0

IN THIS VOLUME


Turbomachinery

1988;():V001T01A001. doi:10.1115/88-GT-4.
FREE TO VIEW

Up to now the reasons of reducing the secondary flow losses by using leaned blades, especially in the case of great incidence angle, have been little concerned with in published references. The experimental results in this paper have shown that the decisive factor reducing secondary flow losses in turbine stator cascades is the static pressure gradient along the blade height inside the cascade channel near the suction surface, especially in the rear part of it, because the negative pressure gradient in the hub region and the positive one in the tip region are beneficial for the boundary layer in both the regions to be sucked into the main stream region, consequently, the accumulation and the separation of the boundary layer have been weakened in both regions. Moreover, the effectiveness of applying the positively or negatively leaned blades is increased with the increase of a incidence angle, in the hub or the tip regions respectively.

Commentary by Dr. Valentin Fuster
1988;():V001T01A003. doi:10.1115/88-GT-19.
FREE TO VIEW

Five types of cascades with different leaned blades have been tested in a low speed annular cascade tunnel. The experimental results show that by using positively leaned blades the hub wall boundary layer in cascade passage can be sucked into main stream zone, and the amount of low energy gas getting into downstream of the cascade is reduced obviously, so that the energy losses downstream is decreased considerably.

Topics: Blades
Commentary by Dr. Valentin Fuster
1988;():V001T01A006. doi:10.1115/88-GT-31.
FREE TO VIEW

The flow field within an axial flow inducer pump near the blade leading edge was explored by laser-Doppler velocimetry to extend the previous studies of the recirculation zone which is observed at low flow rates. Although a considerable region of upstream reverse flow and swirl was observed, the recirculation zone within the impeller was of limited axial extent and was confined to the pressure side of the passage. In an attempt to reduce the flow reversal, a series of perforated disks were placed in front of the inducer. The optimum disk geometry produced minor changes in the pump performance. LDV measurements of the flow field ahead and behind the disk showed considerable reduction of the swirl velocity under reverse flow conditions, with the observed upstream swirl opposite to the inducer rotation.

Topics: Pumps
Commentary by Dr. Valentin Fuster
1988;():V001T01A007. doi:10.1115/88-GT-32.
FREE TO VIEW

Estimating the low Reynolds number and off-design performance of axial turbomachine blades requires an accurate prediction of separation phenomena occurring on the blade surface. This paper discusses a viscous/inviscid interaction analysis of flow over a NACA 65-213 airfoil at a chord Reynolds number of 240,000 using a calculation method of Cebeci et al. The computed characteristics of a mid-chord laminar separation bubble are compared with experimental laser-doppler anemometer measurements of Hoheisel et al. Attention is focused on problems of modeling the laminar-turbulent transition zone within the viscous layer.

A parametric study is undertaken to determine the location and extent of the transition zone which best models the observed separation bubble behavior. The required transition length is almost an order of magnitude smaller than that predicted from conventional transition length correlations. A physical model for this greatly reduced transition length in positive pressure gradient flows is proposed.

The computational model correctly predicts most features of the separation bubble flow, but there are some significant discrepancies at reattachment which point to the need for improved turbulence modeling in this area. The inclusion of transverse pressure gradients associated with flow curvature in the viscous regions also appears very desirable for airfoils operating at Reynolds numbers around 105.

Commentary by Dr. Valentin Fuster
1988;():V001T01A014. doi:10.1115/88-GT-49.
FREE TO VIEW

Predictions of the laminar flow and temperature distribution between coaxial rotating disks of finite radius have been obtained using a computer model solving the Navier-Stokes equation for a laminar fluid of constant properties. To do this, a stream vorticity model in the r-z plane is used in the solution of the Navier-Stokes equation. The velocity fields were obtained for both shrouded and unshrouded disks with or without radial throughflow for either co-rotating or counter-rotating disks. Velocity profiles predicted by this model were compared to experimental data, to a similarity solution and to a large aspect ratio model. The results obtained by this model closely matched the experimental data, and the large aspect ratio solution for the cases considered. The uncoupled energy equation was then solved using the calculated velocity distribution for the temperature distribution between the disks. This was done for two cases: i) two isothermal disks, and ii) one isothermal disk and one adiabatic disk.

Commentary by Dr. Valentin Fuster
1988;():V001T01A021. doi:10.1115/88-GT-59.
FREE TO VIEW

This paper presents a numerical analysis and experimental measurements of flows in the vaneless volute of a radial–inflow turbine. Cases of single and twin entry to the volute were studied, and in particular the case of unequal flows, or partial admission, to the two entries. The distance between the rotor tip and the limit of the internal wall separating the two flows was varied in order to study the effects of interaction of the two streams.

The computational model is based on a quasi–three–dimensional solution of the Euler equations, in which the radial and tangential components of velocity are fully solved, but the axial component is only treated to simulate the mixing of the two streams. The results of the model were compared with published results for a single entry, and extensive new results for a single or twin entry, casing. Even in the single and equal flow twin–entry cases, significant variations in flow properties around the exit circumference of the volutes were observed, most particularly in the region of re–entry near the tongue. For unequal flows the interaction of the two streams was strong and increased with increasing separating wall diameter, and with increasing inequality of flow rates. In extreme cases the flow reverses.

Commentary by Dr. Valentin Fuster
1988;():V001T01A022. doi:10.1115/88-GT-60.
FREE TO VIEW

Based on the method of Casey (1), the present paper makes further development in using Bezier polynomial patches to define the geometrical shape of the flow channels of compressors. There are two advantages in this new method. First, in the geometrical construction process the blade profile and the contour of impeller’s meridional channel can be given by the same number of patches of the same degree so as to avoid more complicated repeat computations. Second, for curved surface construction there are no restrictions of linearity in the spanwise direction in order to match the distribution of flow angles at the inlet better and to fit the needs of various load models. Some examples have shown that the shapes produced by Bezier polynomial patches are general enough to be used in the design of new compressors and in approximation of the geometry of existing ones. The method is particularly suitable for incorporation into a computer-aided design procedure.

