Abstract Particle ingestion into turbine engines is a widespread problem that can cause significant degradation in engine service life. One primary damage mechanism is deposition of particulate matter in internal cooling passages. Musgrove et al. proposed a compact particle separator that could be installed between the combustor bypass exit and turbine vane cooling passage inlet. The design had small pressure losses but provided limited particle separation, and its performance has proved difficult to replicate in subsequent experiments. Borup et al. recently developed a Magnetic Resonance Imaging (MRI) based technique for making full-field, 3D measurements of the mean particle concentration distribution in complex flows. A particle separator based on the Musgrove et al. design was fabricated out of plastic using 3D printing. The primary difference from earlier designs was the addition of a drain from the collector, through which 3% of the total flow was extracted. The separator efficiency was measured at two Reynolds numbers, using water as the working fluid and 33-micron titanium microspheres to represent dust particles. Particle Stokes number was shown to play the dominant role in determining efficiency across studies. MRI was used to obtain the 3D particle volume fraction and 3-component velocity fields. The velocity data showed that flow was poorly distributed between the separator louvers, while the collector flow followed the optimal pattern for particle retention. The particle distribution data revealed that strong swirling flow in the collector centrifuged particles towards the outer wall of the collector and into a partitioned region of quiescent flow, where they proceeded to exit the collector via the drain. Future designs could be improved by re-arranging the louvers to produce a more uniform flow distribution, while maintaining the effective collector design.

Abstract The changing energy landscape leads to a rising demand of more flexible power generation. A system for steam turbines warm-keeping provides the ability to shutdown conventional power plants during periods with a high share of renewable power. Simultaneously, these power plants are ready for grid stabilization on demand without an excessive consumption of lifetime during the start-up. One technical solution to keep a steam turbine warm is the use of hot air which is passed through the turbine. In addition, the air supply prevents corrosion during standstill and also enables the pre-warming after maintenance or long outages. This paper investigates the warm-keeping process of an intermediate pressure steam turbine (double shell configuration) through the use of dynamic numerical Finite-Elements (FE) simulations. As a representative test-case, warm-keeping calculations during a weekend shutdown (60h) are conducted to investigate the temperatures, their distribution and gradients within the rotor and the casing. For this purpose an improved numerical calculation model is developed. This detailed 3D FE model (including blades and vanes) uses heat transfer correlations conceived for warm-keeping with low air mass flows in gear mode operation. These analytical correlations take heat radiation, convection and contact heat transfer at the blade roots into account. The thermal boundary conditions at the outer walls of rotor and casing are determined by use of experimental natural cool-down data. The calculation model is finally compared and verified with this data set. The results offer valuable information about the thermal condition of the steam turbine for a subsequent start-up procedure. The warm-keeping operation with air is able to preserve hot start conditions for any time period. Most of the heat is transferred close to the steam inlet of the turbine, which is caused by similar flow directions of air and steam. Thus, temperatures in the last stages and in the casing stay well below material limits. This allows higher temperatures at the first blade groove of the turbine, which are highly loaded during a turbine startup and thus crucial to the lifetime.

In this paper, we propose a two-step methodology to evaluate the convective heat flux along the rotor casing using an engine-scalable approach based on Discrete Green’s Functions. The first step consists in the use of an inverse heat transfer technique to retrieve the heat flux distribution on the shroud inner wall by measuring the temperature of the outside wall; the second step is the calculation of the convective heat flux at engine conditions, using the experimental heat flux and the Green Functions engine-scalable technique. Inverse methodologies allow the determination of boundary conditions, in this case the inner casing surface heat flux, based on measurements from outside of the system, which prevents aerothermal distortion caused by routing the instrumentation into the test article. The heat flux, retrieved from the inverse heat transfer methodology, is related to the rotor tip gap. Therefore, for a given geometry and tip gap, the pressure and temperature can also be retrieved. In this work, the Digital Filter Method is applied in order to take advantage of the response of the temperature to heat flux pulses. The Discrete Green’s Function approach employs a matrix to relate an arbitrary temperature distribution to a series of pulses of heat flux. In the present procedure, the terms of the Green’s Function matrix are evaluated with the output of the inverse heat transfer method. Given that key dimensionless numbers are conserved, the Green’s Functions matrix can be extrapolated to engine-like conditions. A validation of the methodology is performed by imposing different arbitrary heat flux distributions, to finally demonstrate the scalability of the Green’s Function Method to engine conditions. A detailed uncertainty analysis of the two-step routine is included based on the value of the pulse of heat flux, the temperature measurement uncertainty, the thermal properties of the material and the physical properties of the rotor casing.

