Optimization of air jet pump design for acoustic application

Project Information

OPA

Grant agreement ID: 325977

Closed project

Start date 1 April 2013

End date 31 July 2015

Funded under

• Specific Programme "Cooperation": Joint Technology Initiatives

Total cost € 298 000,00

EU contribution € 212 562,50

Coordinated by

INSTITUTUL NATIONAL DE CERCETARE-DEZVOLTARE TURBOMOTOARE - COMOTI
Romania

Final Report Summary - OPA (Optimization of air jet pump design for acoustic application)

Executive Summary:
The goal of the project was to optimize the design of a subsonic jet pump, to reduce as much as possible the noise level. The optimization of the subsonic jet pump was done from the aerodynamic point of view and two type of numerical analyses were be used: a Reynolds Averaged Navier - Stokes (RAND) analysis as a “fast-screening” procedure to be able to identify trends in the flow behavior when performing a parameter study, and a time dependent analysis using the more time-consuming but more accurate and more reliable Large Eddy Simulation (LES), for only the most promising or thought-provoking configurations.
For the RANS analysis, a set of candidate geometries was defined, based on a baseline geometry provided by the SGO member. Each configuration was derived from the baseline geometry by modifying relevant geometrical parameters for optimization purposes. Next, the fast and robust RANS technique will be used to select a couple of these configurations. The RANS numerical simulations were carried out using the SST k-ω turbulence model and the ANSYS FLUENT commercial software. The numerical results of the simulations were analyzed, particularly from the standpoint of to the production of turbulence kinetic energy, the wall pressure distribution, the uniformity index, and the vorticity components, in order to obtain information on the location and the strength of the acoustic sources as well as on the mixing performance. Based on this analysis, three jet pump configurations showing the greatest potential for noise reduction and increasing mixing performance were selected for further experimental and numerical testing.
Next, the baseline and the selected optimized jet pump demonstrators were designed based on the previously defined geometry and manufactured. A modular design of the demonstrators was be carried out, such that to maximize the number of common components. Also, test rig adaptation parts required to carry out experimental measurements were designed and manufactured. An experimental program aimed for acoustic and aerodynamic measurements was defined. Two types of experiments were planned and carried out: acoustic measurements on the three selected configurations and on the baseline configuration, to quantify and assess the impact of modifications, and aerodynamic tests on the three selected configurations and on the baseline configuration for the determination of the instantaneous flow velocity field using Particle Image Velocimetry (PIV). The acoustic and the aerodynamic data were analyzed.
The directivity tests revealed that the best results were obtained for Configuration 4, which yielded a noise reduction, determined using the global sound level calculated from the levels directivity microphones of 2.5dB(A). The global acoustic intensity show that for Configurations 2 and 3 the area with high noise level shifted towards the ejector exit in comparison with the baseline (Configuration 1). For Configuration 4, the high acoustic intensity area is completely isolated at the injection area. During the intercorrelation measurements, a low vibration level at the exit of the primary flow was obtained for all configurations, At a distance of 4D the pump natural frequencies work for Configurations 1 – 3, while for Configuration 4 a new frequency is excited, because of the injector geometry. The correlation function revealed a substantial contribution of the high frequencies vibration in the sound signal.
The highest axial velocity component value on the centreline is measured for Configuration 1, while the lowest is measured for Configurations 2 and 3. The fastest momentum mixing between the primary stream and the entrained air is registered for Configuration 4 which is an advantage from the noise production standpoint. In the shear layer, large fluctuations of the mean axial velocity are observed, indicating a high level of turbulence in the region, with a maximum value noted for Configuration 3. The tapering of the central air stream is found minimal for Configuration 4. Vortical structures are observed in the mean flow in the transversal and spanwise profiles, due to the detachment of the boundary layer on the mixing segment duct. The vortex is the strongest of Configuration 4. In the transversal direction the central, high velocity region of the flow expands towards downstream, due to the spreading of the primary jet. The spreading rate is slightly larger for Configurations 2 and 3, with only marginal differences between them, and the smallest for Configuration 4. The momentum mixing rate is roughly similar, with a slightly larger value registering for Configuration 4, where the transversal and spanwise velocity profiles indicate a ring shaped jet that progresses radially towards the centreline as it moves downstream.
Finally, unsteady, compressible LES numerical simulations of the confined jet flows characterizing the baseline and the three optimized solutions were carried out, reproducing the experimental tests conditions and geometries. The Dynamic Mode Decomposition (DMD) technique was used for post-processing the LES data, to quantify the developed instabilities in the duct, to characterize the energetic content of the dominant modes, and to identify the flow structures at dominant frequencies correlating the flow and the acoustics. Acoustic propagation at the diffuser outlet, up to 20 exit diameters downstream was also calculated based on the LES data.