Commentary by Dr. Valentin Fuster
1988;():V001T01A023. doi:10.1115/88-GT-61.
FREE TO VIEW

The problem of radial inflow between two plane co-rotating discs with the angular velocity of the fluid at inlet equal to that of the discs is considered. An integral solution technique for turbulent flow, based on that of von Karman (1921), is described. Solutions are shown to be in good agreement with most of the available experimental data. For incompressible flow the pressure drop coefficient is a function of just two non-dimensional parameters; the radius ratio for the cavity and a throughflow parameter. For air flows compressibility can be important and an additional non-dimensional parameter is needed. Results for a wide range of conditions are presented graphically. These show the sensitivity of the pressure coefficient to the governing parameters and provide a quick method for estimating the pressure drop.

Topics: Pressure , Disks , Inflow
Commentary by Dr. Valentin Fuster
1988;():V001T01A024. doi:10.1115/88-GT-62.
FREE TO VIEW

As an extension of the author’s previous work, this paper proposes a mixed iterative method for calculating the interactions of boundary layer with inviscid flow on S1 stream surface. It is shown from the comparison with experimental results that the iteration method given in this paper has a high calculation accuracy. In addition, the rate equation of the entrainment coefficient adapted for the turbulent boundary layer in the S1 stream surface has been established in this paper. The predicted results show that taking this equation as the supplemental equation for solving the turbulent boundary layer on cascade blades can improve the accuracy of calculation of the turbulent boundary layer on cascade blades.

Commentary by Dr. Valentin Fuster
1988;():V001T01A025. doi:10.1115/88-GT-63.
FREE TO VIEW

The flow field inside a seven cavity labyrinth seal with a 0.00127 m clearance was measured using a 3-D laser doppler anemometer system. Through the use of this system, the mean velocity vector and the entire Reynolds stress tensor distributions were measured for the first, third, fifth, and seventh cavities of the seal. There was one large recirculation region present in the cavity for the flow condition tested, Re = 28,000 and Ta = 7,000. The axial and radial mean velocities as well as all of the Reynolds stress term became cavity independent by the third cavity. The azimuthal mean velocity varied from cavity to cavity with its magnitude increasing as the flow progressed downstream.

Topics: Lasers
Commentary by Dr. Valentin Fuster
1988;():V001T01A026. doi:10.1115/88-GT-64.
FREE TO VIEW

The flow field inside an annular seal with a 0.00127 m clearance was measured using a 3-D laser Doppler anemometer system. Through the use of this system, the mean velocity vector and the entire Reynolds stress tensor distributions were measured for the entire length of the seal (0.0373 m). The seal was operated at a Reynolds number of 27,000 and a Taylor number of 6,600.

Topics: Lasers
Commentary by Dr. Valentin Fuster
1988;():V001T01A027. doi:10.1115/88-GT-67.
FREE TO VIEW

A numerical method was developed to consider the two-dimensional flowfield between impeller blades of a given geometry. Solution of the laminar Navier-Stokes equations in geometry-oriented coordinates was obtained for stream functions and vorticities. Velocities and pressures were calculated to determine the output fluid-energy head. The circumferential components of the normal and shear stresses along the blade were evaluated to give the input mechanical-energy head. Performance predictions were obtained for different load conditions. Comparisons were made with the measured velocity vectors of the flowfield of an air-pump impeller and with the measured performance of a production water pump, good agreements were reached.

Commentary by Dr. Valentin Fuster
1988;():V001T01A028. doi:10.1115/88-GT-68.
FREE TO VIEW

A radial mixing calculation method is presented where both convective and turbulent mixing processes are included. The secondary flows needed for the convective mixing are derived from pitch averaged vorticity equations combined with integral methods for the 3D end-wall boundary layers, 3D profile boundary layers and 3D asymmetric wakes. The convective transport due to secondary flows is computed explicitly. The method is applied to a cascade and two single stage rotors. The three test cases show a very different secondary flow behaviour which allows the analysis of the relative importance of the different secondary flow effects. Turbulent diffusion is found to be the most important mixing mechanism, whereas convective mixing becomes significant when overall radial velocities exceed about 5% of the main velocities. The wake diffusion coefficient is found to be representative for the turbulent radial mixing and is the only empirical constant to be determined.

Commentary by Dr. Valentin Fuster
1988;():V001T01A029. doi:10.1115/88-GT-69.
FREE TO VIEW

It is the purpose of this paper to examine the flow fields in an advanced modern transonic rotor design using both axisymmetric and three dimensional techniques. Also, to determine the deviation of the axisymmetric flow from three-dimensional flow field and whether this seriously affects the results. Inviscid Euler solvers are now widely used to analyze transonic flows through turbomachines giving a reasonably accurate indication of the flow field in blade passages. Although viscous effects are important, the inviscid analysis provides significant knowledge of the flow field which is essential to transonic design. The blade-to-blade loading and work distributions are determined quite realistically by the 3-D and quasi-3-D inviscid analyses. Through-flow and blade-to-blade inviscid solutions are presented for a highly loaded transonic rotor. Numerical solutions for various transonic rotor designs operating at peak efficiency are also compared with test data.

Topics: Design , Rotors
Commentary by Dr. Valentin Fuster
1988;():V001T01A030. doi:10.1115/88-GT-70.
FREE TO VIEW

This paper describes the current stage of development of a code aimed at solving the 3D Navier-Stokes equations in any type of turbomachinery geometry. The basic algorithm time marches the fully 3D unsteady equations of motion expressed in finite volume form with a two step explicit / one step implicit method. Full multigrid acceleration is used to reduce solution time and maintain code performance on fine meshes. Turbulence modelling is via mixing-length closure and the widely used Baldwin-Lomax model.