FloMASTER is a 1-D thermo-fluids system simulation tool and its component models depend on the characterisation data of the component performance. Such performance data is mainly based on data banks established from extensive tests exemplified by the books like “Internal Flow” by Miller [1] and “Handbook of Hydraulic Resistance” by Idelchik [2]. One of the key components of the gas turbine secondary air system is the rotating annulus. However, reliable data and correlations for performance characteristics like pressure loss coefficient, torque coefficient, windage and heat transfer for this component are rare and non-existent in the open literature for the case of both walls rotating simultaneously, which is becoming more common in today’s multi-spool military aero engines. To overcome this challenge of lack of reliable performance data and correlations, in this paper the Mentor Graphics 3D CFD tool “FloEFD” is used to model both inner wall rotating and outer wall rotating annulus flow, and to verify the 3D CFD results of performance data in terms of pressure loss coefficient and torque coefficient versus some published test data in the open literature. It is shown that the CFD gives results on pressure loss and torque coefficients that are in good agreement with test data based correlations used in FloMASTER. This demonstrates that 3D CFD can be used as a powerful tool for verifying the existing 1D model, extending the 1D model performance data range and generating new performance data for developing new components where such data is not available from open literature. A future project is to extend this approach to provide performance data for rotating annuli with both walls rotating. Such data will form the basis for developing a new component model for a rotating annulus with both walls rotating.

In the previous studies, the proposed method for gooseneck geometric modeling employed two polynomials to construct the inner-wall and area distribution curves. The inflection point location served as the variable to control the inner-wall polynomial curve, and the peak point location and peak value to control the area distribution polynomial curve. In the effort to be quickly located, the control variables were provided with more geometric meaning. 3D numerical simulations indicated that there existed a total pressure recovery island for given solution area of the three control variables. Consequently, the relationships between the geometric parameters and the total pressure loss were set up. This paper focused on the 2D simplifications to quickly address the control variables for the total pressure island. The studies were conducted in three aspects. First, the simplified model took the constructional blocking effects of struts into account. The baseline of the 2D simplified modeling was set at 30% spanwise near the hub through comparisons of different settings. Therefore, 70% blocking area compensated to outer-wall and 30% to inner-wall along the normal direction of the baseline. The 2D simulation results indicated that the static pressure distribution was consistent with the 3D results, but waves exited at the end walls of both leading and trailing edges due to the geometric changes. Second, the simplification considered the blocking effects of the wake. The wake was converted to boundary layer thickness, and moreover, compensation to the end wall was similar with the constructional blocking of struts. The simulation results revealed that wake blocking had very small impacts to the simplification, although the peak values of static pressure slightly increased at the end wall. Third, smoothing treatments were done for both inner-wall and outer-wall after the above compensating transformations. The results showed that smoothing treatments were very necessary and improved the waves located at end wall on the static pressure distribution which was nearly the same with 3D results. After all the simplifying treatments above, the final 2D results had almost the same total pressure loss distributions with the 3D results, and could save at least 40% calculation time as a quick assessment used to search the reasonable geometric solution areas of inflection point location and peak point location for minimum total pressure loss of the gooseneck.