Project Context and Objectives:
A jet pump is, basically, a duct confined injector with the role of entraining and mixing a secondary jet by and into a primary jet. Air-jet pumps are used in industry to feed and transport a wide range of bulk solids because of their simplicity in structure, lack of moving parts, etc. The design of this type of equipment and associated conveying system still is a "black art" due to the shortage of theoretical analysis and experimental data reported in the literature regarding the design and application of these pumps. A typical jet pump consists of two counter-facing, straight-sided cones separated by a parallel throat section. Within this section the primary flow, in the form of a high velocity jet issuing from the nozzle, entrains the secondary flow, thereby generating the pumping behaviour. A wide variety of jet pump applications exist, ranging from single-phrase gas flows to two-phase flows which employ liquid as the primary fluid and gas as a secondary fluid.
Jet-pump optimization use to be a practical and highly effective method to reduce pumping costs by making some changes to the initial geometry. The optimal jet-pump can be defined as a pump that optimizes particular objectives, while fulfilling system constraints. The efficiency of annular air-jet pumps with multi-hole ring nozzles is less that of central air-jet pumps. One of the key design issues limiting the wider exploitation of jet pump technology is the dramatic reduction of the efficiency and pressure lift ratios that occur due to the onset of thermodynamic shocks within the nozzle. The baseline configuration, representing the current state-of-the-art according to the call topic, is composed of four major parts: a nozzle, a convergent sector, a mixing part, and a diffuser. The flow associated to the baseline configuration is characterized by an induced secondary flow motion in the smaller diameter pipe (i.e. nozzle) after the bend and an acceleration of the flow on the side with the larger radius caused by the bend. On the lower radius side of the bend, just downstream of it, flow separation is possible. Large gradients in the flow variables are to be expected downstream of the nozzle’s lip, where jet-like shear-layers will develop. Flow is entrained from the upstream part of the geometry and accelerated in the lower diameter mixer region. In the baseline configuration, there are a few regions susceptible to flow separation. These are the transition region between the convergent and mixing parts and the slope of the diffuser. Additionally, in the larger diameter duct, the entrained flow upstream of the convergent region will directly impinge on the smaller diameter pipe, which acts as a bluff-body in the flow. A large recirculation region and a wake will form due to this interaction. All these flow phenomena are generating flow instabilities and affect the acoustics of the problem and the performance of the device. The optimized design for the jet pump aims at eliminating or limiting such instabilities without affecting its performance.
Progress beyond the state-of-the-art was attempted through the development of a low noise jet pump that incorporates noise suppression technologies without affecting its performance. A few concepts for the silent, i.e. low-noise jet pump were investigated within this project. One such concept was characterized by the presence of flow mixers (e.g. internal chevrons / vortex generators of a particular form) on the circumference of the nozzle. These internal mixers can have multiple roles (e.g. suppress possible axial and helical modes developed in the mixing duct, enhance mixing and homogenize the flow, avoid flow separation). Distributing the injected mass of air uniformly around the pump using fluidics constitutes an advantage if a more homogeneous flow distribution and good mixing at the outlet of the pump is desired, and one of the studied configurations considers replacing the nozzle by fluidics. The use of fluidic injection aims at for enhancing mixing and controlling the flow without causing thrust penalties or an increase in the resistance of the piping system. Moreover, fluidic jets injected in a cross-flow are generating longitudinal counter-rotating pairs off low structures. Fluidics can be an efficient way for flow control, for enhancing flow mixing, or for suppression of noise.
The performance of the proposed configurations was studied via three approaches:

1. Numerical optimization of the subsonic jet pump, through compressible flow simulations using the commercial CFD ANSYS FLUENT software and the SST k-ω turbulence model, which has the ability in accurately predicting induced secondary flow patterns developed in curved conduits.
Within the Computational Fluid Dynamics (CFD) framework, the RANS formulations with the correspondent turbulence closures are very useful in particular for “fast-screening” of a large number of cases or set-ups. Simulating a large number of set-ups in a short period of time and identifying trends in the flow behaviour when performing a parameter study are necessary processes within any design optimization process. This is the reason for which steady-state RANS techniques are the preferred methods by industry for analysis of compressible turbulent flows.
Within the RANS framework, the two-equation turbulence models are used the most. Probably the most popular two-equation model, the standard k-ε model is robust, fast and gives a reasonable accuracy for a wide range of turbulent flows. It allows determination of the turbulence length and time scales by solving two transport equations for the turbulence kinetic energy and the dissipation rate of turbulence kinetic energy. The k-ε model has the possibility to consider the pressure gradient effects at the near-wall region. Another two-equation model is the standard k-ω turbulence model. It is based on Wilcox’s k-ω model with two transport equations for k and the specific dissipation rate (ω) and incorporates modifications for compressibility and shear flow spreading effects. One of the weak points of the model is the sensitivity of the solutions to k and ω values outside the shear layer, which gives poor accuracy when predicting free-shear flows. An improved turbulence model is the SST k-ω model with compressibility effects. The modified turbulent viscosity to account for the transport of the principal turbulent shear-stress gives it an advantage in terms of performance over both the standard k-ω and k-ε formulations. It behaves as a k-ω model in the inner region of the boundary layers and as a k-ε model in the outer region of the boundary layers, in the free-stream zones. Among the steady state RANS models, the SST k-ω formulation shows good potential in providing accurate data in complex nozzle systems when boundary layers as well as jet shear-layers are present in the flow field. Therefore, the SST k-ω was the preferred model for the RANS simulations included in the project.

2. Acoustic measurements in an anechoic chamber and aerodynamic tests that include far field directivity at the jet pump outlet and PIV measurements inside the mixing part.
The directivity index was evaluated by determining the sound pressure level at distance from the jet pump in a given direction, as well as and the sound pressure level on the surface of the test surface of radius equal to the previous distance. A detailed validation plan was developed and carried out in order to obtain reliable results. For this purpose, several tests were performed with a professional and standardized omnidirectional noise source, in order to evaluate the method uncertainty. The performance parameters that were considered are: uncertainty, precision of repeatability, precision of reproducibility, precision of the acoustic sensor position/robustness, and the measurement domain. The uncertainty evaluation considered all the uncertainty sources. Depending on the identified uncertainty sources these were introduced in the overall uncertainty by considering normal, triangular or other type of distribution. Another aspect considered was the structural vibration of the entire system. For this reason vibro-acoustic analysis with autocorrelations / intercorrelations was used to identify potential components/harmonics of the jet pump and applying vibration insulation solutions if is necessary. This will reduce significantly the noise in the exterior of the system. The mentioned issue can appear when the vibration is transmitted to other exterior components which from their nature take the pulsations and vibrate with higher amplitudes, next leading to high noise in the exterior of the system.
The aerodynamic tests were be carried out by measuring the instantaneous velocity field by means of PIV. PIV is an optical method of flow visualization used in research and industry for obtaining instantaneous velocity measurements and related properties in fluids. Small particles are added to the fluid assuming they follow, to a certain degree, the flow dynamics. The motion of the particles is used to calculate the velocity field of the flow. The first to use particles to study fluids in a systematic manner was Ludwig Prandtl, in the early 20th Century, LDV (Laser Doppler Velocimetry) – obtaining all of a fluid's velocity measurements at a specific point, as well as LSV (Laser Speckle Velocimetry) – measuring instantaneous 2-D velocity field in a selected plane of a flow field, predating PIV.
The method required a so-called "seeding" of the working fluid, which is the insertion of fine solid particles, in this case Titanium oxide, in the flow, and the illumination of the experimental area by a medium intensity coherent light sheet. The LASER beam that creates the coherent light sheet is emitted simultaneously with the triggering of two fast cameras that record the image thus formed. By the analysis of the two images thus captured, the displacement of the solid particles in the flow, reflecting the LASER light, is determined. The movement of a cloud of particles is followed on the entire captured image and then scaled through image enlargement, divided by the laser frequency to obtain the flow velocity at each point. Depending on the flow velocity and the optical zoom of the lenses, the interval between the two laser beams must be chosen carefully to obtain an accurate movement of the particles on the CCD camera image. From the time difference between two laser pulsations and the movement of the particles, the velocity components are computed. The first important operation in preparing a PIV measurement is the calibration of the CCD cameras on the measurement zone. This zone has to be parallel with the major component of the velocity, while the CCD camera visual range has to be perpendicular on the measuring zone. Once this is set up, the second important activity is the laser sheet orientation exactly on the measuring zone in order to illuminate the previously injected particles. The PIV recording is split in smaller areas named interrogation windows. For good results the necessary number of particles per interrogation window to be around 10 and this is why the particle injection device must not only has to assure a constant mass flow of particles but also a certain density.