The generality and robustness of the code is demonstrated by application to five different test cases, three axial and two radial configurations. Also included is a grid independence study which demonstrates near grid independent solutions for transonic compressor cascade flow (albeit with the actual result subject to transition modelling constraints).

For two of the axial cases (transonic compressor in cascade, secondary flow in a high speed compressor) and one radial case (Eckardt high speed impellor) sufficient mesh is employed for the predictions to be essentially quantitative. The other two cases (radial inflow turbine with clearance and compressor stator with hub clearance) are really simulations rather than predictions, but are included as the flows are novel and provide much physical insight.

Topics: Turbomachinery
Commentary by Dr. Valentin Fuster
1988;():V001T01A036. doi:10.1115/88-GT-82.
FREE TO VIEW

A non-isentropic potential method on a S2 stream surface has been developed for the design and analysis of transonic compressors with shocks, in which the entropy increase across a shock may be directly calculated from the momentum equations in the divergence form. The numerical results show that the non-isentropic shock is weaker and placed one or two meches further upstream compared to the classical potential calculation, and is in good agreement with the experimental data.

Commentary by Dr. Valentin Fuster
1988;():V001T01A037. doi:10.1115/88-GT-83.
FREE TO VIEW

This paper describes an algorithm for computing two-dimensional transonic, inviscid flows. The solution procedure uses an explicit Runge-Kutta time marching, finite volume scheme. The computational grid is an irregular triangulation. The algorithm can be applied to arbitrary two-dimensional geometries. When used for analyzing flows in blade rows, terms representing the effects of changes in streamsheet thickness and radius, and the effects of rotation, are included. The solution is begun on a coarse grid, and grid points are added adaptively during the solution process, using criteria such as pressure and velocity gradients.

Advantages claimed for this approach are (a) the capability of handling arbitrary geometries (e.g., multiple, dissimilar blades), (b) the ability to resolve small-scale features (e.g., flows around leading edges, shocks) with arbitrary precision, and (c) freedom from the necessity of generating “good” grids (the algorithm generates its own grid, given an initial coarse grid).

Solutions are presented for several examples that illustrate the usefulness of the algorithm.

Topics: Inviscid flow
Commentary by Dr. Valentin Fuster
1988;():V001T01A038. doi:10.1115/88-GT-89.
FREE TO VIEW

This paper is a contribution to the study of the generation of unsteady forces on turbine blades due to potential-flow interaction and viscous-wake interaction from upstream blade rows. A computer program is used to compute the unsteady forces on a rotor. The accuracy of the computer program is tested by comparing the results of a steady-flow calculation case and of an unsteady-flow calculation case with theory and experiment respectively. Results are shown for typical stator-to-rotor-pitch ratios and stator outlet-flow angles. These results show that the first spatial harmonic of the unsteady force may decrease for higher stator-to-rotor-pitch ratios. This trend is explained by considering the mechanisms by which the unsteady forces are generated. In this paper the mechanism by which the potential-flow interaction affects the flow field to generate these unsteady forces is shown to vary with the stator-to-rotor-pitch ratio and with the outlet flow angle of the stator.

Commentary by Dr. Valentin Fuster
1988;():V001T01A039. doi:10.1115/88-GT-90.
FREE TO VIEW

This paper is a contribution to the study of the generation of unsteady forces on turbine blades due to viscous wake interaction and potential-flow interaction from upstream blade rows. A computer program is used to compute the unsteady forces on a rotor. Typical results for isolated viscous-wake interaction (no potential-flow interaction) are shown. These results indicate that the first spatial harmonic of the unsteady force may decrease for higher stator-to-rotor-pitch ratios. This trend is explained by considering the mechanisms by which the unsteady forces are generated. The mechanism by which the viscous wakes affect the flow field to generate these unsteady forces is shown to vary with the stator-to-rotor-pitch ratio and with the outlet flow angle of the stator. It is also shown that by varying the axial gap between rotor and stator one can attempt to minimize the magnitude of the unsteady part of the forces generated by the combined effects of viscous-wake interaction and potential-flow interaction.

Topics: Wakes , Gas turbines , Blades
Commentary by Dr. Valentin Fuster
1988;():V001T01A040. doi:10.1115/88-GT-92.
FREE TO VIEW

Results of a continious effort at such a correlation, involving the most important effects of tip clearance on overall efficiency deterioration, are presented. The term tip clearance refers to the clearance between the vane tips of a half shrouded impeller and the front casing. In an attempt to develop a simple and more universal approach for the estimation of tip clearance losses in centrifugal impellers, the possibility of correlation through specific speed is examined. A tip clearance loss sensitivity factor is introduced and its likely dependence on specific speed presented. The weakness and ambiguity of the correlcation are discussed.

Commentary by Dr. Valentin Fuster
1988;():V001T01A041. doi:10.1115/88-GT-93.
FREE TO VIEW

The flow within a centrifugal rotor has strong characteristics of three-dimensional effect. A procedure called “stream-surface coordinates iteration” for the calculation of complete three dimensional flow in turbo-machinery is first described. Splitter blade techniques have been used in many rotors, especially in centrifugal compressors and pumps with high flow capacity. The difficulty of the calculation of the flow field for this type of rotor lies on that the mass flow ratio between the two sub-channels is unknown for the given total flow capacity. In the second part of this paper, an assumption about how to determine this mass flow ratio and a procedure to calculate the complete three-dimensional flow are presented. Finally, some design criteria about the splitter blades are put forward. Experimental data from two centrifugal pump impellers equipped with different splitter blades are also given to demonstrate the availability of the present calculation method.