The emissions characteristics of a model gas turbine combustor characterized by the enhancement of the reactions of the secondary mixtures of lean to ultra-lean compositions by the burned gas from the primary stage were investigated at atmospheric pressure. A cylindrical quartz tube with the bottom end closed was used as the combustion chamber and the mixture injector was extending co-axially into the combustion chamber to the bottom wall. The stagnation point reverse flow combustion was used for the primary stage where the primary mixture was injected toward the end wall. The secondary mixture was injected radially from the multiple holes in the outer wall of the injector into the burned gas from the primary stage. The effects of the positions for injection of the primary and secondary mixtures as well as mixture temperatures on NOx, CO and HC emissions were investigated over a range of the primary and secondary mixture equivalence ratios at atmospheric pressure.

Demands on the high by-pass ratio of modern civil engine, an increasing radial offset between the booster and the high pressure compressor highlights the importance of S-shaped gooseneck. A desirable balance between the length and the radial offset by optimizing inner-wall and outer-wall for gooseneck will save the length of low pressure rotor and improve the situation to deliver the flow through it. This paper presented a geometric modeling method which adopted polynomial curves to construct inner-wall and section area distribution for goosenecks. With the given boundary dimensions of gooseneck, the location of inflection point served as a variable to control the inner-wall curve, and the peak value and its location served as the other two variables to control the area distribution curve. A series of 3-D models were built from the baseline gooseneck model in order to analyze the relationships between the geometric parameters and the aerodynamic performance. The 3-D numerical simulation results indicated that the peak value of area distribution has the most important effects on the total pressure loss of gooseneck. An appropriate peak value of area distribution could be obtained by changing the normalized Mach number at peak point to weaken the flow separation near the outer-wall and after the struts. Inner-walls with large curvature near the entrance of gooseneck, and area distributions with an appropriate peak location, would improve the streamline of outer-wall and restrain the flow separation. Subsequently, the relationships between the geometric control variables and the total pressure loss were established. Moreover, taking the constructional blocking effects of struts into account, a series of simplified 2-D flow paths were modeled and the relationships were reestablished. The 2-D results indicated that the simplified model could also reveal the flow field near the inner-wall address the correct peak value. But the optimum peak location and the inner-wall inflection point were not consistent with the 3-D results due to the geometric changes on the outer-wall. The equivalent method will be further improved in the future research.

The development and limitations of a numerical modelling framework applied to an aero-engine air/oil separator are presented here. Oil enters the device in the form of dispersed droplets and primary separation occurs by centrifuging larger droplets towards the outer walls, whereas secondary separation occurs by partially coalescing and centrifuging smaller droplets within a porous material, namely an open-cell metal foam. The work described here is part of a study led jointly by the University of Nottingham (UNott) and the Karlsruhe Institute of Technology (KIT) in the Engine Breakthrough Components and Subsystems (E-BREAK) project. The main objectives for UNott have been to define a CFD methodology able to provide an accurate representation of the air flow behaviour and a qualitative assessment of the oil capture within the air/oil separator. The feasibility of using the current state-of-the-art modelling framework is assessed. Experimental measurements of the overall pressure drop and oil capture performed at KIT are used to validate the simulations. The methodology presented here overcomes some limitations and simplifications present in previous studies. A novel macroscopic model for the secondary oil separation phenomena within metal foams is presented. Experiments and simulations were conducted for three different separator configurations, one without a metal foam, and two with metal foams of different pore sizes. For each configuration, a variation of air flow, shaft speed and droplet size was conducted. The focus was on the separation of droplets with a diameter smaller than 10 μm. Single-phase air flow simulation results showed that overall pressure drop increases with both increased shaft speed and air flow, largely in agreement with the experiments. Oil capture results proved to be more difficult to be captured by the numerical model. One of the limitations of the modelling set-up employed here is not capable of capturing droplet re-entrainment due to accumulation of oil inside the metal foam, which is believed to play a significant role in the separation phenomena.

