinery flows: observations on prestall instabilities

The present paper introduces a novel approach for measuring near wall flow velocities by utilizing a sublayer surface fence probe. Hereby, a difference in static pressure builds up over a microscopic obstacle within the viscous sublayer. An analytical model of the angular dependent pressure difference is employed to derive information about the flow direction. Furthermore, a computational preston tube approach has been used to calibrate the surface fence probe with regard to a flow velocity to be assigned to half of the fence height. Through the use of a sophisticated analyzing algorithm the flow direction and its velocity can be determined as a function of time. As a proof of concept measurements were conducted within the radial gap of an annular compressor rig yielding both mean and time resolved near wall flow fields. Former are in very good compliance with oil flow visualizations proving the methods accuracy. The experimental results give unprecedented insights into an unsteady flow phenomenon that arises when an axial compressor is operating close to its stalling limit. The presented technique allows for investigating turbomachinery areas which formerly were observable only by computational means.Graphical

The Current State of High-Fidelity Simulations for Main Gas Path Turbomachinery Components and Their Industrial Impact

Over the past two decades high-fidelity simulations have become feasible for most main gas path turbomachinery components. This paper introduces the key challenges of simulating and modelling turbomachinery flows and presents an overview of possible simulation strategies. A comprehensive overview is given of which particular design challenges have to date been addressed by high-fidelity simulations, with a particular focus on high-pressure and centrifugal compressors, and high-pressure and low-pressure turbines. Recommendations are provided for how quality and accuracy can be ensured for high-fidelity simulations, using direct numerical simulations of a low-pressure turbine as a case study. It is argued that industrial impact from high-fidelity simulations can be achieved in two ways, either by conducting design-of-experiment-like studies that can provide designers with insight into certain physical mechanisms and phenomena, or by helping mature and improve lower-order models. The latter is discussed with particular emphasis on recent advances in machine learning for model assessment and improvement, and the potential of one selected data-driven approach is demonstrated on predictions of wake mixing for low-pressure and high-pressure turbines.

Sir William Rede Hawthorne. 22 May 1913—16 September 2011

William Rede Hawthorne was a pioneer in gas turbine aerodynamics and thermodynamics, a sought-after technology advisor to industry and government, and a generous and enthusiastic teacher who encouraged students to excel. His outstanding contributions included resolution of combustion problems that limited the operation of the original Whittle jet engine, early in-depth descriptions of compressible channel flow that still inform engineers today, innovative and wide-ranging analyses of secondary flows in turbomachinery that defined the field, and creation of some of the first notes on gas turbine cycle analysis. A theme in the many areas of engineering in which he had impact was the satisfaction from the growth of understanding that can accompany making things work—in his words, ‘machines produce ideas just as surely as ideas produce machines’. A Cambridge graduate, he was a professor at MIT when, in 1951, he was recruited to a newly established chair at Cambridge, where he later had leadership roles as head of the engineering department (1968–73) and Master of Churchill College (1968–83). He retained strong ties to MIT, however, and fostered lasting collaborations between the two universities. Among his numerous awards and honours were the US Medal of Freedom (1947), a Royal Society Medal (1982) and a knighthood (1970) for ‘services to thermodynamics’, a citation that pleased him greatly.

A two-level computational graph method for the adjoint of a finite volume based compressible unsteady flow solver

Abstract The adjoint method is a useful tool for finding gradients of design objectives with respect to system parameters for fluid dynamics simulations. But the utility of this method is hampered by the difficulty in writing an efficient implementation for the adjoint flow solver, especially one that scales to thousands of cores. This paper demonstrates a Python library, called adFVM, that can be used to construct an explicit unsteady flow solver and derive the corresponding discrete adjoint flow solver using automatic differentiation (AD). The library uses a two-level computational graph method for representing the structure of both solvers. The library translates this structure into a sequence of optimized kernels, significantly reducing its execution time and memory footprint. Kernels can be generated for heterogeneous architectures including distributed memory, shared memory and accelerator based systems. The library is used to write a finite volume based compressible flow solver. A wall clock time comparison between different flow solvers and adjoint flow solvers built using this library and state of the art graph based AD libraries is presented on a turbomachinery flow problem. Performance analysis of the flow solvers is carried out for CPUs and GPUs. Results of strong and weak scaling of the flow solver and its adjoint are demonstrated on subsonic flow in a periodic box.