Commentary by Dr. Valentin Fuster
1988;():V001T01A044. doi:10.1115/88-GT-100.
FREE TO VIEW

To help with the design of axial flow steam turbine blading a suite of flow analysis programs has been adapted and developed. Effort has been concentrated on improving the blade-to-blade analysis and developing a two-dimensional blade-to-blade profile design method. The development and verification of the analysis program have already been reported in detail (Willis, 1987 and Willis and Goulas, 1987). This paper presents the design or inverse solution. The analysis method uses an inviscid stream function solution coupled with an integral boundary layer calculation. In the design program the required changes in the blade geometry are effected via a transpiration type model. It is therefore a ‘profile refinement’, rather than an ‘original’ design procedure, and is necessarily an iterative solution. A required velocity distribution may be specified over only part of the blade surface. Two examples are presented in this paper to illustrate the capability of the design program.

Commentary by Dr. Valentin Fuster
1988;():V001T01A045. doi:10.1115/88-GT-101.
FREE TO VIEW

An elliptic flow calculation procedure has been used to model 3-D flow in the NACA 48–inch centrifugal impeller. The results demonstrate that fully elliptic steady flow calculations can be performed at design and off-design conditions. The calculations reproduce the measured overall performance and most of the features of the loss distributions observed in the NACA flow study. They give further insight into the complex 3-D flow with leading-edge separation and tip leakage.

The calculated secondary flow patterns are presented and used to explain the convection of vortices in a more recent laser anemometry study of a centrifugal compressor impeller.

Commentary by Dr. Valentin Fuster
1988;():V001T01A046. doi:10.1115/88-GT-103.
FREE TO VIEW

The flow through an impeller of a low speed radial-inflow turbine has been analysed using a fully three-dimensional viscous program and good correlations with instantaneous measurements of casing static pressure and exit flow distribution have been obtained. The flow at the exit of the turbine shows a pronounced non-uniformity with a wake region of high absolute flow angle near the casing. The predictions show that the flow is fully attached inside the impeller, while secondary flows can be observed especially in the exducer moving low momentum fluid towards the casing-suction corner. The presence of these secondary flows is discussed with reference to classical secondary flow theory. However, the comparison of measurements and numerical predictions indicate that the wake flow pattern is only partly due to the secondary flow. It is shown that in fact the tip leakage flow also plays a significant role in the wake generation and correspondingly some modelling of the leakage flow is essential in any attempted numerical simulations.

Commentary by Dr. Valentin Fuster
1988;():V001T01A048. doi:10.1115/88-GT-111.
FREE TO VIEW

This paper reports results from investigations conducted to determine the effect of stage loading on endwall flows in a low speed axial flow compressor. These investigations consisted of two sets of measurements. The first set consisted of radial tranverse of flow properties at the rotor inlet and exit, at five flow coefficients. These measurements are used to determined the boundary layer integral parameters. The displacement thicknesses at the rotor hub and tip agree reasonably well with Smith’s (1970) correlation for multistage axial compressors. The second set consisted of measurements of static pressures on the rotor blade at four flow coefficients. From these measurements lift coefficient is determined. Also loss of lift coefficient near the tip is calculated and is attributed mainly to the tip leakage flows.

Commentary by Dr. Valentin Fuster
1988;():V001T01A049. doi:10.1115/88-GT-118.
FREE TO VIEW

A method for solving various aerodynamic problems of transonic cascade design is developed, in which the coordinates equation is derived directly by inverting the stream function and the principal equation of the stream function. The corresponding “boundary conditions are obtained similarly. The coordinates equation has the same type property as that of the original equation and the only difference for various aerodynamic problems is the change of the boundary conditions on the cascade surfaces. Numerical results show that this unified method offers a useful tool for transonic cascade design.

Commentary by Dr. Valentin Fuster
1988;():V001T01A051. doi:10.1115/88-GT-124.
FREE TO VIEW

A development is undertaken to determine possible configurations for multiple row or tandem blades for blowers. For the individual rows of the airfoils, changes in blade number, blade camber, and chord length are analyzed. Modifications in the arrangement between two blade rows are made and tests are presented to determine the optimum design. Tests are shown regarding the effects of changes in blade number and blade solidity. The tests indicate that two row blades are capable of larger flow deflection with associated flow deceleration than single airfoils. The development makes it possible to reduce the number of stages for multistage blowers.

Topics: Blades
Commentary by Dr. Valentin Fuster
1988;():V001T01A052. doi:10.1115/88-GT-127.
FREE TO VIEW

Finite difference solutions are presented for turbulent flow in the cavity formed between a rotating and a stationary disc, with and without a net radial outflow of fluid. The mean flow is assumed steady and axisymmetric and a mixing length model of turbulence is used. Grid dependency of the solutions is shown to be acceptably small and results are compared with other workers’ experimental data. Theoretical and measured disc moment coefficients are in good agreement, while theoretical and measured velocities are in reasonable agreement. It is concluded that the mixing-length model is sufficiently accurate for many engineering calculations of boundary layer dominated flows in rotating disc systems.

Commentary by Dr. Valentin Fuster
1988;():V001T01A053. doi:10.1115/88-GT-132.
FREE TO VIEW

The problem of base pressure in the absence of base bleed has been investigated for a symmetrical model with a square-cut trailing edge. The model was mounted at zero angle of attack in a transonic wind tunnel covering a range of mainstream Mach number from 0.6 to 1.3. The role of vortex shedding was found to be of great importance in the prediction of base pressure. A semi-theoretical analysis for the prediction of base pressure in subsonic and transonic speeds which includes the effect of vortex shedding is proposed.

Topics: Pressure
Commentary by Dr. Valentin Fuster
1988;():V001T01A054. doi:10.1115/88-GT-134.
FREE TO VIEW

A new procedure employed in computer-aided design of centrifugal compressor stage to determine its over-all dimensions is described in this paper. By the use of the COMPLEX METHOD, the arbitrary number of variables to be optimized can be specified to remove the hidden danger of the local optima which stems from adopting a few, for example two or three, variables to be optimized. This procedure is available for any complicated implicit nonlinear objective function and ensures establishment of a true optimum solution. Numerical calculations have been carried out by using the computer program described here to check the ability of the optimization method. The results obtained by the calculations agree fairly well with that obtained by experiments.