3. Unsteady simulations of the jet pump, through compressible LES, using an in-house developed node based finite volume solver designed to solve complex compressible flow phenomena in a high Mach-number range. Dynamic Mode Decomposition analysis was be used for the post-processing of the LES data, and acoustic propagation calculation numerical methods will be used to simulate the acoustic far field pressure.
The existent in-house numerical tools allow the investigation of the compressible flow and the acoustics associated with the baseline and the RANS optimized configurations design. The full compressible Navier-Stokes equations together with the equation of state were be solved in three dimensions. The solver is capable to make use of agglomeration multigrid to increase convergence for steady-state solutions. The flowing medium is modelled as a calorically perfect gas using the perfect gas law, where the temperature dependence of viscosity is modelled using the Sutherland's formula. The time integration used by the algorithm is a four-stage Runge-Kutta explicit time stepping scheme of second order accuracy. An implicit residual smoothing algorithm is available for time-dependent calculations to accelerate the computation by allowing a larger time-step. For the spatial discretization, second order central difference finite volume scheme is employed. A blend of second and fourth order schemes is chosen as an artificial dissipation procedure to avoid separation of solutions. The Smagorinsky subgrid-scale model is implemented to model the unresolved small turbulent scales. Non-reflecting characteristic boundary conditions are implemented to assure that all waves propagate smoothly out of the domain. The solver is parallelized with Message Passing Interface (MPI) for execution on distributed memory machines.
To gain insights into the three-dimensional and unsteady structures present in the flow field inside of the jet pump, the DMD is used. DMD allows finding the spatial shape of the modes that are associated with certain (given) frequencies of interest found in the flow field at different monitoring locations. This allows describing the temporal evolution of the instabilities in the flow field. Such a tool is very important in particular to the project, where flow control techniques (e.g. chevrons, vortex generators, fluidics) were used for suppressing certain instabilities that develop in the flow field. With DMD, the quantification and the visualization of the developed modes and instabilities at baseline is possible, and the efficiency of the low-noise jet pump in suppressing the unwanted modes by using flow control techniques (e.g. chevrons, mixers, vortex generators) is proved.
The jet pump configurations present a strong interaction and coupling between the flow, the developed acoustics, the solid structures in the flow, and the geometry of the duct. Among the acoustic sources of interest developed in such a situation the acoustic sources due to velocity variations (quadrupole due to turbulence effects), those due to entropy effects, and those due to unsteady pressure loads on the solid surfaces (dipole acoustic source) can be mentioned. Also, when a wave propagates through a channel with a variation in the cross-sectional area’s diameter (e.g. the convergent-mixing-divergent regions of the jet pump), the wave is partly reflected by the surrounding walls at the edge of the constriction and partly transmitted through the duct. For low frequencies, some of the acoustic energy is converted to vorticity in the developed shear-layers (e.g. downstream of trailing edge of the nozzle). However, for higher frequencies under certain conditions, the acoustic wave can interact with the flow in such a way that it extracts energy from the flow, resulting in an increase of acoustic power. If this phenomenon couples to the resonances in the system a high tonal noise (whistle) can be generated. Using compressible LES and DMD technique, the interaction between the flow and the developed acoustic waves was studied in detail. This is because a direct computation of the noise is performed with the LES flow calculations, i.e. the acoustics are obtained directly from the compressible flow solution. However, in turbulent flows the excited acoustic waves are accompanied by turbulent fluctuations. In order to get accurate results, the so called hydrodynamic generated noise had to be suppressed. The acoustic perturbations are characterized by having harmonic time dependence, they propagate with the sound speed, and for low frequencies the acoustic variables are constant over the duct’s cross-sections. These characteristics separate the acoustic perturbations from the hydrodynamic fluctuations that are associated with the turbulent flow. Thus, using this information and following the characteristics based filtering method the suppression of the flow fluctuations from the acoustic ones was achieved and acoustic data became available at any location in the jet pump.
For simulating the acoustic propagation problem in the far-field, the near-field flow domain was extended to the far-field and the acoustic perturbations overlapping the flow field were extracted from the compressible LES solution.

Considering this context, the project objectives were as follows:
1. Numerical optimization of a low noise subsonic jet pump to reduce the noise level;
2. Development of a jet pump demonstrator with several nozzles and diffuser shapes, up to a Technology Readiness Level of 4;
3. Acoustic measurements and aerodynamic tests on the demonstrator, in order to quantify the optimization effects on the noise level;
4. Unsteady simulation of a jet pump simplified geometry and comparison of the numerical data with the demonstrator experimental data.