Application and adaption of Dynamic Mode Decomposition to experimental and numerical data of turbomachinery flow

AbstractUnsteady turbomachinery flows are challenging in terms of resolving turbulent characteristics and instability mechanisms like rotating stall, rotor‐stator interaction and von Karmán vortex streets. Furthermore, in part load conditions these flow phenoma can induce vibrations leading to resonance problems and high cycle fatigue. Flowfield measurements of a centrifugal fan were taken by high‐speed particle image velocimetry (PIV). CFD results, calculated as unsteady Reynolds averaged Navier Stokes (URANS) and based on Menter's shear‐stress transport (SST) model, are the source of numerical data. In former examinations discrepancies between experiment and simulation in time‐averaged velocity profil data for part load conditions have been stated. The source of theses discrepancies is still unknown. Hence, a new approach is needed for further examination of measured and calculated flowfields. The Dynamic Mode Decomosition is employed in order to compare experimental and numerical datasets. Information about coherent structures and the dynamical behaviour are contained in the eigenvalues and their associated eigenvectors offered by DMD. While the eigenvectors are describing the coherent structures, the associated eigenvalues describe the dynamic behaviour of the structure. The eigenvectors are denoted as DMD‐Modes. The results acquired by DMD based on experimental and numerical data are compared to investigate congruent behaviour and therefore if it is possible to obtain reliable information about dynamic behaviour of a system even with URANS simulations. Comparisons between singular value and eigenvalues are presented and discussed.

On the dynamic modeling of Brayton cycle power conversion systems with the CATHARE-3 code

As the share of intermittent energy sources is increasing, the dynamic modeling of thermal power plants including their power conversion system gets researchers' attention. Recent applications require transient simulations of high-pressure nitrogen closed Brayton cycle for which the ideal gas assumption is no longer verified. In this work, the REFPROP real gas equation of state and pressure drop correlations suitable for high Reynolds numbers have been implemented in the CATHARE-3 code. Moreover, new real gas turbomachinery and sonic flow models have been developed and integrated in the code. Relative errors obtained for the nominal state of a high-pressure nitrogen closed Brayton cycle are in the range −3%/+0.8%. A detailed analysis of real gas effects is carried out on the heat exchangers heat flux and the piping pressure losses. The new sonic flow computed during a loss of coolant accident is in the best possible agreement with literature experimental results. With regard to the turbomachinery, the new real gas model creates a pressure dependence that brings compressors closer to the choke region when the pressure drops in the cycle. This work is expected to provide an efficient and reliable simulation tool for transient analysis of real gas closed Brayton cycles.

A modified IWAN model for micro-slip in the context of dampers for turbine blade dynamics

Abstract In turbine blade systems, under-platform dampers are widely used to attenuate resonant vibrations and prevent high cycle fatigue failure depending on friction contact. To effectively predict dynamic response of blade-damper system, a numerical model of friction contact between damper and blade is required. The majority of existing friction contact model either cannot reproduce micro-slip behavior or include too many contact parameters, which restricts their application. The proposed model, a modification of the IWAN model, not only describes the tangential micro/gross-slip and normal contact/separation motions, but also has the same number of parameters as the classical gross-slip model. A group of microslip-dominated experimental results for a laboratory crossed curve-flat under-platform damper are shown, and a specially developed procedure is used to determine the elastic contact parameters in order to validate the proposed model. It is shown that the proposed model is able to correctly represent the damper’s energy dissipation under low vibration amplitudes. The relative difference of energy dissipation between experiment and simulation by the proposed model is about 1.15%, far less than 24% according to the gross-slip model. Furthermore, the relative simple expression of the model is fully compatible with the current state-of-the-art in the numerical analysis of resonance damping of bladed disks for axial flow turbo-machinery.