Commentary by Dr. Valentin Fuster
1988;():V001T01A055. doi:10.1115/88-GT-136.
FREE TO VIEW

In this work, a family of methods, both blade-to-blade surface and 3D, based on the numerical integration of the unsteady Euler equations, are used in studying various aspects of the unsteady aerodynamics of vibrating compressor cascades, in the supersonic flutter region. Most aerodynamic methods assume a traveling wave assembly mode of structural vibration, and suppose that the associated chorochronical periodicity is also encountered in the flowfield. This hypothesis has been tested by simulating the flow in a full annular cascade and has been verified in all the cases we have studied. When analyzing the aeroelastic stability of cascades, some assembly modal basis must be used. This, especially in presence of mistuning, relies on modal superposition hypotheses and linearity assumptions. The superposition assumption seems to be justified and the linear range of the amplitude-response is fairly large, although it varies greatly with frequency. Finally, an assessment of the importance of unsteady 3D effects is attempted using the 3D method.

Commentary by Dr. Valentin Fuster
1988;():V001T01A056. doi:10.1115/88-GT-137.
FREE TO VIEW

An experimental study on the production and development of streamwise vortices in a compressor cascade is reported.

At four locations inside and one location outside the blade passage, the mean velocity components, three turbulent intensities and three Reynolds stresses were measured with a “x” hot wire probe. The results obtained describe the flow structure in the corner between the end-wall and blade suction surface in detail. Besides a passage vortex within the passage, there exist a shed corner vortex pair and a secondary vortex pair in the corner. The characteristics of two vortex pairs were different from that of the passage vortex.

The mechanism causing the shed corner vortex pair and secondary vortex pair and the effect of these vortices on the cascade losses are discussed.

Commentary by Dr. Valentin Fuster
1988;():V001T01A059. doi:10.1115/88-GT-151.
FREE TO VIEW

Experiments were conducted with glue-on hot-film gages in a large-scale axial turbine cascade to identify transition and/or separation on the suction surface of the blade. Standard strain-gage type temperature sensors were adapted and used as the gages and Transition and separation were identified by examining the mean and RMS voltage output. To assist with interpreting the output of the gages, surface oil-flow visualizations were used. Results of this study showed that transition and separation could be easily identified with the hot-film gages. Depending upon the Reynolds number and free stream turbulence level, the suction surface boundary layer was found to undergo bubble-induced transition, natural transition, or a combination of both; i.e, a transition which started naturally but ended abruptly with a bubble.

Commentary by Dr. Valentin Fuster
1988;():V001T01A060. doi:10.1115/88-GT-152.
FREE TO VIEW

A large-scale, low-speed, axial turbine cascade designed for high-loading and high-turning was tested over a range of Reynolds number, turbulence level, and incidence angle. End wall suction was applied to provide two-dimensional flow over a large spanwise region of the airfoil. In all, thirty-six test conditions were examined.

Overall cascade performance including mass-averaged loss coefficients at each test flow condition was determined from detailed five-hole pressure probe traverses in an exit plane of the cascade. In addition, using glue-on hot-film gages and surface oil-flow visualizations, transition and/or separation was identified over the suction surface of the airfoil. The measured transition start and end points were compared against predictions using existing transition models. Also, the measured losses were compared against predicted losses from boundary layer calculations based on finite difference analysis.

Commentary by Dr. Valentin Fuster
1988;():V001T01A061. doi:10.1115/88-GT-153.
FREE TO VIEW

Experiments were carried out in a model vaneless diffuser rig to investigate the rotating stall phenomenon and its relation to diffuser geometry. The experimental rig consisted of an actual impeller which was used to deliver the flow to the vaneless diffuser. Mass flow rate through the system could be adjusted by varying the rotational speed of the impeller at a fixed inlet opening or by changing the inlet opening at a fixed impeller speed. The flow exited to room condition. As such, the rig was designed to investigate the fluid mechanics of vaneless diffuser rotating stall only. Attention was focused on the effects of diffuser width and radius on rotating stall. Three diffuser widths and three outlet radii were examined. The width-to-inlet radius ratio varied between 0.09 and 0.142 while the outlet-to-inlet radius ratio varied between 1.5 and 2.

Results showed that the critical mass flow rate for the onset of rotating stall decreases with decreasing diffuser width. The critical mass flow rate is affected also by the diffuser radius ratio; larger radius ratios resulted in smaller critical mass flow rates. The ratio of the speed of rotation of the stall cell to impeller speed is found to decrease with increasing number of stall cells. This relative speed also decreases with increasing diffuser radius ratio, but it is largely independent of the diffuser width.

Commentary by Dr. Valentin Fuster
1988;():V001T01A063. doi:10.1115/88-GT-187.
FREE TO VIEW

Effects attributable to shock wave movement on cascade flutter were examined for both turbine and compressor blade rows, using a controlled-oscillating annular cascade test facility and a method for accurately measuring time-variant pressures on blade surfaces. Nature of the effects and blade surface extent influenced by the shock movement were clarified in a wide range of Mach number, reduced frequency and interblade phase angle. Remarkable unsteady aerodynamic force was generated by the shock movement and it significantly affected the occurrence of compressor cascade flutter as well as turbine one. For turbine cascade the interblade phase angle remarkably controlled the effect of the force, while for compressor one the reduced frequency dominated it. The chordwise extent on blade surface influenced by the shock movement was suggested to be about 6% chord length.