Project Results:
Due to the need to include figures and table, the description of the main scientific and technical results of the project can be found in the attached document "ST_results.pdf"

Potential Impact:
The impact of the project can be divided into three categories: social, environmental, and economical.
From the social point of view, the project fostered a knowledge increase in the field of jet pump noise propagation and aerodynamics, resulting from both a numerical, and an experimental approach.
New directions of research were opened for the consortium partners, such as jet mixing, secondary flows, and passive noise reduction methods, and the partnerships between the consortium partners was consolidated and further developed.
These new research directions opened up by the project are provided and are expected to further provide new high skilled jobs. At COMOTI, the anechoic chamber used in the project was upgraded for the project needs and was accepted by the Romanian Government and a National Interest Infrastructure, which also allowed the hiring of new personnel.
The project results were disseminated among the student body at KTH and at the partner universities by COMOTI, as well as in public events attracting the young. This will increase the attractiveness of the field for the future young researchers.
The results of the project also provide new scientific information that will support further research studies in the field.
From the environmental point of view, an optimization of the jet pump directly led to a decrease of sound pollution. The reduction in the noise level of the fan is most important impact of this project. Since the jet pump that is the subject of this project is part of the air conditioning system of the aircraft, its location is required to be closed to the passenger and crew cabin, and therefore, the emitted noise can be expected to easily propagate there. Exposure to noise constitutes a health risk, and can induce hearing impairment, hypertension and ischemic heart disease, annoyance, sleep disturbance, and decreased mental performance. Hence, a reduction in the fan noise can reduce the health hazard for the aircraft passengers and crew. In this respect, it must be noted that the project finalized by designing, manufacturing new air jet pump optimized solutions and experimentally demonstrated a noise reduction of 2.5 dBA provided by on of the new solution.
As for the economical point of view, the successful completion of this project led to a new market sectors penetration for the consortium partners, namely the low noise subsonic jet pumps market.
The dissemination of the project results promoted the capabilities and the infrastructure of the consortium, aiming at attracting new economic and research contracts and strengthening the economic viability of the consortium partners. Particularly for the anechoic chamber of COMOTI, the economic impact of the project was very important, since it fostered its upgrade and re-commissioning after a significant time of lack of activity, and it allowed the publication of experimental results gathered by using it. Thus, coupled with its upgrading to a National Interest Infrastructure status, mentioned earlier, attracted already economic contracts for its use, together with its personnel expertise, gained also through the project, by interested third parties. By this, a further improvement of the noise reduction solutions used by the aerospace, automotive, and other industries, may be expected, strengthening the impact of the project.

The main dissemination of the project results included four scientific papers so far:
• M. Mihaescu, B. Semlitsch – “Steady-state and unsteady simulations of a high velocity jet into a Venturi shaped pipe”, 4th Joint US-European Fluids Engineering Division Summer Meeting, Chicago, IL, USA, 3-7 Aug 2014.
• B. Gherman, I. Malael, M. Mihaescu, I. Porumbel – “Jet pump optimization through Reynolds averaged Navier - Stokes analysis”, 22nd AIAA Computational Fluid Dynamics, Dallas, TX, USA, 22-26 June 2015
• L. Dragasanu, M. Deaconu
• F.G. Florean, A.C. Petcu, I. Porumbel, G. Dediu – “PIV Measurements in Low Noise Optimized Air Jet Pump Demonstrators”, AIAA Aviation and Aeronautics Forum and Exposition 2016, Washington. DC, USA, 13 – 17 June, 2016.
Since the aerodynamic measurements were completed close to the end of the project, these results were not yet fully exploited in scientific and technical publications, due to the lack of time. Therefore, further publications will follow after the completion of the project.
The results of the projects were also included in a Chapter of a Ph.D. Dissertation:
• B. Semlitch – “Large Eddy Simulation of Turbulent Compressible Jets”, Ph.D. Dissertation, KTH Royal Institute of Technology, School of Engineering Sciences, Mechanics, Stockholm, Sweden, 2014
The project results have also been presented at a workshop, as follows
• B. Semlitsch, M. Mihaescu – “Functional Modelling of Aircraft Electric Power Systems using Dynamic Phasors”, Annual Linne FLOW Center Meeting, 8-9 January 2014
• B. Semlitsch, M. Mihaescu – “Compressible Jet Flows by Large Eddy Simulations”, Annual Linne FLOW Center Meeting, 8-9 January 2014
A poster presenting the project results was also displayed during an event aimed at enhancing the interaction between the general public and the research world, and is an annual event taking place in over 300 European cities (Researchers for Humanity):
• L. Dragasanu, B. Gherman, F.G. Florean, I. Porumbel – “Optimization of air jet pump design for acoustic applications”, European Researchers’ Night – Romania, Bucharest, Romania, 25th September, 2015.

Related documents

• final1-st_results.pdf