Ring vanes were recognized as being beneficial for increasing pressure recovery and reducing stall in wide-angled diffusers. Augmentation devices can also reduce swirl-induced flow distortion. A three-point ring vane was inserted into a short annular diffuser with a length-to-inlet-height ratio of 14.5 and area ratio of 4. Numerical optimization studies were completed using the realizable k-ε turbulence model at zero and 12 deg inlet swirl that established a preferential location of the ring vane leading edge. Parametric computational fluid dynamics studies evaluated the influence of the dominant variables — overall length and angle — at 0 to 12 deg inlet swirl. Performance was dependent on the ability of the ring vane to reduce outer wall stall without swirl or the central toroidal recirculation zone with swirl and the ring vane wake. Shorter vanes angled towards the centre reduced back pressure with no swirl whereas longer outward-angled vanes were preferred with 12 deg swirl.

A new aerodynamic design strategy of the S-shaped transition duct between two compressor components was studied. Based on the controlled wall pressure gradient distribution and the wall velocity distribution, a semi-inverse problem of the transition duct was proposed, the corresponding inverse and direct approach codes were developed. To verify the feasibility of this method, two axial-centrifugal compressor transition ducts were designed. The results show that the static pressure distribution on the inner wall and the duct geometry both can be controlled freely by adjusting the inverse design parameters. The designed inner wall pressure distribution can be realized through a numerical matching procedure of the outer wall geometry based on the direct problem. The new design method is practicable that, without searching the optimal solution of the static pressure distribution of the inner wall, the total pressure coefficient can be at least 0.92.

A highly loaded transonic fan with a splittered rotor and maximum pressure ratio of 2 has been tested at the Naval Post-graduate School. Temperatures on the casing outer wall have been measured using thermocouples in order to assess the heat transfer. An axisymmetric heat transfer analysis has shown that the heat transfer downstream of the rotor through the casing leads to an overstatement of the experimental efficiency by 0.36% and 0.29% for 100% and 90% speed respectively. Just above the rotor is an abradable material that can be considered adiabatic. Thus heat escapes only aft of the rotor and upstream of the total pressure and total temperature probes. For rotors without an adiabatic casing, the effect would be considerably higher. This has a large effect on the exit total temperature profiles which are often not matched well with CFD using an adiabatic boundary condition. In addition, heat transfer coefficients have been determined that can be used to assess other compressor applications.

The trend for higher overall pressure ratios means that turbine entry temperatures are continually increasing. Furthermore, the development of lean, low-emission combustion systems reduces the availability of cooling air and is accompanied by new problems at the combustor/turbine interface. For example, the exit temperature traverse differs from that found in traditional rich-burn combustors with increased swirl and a much flatter profile. Effectively cooling the turbine components is becoming increasingly difficult. One solution is to employ cooled cooling air (CCA) where some of the compressor efflux is diverted for additional cooling in a heat exchanger located in the by-pass duct. An example CCA system is presented which includes an off-take within the dump cavity and the addition of radial struts within the pre-diffuser through which the cooled air is returned to the engine core. This paper addresses the impact this CCA system has on the combustion system external aerodynamics. This included the development of a fully annular, isothermal test facility which incorporated a bespoke 1.5 stage axial compressor, engine relevant outlet guide vanes, pre-diffuser and combustor geometry. A datum aerodynamic performance was established for a non-CCA configuration with a clean, un-strutted pre-diffuser. Results for this baseline CCA system demonstrated that inclusion of a bleed in the dump cavity had limited effect on the overall flow field. However, the inclusion of struts within the pre-diffuser caused a reduction in area ratio and a notable increase in system loss. Consequently an alternative pre-diffuser was designed (using CFD) with the aim of increasing the area ratio back to that of the un-strutted datum. A so-called hybrid diffuser was designed in which the CCA bleed was moved to the pre-diffuser outer wall. The bleed was then used to re-energize the boundary layer, preventing flow separation, enabling the area ratio to be increased close to the datum value. The mechanisms of the hybrid diffuser are complex; the geometry of the off-take and its location with respect to the OGV and strut leading edge were seen to be critical. Experimental evaluation of the final design demonstrated the effective operation the hybrid diffuser with the result that the system loss returned to a level close to that of the datum. Only small differences were seen in the overall flow field.