Commentary by Dr. Valentin Fuster
1988;():V001T01A065. doi:10.1115/88-GT-189.
FREE TO VIEW

Two methods for acquiring transient pressure measurements from a high speed multi-stage compressor are presented. Data were obtained from upstream, inter-stage and downstream measurement locations on the compressor during unsteady operation. The two methods of data acquisition were real time digital conversion of close coupled pressures and frequency modulated (FM) analog recording of high response measurements. The close coupled measurements provide for a nominal frequency response of 70 Hz while the high response measurements provided 200 Hz nominal response. A description of both acquisition systems is provided with discussion of the limitations involved in both methods. Examples and comparisons of data obtained by both methods are presented.

Commentary by Dr. Valentin Fuster
1988;():V001T01A066. doi:10.1115/88-GT-190.
FREE TO VIEW

The axial clearance between the tip of the blades of an unshrouded impeller and its stationary shroud has been varied to study its effect on overall compressor performance. The compressor under investigation consisted of an inlet nozzle, a 3D open impeller with full inducer, a parallel-wall vaneless diffuser and a collector. High-accuracy overall performance data were obtained for this compressor.

The experiments were carried out in a closed-loop centrifugal compressor test rig with the impeller running at a rotational Mach number u2/a0 = 1.39. The impeller tip diameter was 0.516 m, its tip width 0.021 m and the impeller blade exit angle was 30 degrees from radial.

Assuming a linear relationship, the experimental data indicates a pressure ratio decrease of 0.77 percent, an efficiency loss of 0.31 points, an input head reduction of about 0.25 percent and an output head reduction of about 0.65 percent for each percent increase in clearance ratio. However, the data seems to indicate a non-linear effect showing stronger performance sensitivity at smaller clearances.

The test data are compared against a clearance loss model. Improved performance prediction is obtained by including the effect of clearance on impeller work input.

Commentary by Dr. Valentin Fuster
1988;():V001T01A067. doi:10.1115/88-GT-197.
FREE TO VIEW

The paper presents the flow measurements within the leg portion of the horseshoe vortex. An uncambered blade of constant thickness downstream with a half-circular nose of an idealized turbine blade was installed on a flat plate. The six components of the Reynolds stresses were measured in addition to the three mean velocity components at two cross-sectional planes by a triple wire probe. The predominant vortical motion of the secondary flow occurs at the corner of the blade and the endwall. The effect of the penetrative motion of the free-stream toward the corner region induced by the vortex on the Reynolds stress is found in u2 profiles near the blade, but not in v2 profiles. The diffusion of the Reynolds stresses is observed in the crossflow direction.

Topics: Wire , Vortices , Blades
Commentary by Dr. Valentin Fuster
1988;():V001T01A075. doi:10.1115/88-GT-216.
FREE TO VIEW

The effects of tip clearance changes on efficiency in axial compressors are typically established experimentally. The ratio of change of efficiency with change of clearance gap varies significantly for different compressors in the published data. An analysis of this sensitivity range in terms of the blade and stage design parameters was initiated. The analysis revealed that the sensitivity range largely resulted from a derivation at constant flow of the efficiency decrement. It was also found that a generalized loss method of generating the sensitivities produced a much improved correlation of the change in efficiency with change in clearance over a variety of machines, configurations and speeds.

Commentary by Dr. Valentin Fuster
1988;():V001T01A076. doi:10.1115/88-GT-217.
FREE TO VIEW

The low speed flow through the shrouded de-Havilland Ghost centrifugal impeller is computed using an incompressible elliptic calculation procedure. The three dimensional viscous flow equations are solved using the SIMPLE algorithm in an arbitrary generalised coordinate system. A non-staggered grid arrangement is implemented in which pressure oscillations are eliminated using an amended pressure correction scheme. Flow computations are performed at ‘nominal’ low speed design and above design flow rates, and (on the coarse grids used in the calculations) good agreement is obtained with the experimentally observed jet-wake structure of the flow.

Commentary by Dr. Valentin Fuster
1988;():V001T01A077. doi:10.1115/88-GT-218.
FREE TO VIEW

Recently, a singular integral method was developed for solving two-dimensional compressible potential flows. The original formulation limited the applicability to single bodies in a uniform freestream. Currently, modifications are being made so that the flow field around multiple bodies can also be calculated. Because of the unique manner in which the integral equations are formulated, they are especially well-suited for analysis and inverse design calculations of a row of similar airfoils. A brief description of the formulation is given in this paper and some preliminary results are included.

Topics: Airfoils
Commentary by Dr. Valentin Fuster
1988;():V001T01A078. doi:10.1115/88-GT-219.
FREE TO VIEW

For many years, there has been a significant effort, both experimental and theoretical, to better understand the rotating stall phenomenon. For very practical reasons, most of the experimental investigations have focussed on laboratory compressors with very low stage pressure ratios. The aim of the present study was to extend the range of available data to industrial-size compressors operating in a typical real life environment. This paper reports results of detailed flow measurements made on the first four stages of a 10 stage compressor operating as part of a turbojet engine mounted on a test stand. Hot sensor anemometer measurements made at a number of axial and tangential locations showed clear evidence of rotating stall in the front stages during part-speed operation of the engine. Stall cell configuration and rotative speed, and details of flow speed and angle at hub, mid and tip radii are presented in the paper. On the basis of the measurements it is concluded that (1) although rotating stall has its origins in a flow instability, it is a highly reproducible phenomenon, (2) reverse flow can occur within the cells, as has been reported by several other observers and (3) the cells retain an axial (as opposed to helical) configuration on passing from stage to stage through the compressor.

Commentary by Dr. Valentin Fuster
1988;():V001T01A079. doi:10.1115/88-GT-220.
FREE TO VIEW

Problems of compressor instability have been of concern not only to aircraft engine designers but also for example to operators of industrial power plants. Two types of flow instabilities are known and in this paper only one of them, the surge phenomenon, will be discussed in detail. For the occurence of these heavy pressure fluctuations the volume of the neighbouring plenum chamber is the important quantity.