Volumetric Velocimetry Measurements of Film Cooling Jets

This paper presents volumetric velocimetry (VV) measurements for a jet in crossflow that is representative of film cooling. VV employs particle tracking to nonintrusively extract all three components of velocity in a three-dimensional volume. This is its first use in a film-cooling context. The primary research objective was to develop this novel measurement technique for turbomachinery applications, while collecting a high-quality data set that can improve the understanding of the flow structure of the cooling jet. A new facility was designed and manufactured for this study with emphasis on optical access and controlled boundary conditions. For a range of momentum flux ratios from 0.65 to 6.5, the measurements clearly show the penetration of the cooling jet into the freestream, the formation of kidney-shaped vortices, and entrainment of main flow into the jet. The results are compared to published studies using different experimental techniques, with good agreement. Further quantitative analysis of the location of the kidney vortices demonstrates their lift off from the wall and increasing lateral separation with increasing momentum flux ratio. The lateral divergence correlates very well with the self-induced velocity created by the wall–vortex interaction. Circulation measurements quantify the initial roll up and decay of the kidney vortices and show that the point of maximum circulation moves downstream with increasing momentum flux ratio. The potential for nonintrusive VV measurements in turbomachinery flow has been clearly demonstrated.

On the challenge of five-hole-probe measurements at high subsonic Mach numbers in the wake of transonic turbine cascades

Five-hole-probes are common use in turbomachinery flow investigations, even though, inserting a probe into a flow field inevitably induces perturbations to the flow which can falsify the measurement results, especially when exposed to transonic flows. The objective of the investigations presented here is to evaluate the Mach number measurements of a five-hole-probe (5HP) in the wake flow of a transonic turbine cascade at engine relevant Reynolds numbers by comparing them to the results of particle image velocimetry (PIV). Furthermore, PIV measurements were performed with inserted probe to investigate the influence of the probe on the wake flow field. Together with a sensitivity study of 5HP measurements in flow regimes close to Ma = 1, the results demonstrate how the measurement uncertainty can be improved in high subsonic flow regimes.

A characteristic inlet boundary condition for compressible, turbulent, multispecies turbomachinery flows

A methodology to implement non-reflecting boundary conditions for turbomachinery applications, based on characteristic analysis is described in this paper. For these simulations, inlet conditions usually correspond to imposed total pressure, total temperature, flow angles and species composition. While directly imposing these quantities on the inlet boundary condition works correctly for steady RANS simulations, this approach is not adapted for compressible unsteady Large Eddy Simulations because it is fully reflecting in terms of acoustics. Deriving non-reflecting conditions in this situation requires to construct characteristic relations for the incoming wave amplitudes. These relations must impose total pressure, total temperature, flow angle and species composition, and simultaneously identify acoustic waves reaching the inlet to let them propagate without reflection. This treatment must also be compatible with the injection of turbulence at the inlet. The proposed approach shows how characteristic equations can be derived to satisfy all these criteria. It is tested on several cases, ranging from a simple inviscid 2D duct to a rotor/stator stage with turbulence injection.

Aero-thermal flow characterization downstream of an NGV cascade by five-hole probe and filtered Rayleigh scattering measurements

The characterization of pressure, temperature and velocity fields in turbomachinery flows typically relies on well-proven probe-based technology such as pneumatic probes, hot-wire sensors or thermocouples. These devices have to be introduced into narrow flow channels and by that obstruct part of the duct at the actual measuring position, which can significantly alter the aerodynamic performance of the components under investigation. In this contribution, measurement results of a commercially available five-hole probe with mounted temperature sensor and laser-optical-filtered Rayleigh scattering diagnostics are acquired downstream of a nozzle guide vane cascade with lean-burn combustion representative inlet flow distortions. Pressure results obtained by both methods are found to be on a similar absolute level. However, the unavoidable movement of optical elements following the facilities’ heat-up procedure prevents a sufficient sensitivity of pressure data obtained by filtered Rayleigh scattering and, therefore, an identification of certain flow structures related to probe interference on the static pressure field. Concerning temperature data, results of both methods, despite the probe data being shifted in lateral direction, are found to be in good agreement in topology as well as absolute level. The strongest intrusive influence of the probe body as well as a bias to probe readings in certain flow regions is identified when comparing velocity data by means of the optical Doppler frequency-shift. Whereas filtered Rayleigh scattering results follow a smooth and continuous trend, five-hole probe data are characterized by steep gradients inside the airfoil wakes as well as differing slopes in the flow passages. These findings are backed by a comparison of five-hole probe velocities with a CFD solution, revealing an erroneous reading of axial velocities inside the wake areas. The study concludes that the presented filtered Rayleigh scattering diagnostics, with estimated accuracies of 50–90 hPa, 2–3 K and 2–3 MHz in pressure, temperature and Doppler frequency-shift, is a viable alternative to conventional probe-based measuring approaches in characterizing the aero-thermal properties of turbomachinery flows.