The high efficiencies of small radial turbines have led to their application in space power systems and numerous APU and shaft power engines. Experimental and analytical work associated with these systems has included examination of blade shroud clearance, blade loading, and exit diffuser design. Results indicate high efficiency over a wide range of specific speed and also insensitivity to clearance and blade loading in the radial part of the rotor. The exit diffuser investigation indicated that a conventional conical outer wall may not provide the velocity variation consistent with minimum overall diffuser loss. A list of recently published NASA radial turbine reports is included.

Experimental investigations brought to light the possibilities of broadening the operating range of axial flow compressors by means of outer wall casing treatments such as grooves, honeycombs or cavities. The industrial use of these techniques is however limited by uncertainties on the effectiveness of these treatments as, while they are very beneficial in some cases, they deteriorate the performance of the machine in some others. A theoretical investigation has been conducted on the flow field in cavities most commonly used, and laws for determining the groove depth have been established which account for the surge margin improvement obtained in some cases and make it possible to dismiss inefficient geometric configurations. A comparison between theory and experiment is presented, based on test results published in the literature and on an experimentation on two industrial compressors.

A systemmatic series of tests has been conducted on a family of annular diffusers where the outer casing is maintained at constant diameter. Such a diffuser is typical of turbine exits. Data, in the form of static pressure recovery coefficient is plotted against diffuser length for several different designs of centerbody closure. It has been shown that such diffusers can have short length centerbodies for which a set of design guides has been established.

The detailed heat/mass transfer distributions in and around the sharp 180 degree turn of a three-pass square channel were determined by using the naphthalene sublimation technique. The top, bottom, inner (divider) and outer walls of the test channel were naphthalene plates. For the case of rib-roughened tests, the ribs of square cross section were glued periodically in-line on the top and bottom walls of the naphthalene channel in a required distribution. The rib height-to-hydraulic diameter ratios (e/D) were 0.063 and 0.094, whereas the rib pitch-to-height ratios (P/e) were 10 and 20, respectively. The channel Reynolds numbers varied from 30,000 to 60,000. The results showed that, for both the smooth and the ribbed channels, the Sherwood numbers after the sharp 180 degree turn were higher than that before the sharp 180 degree turn; after the turn the Sherwood numbers of the inner wall were higher than that of the outer wall. The results also indicated that the Sherwood numbers on the top, outer and inner walls of the rib roughened channel were higher than that of the smooth channel.

Three groups of the annular turbine cascades with diameter-blade height ratios d/1 = 1.9, 3, and 5 respectively are tested. Each group includes five types of cascades equipped with different kinds of leaned blades. For each type of the cascades, the distributions of total pressure, static pressure and exit flow angle along the pitch and the blade height are surveyed in detail. The experimental results show that the smaller the diameter-blade height ratio of the cascade, the more effective application of positively leaned blade to reduce the secondary flow loss, when the inner and outer walls of the cascade are cylindrical. Under this condition the effect of the blade leaning on the energy loss distribution along the blade height is notable. There is an optimal lean angle corresponding to the minimum of the overall energy loss in each cascade with leaned blades. The function-relation between the optimal lean angle and the diameter-blade height ratio is monotonic.