Here the basic compression system consisting of a compressor, a duct and a plenum will be analysed in order to show the complete unsteady pressure field. The theoretical model is based on the application of mass, momentum and energy conservation equations of one-dimensional flow. The set of differential equations of hyperbolic type is solved numerically by a ‘predictor-corrector’ scheme. The method itself is an explicit procedure first given by MacCormack /9/. At the interfaces between two components the coupling conditions for pressure and flow velocity are taken from the method of characteristics. The behaviour of the compressor is determined by its performance map, which describes the relationship between pressure and mass flux for all conditions from stable performance to inverse mass flow. For the dynamic change of the pressure development a special compressor characteristic taking into account the unsteady effects is used. This model function is chosen according to a time lagging element often employed in control theory.

To demonstrate the capability of the described method variations of different parameters such as mass flow rate, volume behind the compressor, ect. will be presented. The pressure-time history will be compared with experimental data published in the literature. Also the calculated surge frequencies, will be discussed. Furthermore it will be examined how a working point of the compressor moves from a stable part of the characteristic to another one, stable or unstable. Finally the general time dependent surge performance is considered in comparison to the quasi-steady characteristic.

Topics: Compressors , Surges
Commentary by Dr. Valentin Fuster
1988;():V001T01A080. doi:10.1115/88-GT-224.
FREE TO VIEW

Two-phase flow of wet steam is nowadays a subject of intensive research for different reasons. Prediction of erosion damage to turbine blading is one of important elements of this research. In the paper, a method of predicting this damage as a function of time is presented briefly. The emphasis is however put on the statistics of the accuracy of damage prediction as compared to field measurements. Also comments on erosion induced turbine efficiency deterioration are presented. The paper relates to the authors experience presented in references [1] and [2].

Commentary by Dr. Valentin Fuster
1988;():V001T01A081. doi:10.1115/88-GT-225.
FREE TO VIEW

The wall heat transfer resulting from small diameter holes drilled at 90° through gas turbine combustion chamber and turbine blade walls is considered. Available information is briefly reviewed and shown to generally omit the hole approach surface heat transfer and to relate only to the internal hole heat transfer. Experimental techniques are described for the determination of the overall heat transfer in a metal plate with a large number of coolant holes drilled at 90°. The results are compared with conventional short-tube internal heat transfer measurements and shown to involve much higher heat transfer rates and this mainly resulted from the additional hole approach flow heat transfer.

Commentary by Dr. Valentin Fuster
1988;():V001T01A082. doi:10.1115/88-GT-226.
FREE TO VIEW

Detailed measurements of mean velocity, turbulence intensity and Reynolds stresses have been performed in the passage of a cascade of turbine rotor blades. By using the experimental values of the mean velocity, the turbulence quantities are computed with three different turbulence closure models. The results are analysed and compared with the experimental data. The capability of the closure models to describe the turbulence development associated with secondary flows in a turbine cascade is discussed.

Commentary by Dr. Valentin Fuster
1988;():V001T01A086. doi:10.1115/88-GT-237.
FREE TO VIEW

In order to facilitate the aerodynamic design for the computer-aided design (CAD) of centrifugal compressor, a unified direct problem method for the flow calculations along S1 and S2 stream surfaces has been presented in the present paper. A single stream function equation expressed by nonorthogonal curvilinear coordinates and the unified matrix direct solution for the governing equation have been used. This method greatly simplifed the quasi-three dimensional and full dimensional computing program, meanwhile it improved the computing accuracy and the convergence rate. Numerical examples have illustrated the advantages of the new technique for CAD of centrifugal compressors.

Commentary by Dr. Valentin Fuster
1988;():V001T01A088. doi:10.1115/88-GT-248.
FREE TO VIEW

The aerodynamic blade cascade design of a two stage axial supersonic turbine is investigated by using an inviscid flow computation method. The flow inside such kind of turbine is characterized by high inlet Mach numbers and large deflection angles as well as supersonic flow matching between stator and rotor. Most of the computed flow configurations give strong shock waves in the blade-to-blade channels, which let anticipate boundary layer separation phenomena. The inverse mode calculation is applied in order to avoid adverse pressure gradients on the walls. The semi-inverse method allows us to get the geometry of a blade profile corresponding to a given pressure distribution on the suction side and the pressure side, the solidity being fixed. A new design of the turbine blade cascade is then considered in order to achieve the desired velocity diagram.

Commentary by Dr. Valentin Fuster
1988;():V001T01A091. doi:10.1115/88-GT-255.
FREE TO VIEW

In order to analyze axial-flow compressor flutter, methods are required that compute the unsteady flow through vibrating cascades. A 3D fully nonlinear method has been developed by numerically integrating the 3D unsteady Euler equations, in the time-domain. The equations are discretized in a moving grid, which conforms with the vibrating blades and are integrated using the explicit MacCormack scheme, in finite-difference formulation. The method assumes a traveling-wave assembly mode of vibration. In this manner, the flow is computed in a single channel by applying the corresponding chorochronical periodicity condition at the permeable pitchwise limits. The blade vibratory mode is an input to the method obtained by a standard finite element method structural analysis code. A number of results are presented, for a transonic fan rotor, illustrating the possibilities of the method, both in started and unstarted supersonic flow conditions.

Commentary by Dr. Valentin Fuster
1988;():V001T01A092. doi:10.1115/88-GT-256.
FREE TO VIEW

Flow visualisation and microscopic static pressure measurements were done in the tip clearance region of an annular turbine cascade with a rotating outer casing to simulate the relative motion at the tip of an axial rotor. The effect of relative motion did not have a significant effect on the blade gap pressure distributions. As in previous studies the narrow deep pressure depression on a sharp pressure edge was seen. It was confirmed that the width of the gap separation bubble depends on clearance and a correlation with flow visualisation showed that at the reattachment line there is the expected slight pressure peak. The separation bubble, which is thought to contribute a major part of the leakage loss, was shown to disappear when the pressure surface tip is give a radius of 2.5 gap widths.

Commentary by Dr. Valentin Fuster
1988;():V001T01A093. doi:10.1115/88-GT-260.
FREE TO VIEW

A secondary flow calculation method is presented in this work, which makes use of the meridional vorticity transport equation. Circumferentially mean flow quantities are calculated using an inverse procedure.

The method makes use of the mean kinetic energy integral equation and calculates simultaneously hub and tip secondary flow development. Emphasis is placed upon the use of a coherent two-zone model and particular care is taken in order to describe adequately the flow inside an unbounded (external), semi-bounded (annulus) and fully-bounded (bladed) space. The velocity field, the losses and the defect forces receive particular attention.

Comparison between theoretical and experimental results is presented.

Commentary by Dr. Valentin Fuster
1988;():V001T01A094. doi:10.1115/88-GT-262.
FREE TO VIEW

In this paper, a full Inverse problem method — a prescribed wall-velocity method (PWVM) — and a hybrid problem method are suggested in solving S2 stream surface. In these methods two walls or one wall coordinates of the meridional through flow section of turbomachine can be obtained when the velocities along the hub and casing walls or one of them are prescribed. The new form of the governing equations in terms of grid coordinates are obtained. These equations are different with those used for the S1 stream surfaces. The suitable wall-velocities can be selected and prescribed by designer and the coordinates of hub and casing walls will be obtained by solving these new equations on S2 stream surfaces. The boundary conditions and the solution of these equations will be presented too.

The programs have been written and several examples are calculated. The results seems to be satisfactory. A comparison between the solutions of this method and the ordinary method, which is a method given the coordinates of hub and casing walls, is also presented in this paper.

Topics: Turbomachinery
Commentary by Dr. Valentin Fuster
1988;():V001T01A095. doi:10.1115/88-GT-265.
FREE TO VIEW

A new type of compressor cascade, called the vortex flow cascade (shortly VFC), has been developed in the paper. The VFC is made up of the normal compressor cascade (shortly NCC) with NACA-65-0010 profile and vortex generator. Experiments are conducted for researching the effects of a large scale streamwise control vortex on the flow structure inside cascade passage. The results are encouraging. Based on the present investigation the vortical flow pattern and loss mechanism of VFC have been discussed.

Commentary by Dr. Valentin Fuster
1988;():V001T01A096. doi:10.1115/88-GT-269.
FREE TO VIEW

A straight compressor cascade of aspect ratio 2 was tested in a low speed tunnel within Reynolds number Re1 = 45 000 – 150 000 and inlet flow angle α1 = 35° – 48°. The profile of the blade was NACA 65-12-10. The purpose of the paper was to obtain data on three–dimensional flow in a straight cascade at low Reynolds numbers. Some experimental results on secondary flow have been made into simple correlation relations.

Commentary by Dr. Valentin Fuster
1988;():V001T01A099. doi:10.1115/88-GT-276.
FREE TO VIEW

Blade-cooling air for a high-pressure turbine is often supplied from pre-swirl nozzles attached to a stationary casing. By swirling the cooling air in the direction of rotation of the turbine disc, the temperature of the air relative to the blades can be reduced. The question addressed in this paper is: knowing the temperatures of the pre-swirl and disc-cooling flows, what is the temperature of the blade-cooling air?

A simple theoretical model, based on the Reynolds analogy applied to an adiabatic rotor-stator system, is used to calculate the pre-swirl effectiveness (that is, the reduction in the temperature of the blade-cooling air as a result of pre-swirling the flow). A mixing model is used to account for the ‘contamination’ of the blade-coolant with disc-cooling air, and an approximate solution is used to estimate the effect of frictional heating on the disc-cooling air.

Experiments were conducted in a rotor-stator rig which had pre-swirl nozzles in the stator and blade-cooling passages in the rotating disc. A radial outflow or inflow of disc-cooling air was also supplied, and measurements of the temperature difference between the pre-swirl and blade-cooling air were made for a range of flow rates and for rotational Reynolds numbers up to Reθ = 1.8 × 106. Considering the experimental errors in measuring the small temperature differences, good agreement between theory and experiment was achieved.

Topics: Cooling , Rotors , Blades , Stators
Commentary by Dr. Valentin Fuster
1988;():V001T01A104. doi:10.1115/88-GT-294.
FREE TO VIEW

Turbines and compressors operating in polluted atmosphere with solid particles are subjected to performance deterioration. This paper presents an investigation carried out on two-stage gas turbine with blunt leading edge blades and on a single-stage axial flow compressor to study the effects of particulates and erosion on performance deterioration.

Commentary by Dr. Valentin Fuster
1988;():V001T01A106. doi:10.1115/88-GT-299.
FREE TO VIEW

This paper describes numerically calculated results obtained for a viscous driven enclosed cavity flow and comparison with the experiments of Daly and Nece (Ref. 4). The sensitive prediction of moment coefficient is chosen for comparison purposes. Some discussion of the impact of k-ε turbulence modeling is also included. A second configuration demonstrating the degree of geometric complexity which can be handled is also presented.

Commentary by Dr. Valentin Fuster
1988;():V001T01A109. doi:10.1115/88-GT-323.
FREE TO VIEW

An experimental investigation of rotating stall and surge was carried out on a four stage axial flow compressor. Results of flow and blade temperature measurements in the compressor are presented. Internal temperature levels during rotating stall and surge are considerably higher than those obtained during unstalled compressor operation. In the pure rotating stall regime, the temperature is almost identical in all compressor stages and depends only on rotor speed and mass flow rate. During surge, the highest temperature is found at the tip diameter prior to the first stage rotor. The absolute level depends on rotor speed, mass flow rate (i.e. throttle position) and on the number of compressor stages. A model of the temperature changes in the multistage compressor during the surge cycle has been derived from the experiments.

Commentary by Dr. Valentin Fuster

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In