An experimental investigation has been carried out to study the performance of an integrated design of compressor outlet guide vane and combustor pre-diffuser system. The trailing edge of each OGV was located within the outwardly canted diffuser by a distance equal to 27% of the diffuser axial length. In order to obtain representative inlet conditions a rotor, providing a fully sheared velocity profile and an air outlet angle of approximately 40°, was located upstream of the OGVs. Compared with the measured performance when the trailing edge of each OGV was located at diffuser inlet, a small increase in total pressure loss and a corresponding decrease in static pressure recovery was observed for the shortened system. This change in performance reflected a deterioration in the flow conditions along the outer wall, with transitory stalling of the flow being observed at diffuser exit. By leaning the blades in a circumferential direction through angles of 10° and 15° the outer wall flow conditions could be progressively improved, although at the largest angle tested stalling of the flow occurred at the hub of each OGV. However, at a lean angle of 10° the performance, in terms of loss and flow stability, could be virtually restored to the levels obtained when the trailing edge of each OGV was located radially at diffuser inlet.

A pilot hot gas particulate removal system based on positive porous ceramic filters has been tested on the Grimethorpe Pressurized Fluidized Bed Combustor facility. The filters are in the form of closed ended tubes, 1.5 m long: these are generally called ‘candles’. The dust accumulates on the outside of the candles, and is periodically removed by a pulse of air into the candle interior, which then flows outwards through the candle wall in the reverse direction to the normal flow of the combustion gas. The EPRI system contained a maximum of 130 candles, which is approximately equivalent to the requirement for 7 MW(e) capacity, depending on the filter operating parameters. The filter unit operated for a total of 860 hours under PFBC conditions, of which 790 hours were at defined process conditions, typically 850°C and 10 bar. The amount of gas flowing through each filter element was varied, and the time between cleaning pulses was also varied. The pressure drop through each filter element rose as the dust accumulated on the outer wall, and recovered after the cleaning pulse. However, the post-cleaning pressure drop does not recover to the original clean candle value, but increases with time. It is believed that a steady-state value is attained, but the exposure in the Grimethorpe test series was insufficient to establish this unequivocally. During the test, five candles failed. This appears to have been due to mechanical shock, as a result of candles lifting because of excessive pressure differentials across their support plate, and dropping back. The failures are not believed to be intrinsic to the technology. However, in addition a reduction in the strength of the candles with time of exposure was observed. This might also attain a steady state value, but this too could not be established on the basis of the tests reported in this paper. This is clearly a matter of importance, and further work will be required to determine the suitability of the clay-bonded silicon carbide medium used in these tests for this application. A number of deficiencies in the design of the unit emerged with the operating experience, and suggestions have been made for improvements. However, it is clear that further work on design optimization is required. The pulse cleaning air usage in the tests was greater than would be economically acceptable in a practical system. Further work needs to be done to optimize the cleaning cycle. Overall, the test was very successful, and when operating properly the filters removed essentially all of the dust in the gas exiting from the combustor. Apart from the issues with the candle strength and the pulse cleaning air usage, the other problems were not believed to be of major importance in the further development of the technology. This paper will summarize the test results, emphasizing the problems of candle durability and the pulse cleaning system.

A computational procedure based on the solution of fully elliptic Navier-Stokes equations on a body-fitted non-orthogonal grid was used to obtain flow fields in annular diffusers with a suction slot at the inner and outer walls. The turbulence effects were simulated by high Reynolds number form of the k-ε model. The calculation method was used to modify an industrial gas turbine (GE MS · 7001F) compressor/combustor annular diffuser to allow extraction of compressed airflow for coal gasification in simplified IGCC Systems. The air for gasification was extracted through a suction slot on the outer wall of the diffuser which was curved to improve the overall performance and to avoid flow separation; both of these insured by providing accelerated flow through the suction slot and nearly constant wall pressure downstream of the slot. Suction slot and outer wall geometries to result in the above conditions were determined by a trial and error procedure. The diffuser’s performance was further improved by extracting 6% of the compressed air through a slot at the inner wall, kept straight due to structural constraints. The resulting diffuser arrangement was relatively insensitive to the upstream disturbances.