This work overviews a physics-informed implementation of filtered Rayleigh scattering (FRS) to reconstruct the mean field of turbulent jets, despite the presence of noise in the FRS data. The physics-informed FRS methodology is applied to three different jet flows: a noise-free simulated test case, a Mach 0.8 jet flow with low levels of noise, and a Mach 1.25 jet flow with high levels of spectral noise. Using sparse observations of FRS signals, the physics-informed FRS is able to predict the mean axial velocity fields within 3% accuracy and can reconstruct continuous mean field profiles in high noise scenarios where traditional FRS processing introduces discontinuities. To the best of the authors' knowledge, this is the first implementation of a physics-informed neural network that relies exclusively on measured FRS data and is trained using two key components: (i) the compressible Reynolds-averaged Navier-Stokes residuals, and (ii) the residual between the measured FRS signals and a modeled FRS signal derived from the physics-informed FRS model predictions, where the FRS measurement model is incorporated directly into the training process.

Particulate induced erosion has proven to be detrimental to turbomachinery over the past several decades. Even with this longevity, the complexities associated with particles and engine surfaces have inhibited the development of simple and efficient particle bounce models. Typically, particle behaviors are statistically deduced through the study of the coefficient of restitution (COR) for impacts on aerospace materials. This work builds upon previous developments to derive a normal COR model that better accounts for the effects of particle geometries in particle-surface interactions. The model incorporates the jagged, irregular shapes directly measured from particles while also modelling the deformation of the particle and surface individually. As a result, the model natively predicts surface deformation that occurs due to particle impingement. While this new model has been developed to apply to general particle impact conditions, it is compared with measured normal coefficients of restitution for 212 mu m - 250 mu m sieved quartz normally incident at speeds between 20 m/s and 120 m/s on various aerospace-grade materials. Moreover, predicted erosion stemming from the model is compared directly with surface deformation measurements taken from the particle impact test procedure. It is shown that utilizing measured particle geometries in normal COR models leads to more robust predictions that accurately predict average COR values and estimate some of the stochasticity. It is also demonstrated that surface measurements of plastic deformation present on tested surfaces used in COR tests align with the predicted deformation. Therefore, these measurements can also serve as an auxiliary validation of the normal COR model.

Abstract A physics-informed filtered Rayleigh scattering model is applied to a supersonic jet flow with a core flow at ambient temperature and a heated bypass flow. This model serves as a surrogate for larger-scale flows like turbofan engine exhaust. The limited spatial resolution allows for a high uncertainty test case for the force decomposition methodology. The method is evaluated using a CFD simulation and measured FRS data. The simulated jet flow provides a noise-free baseline, while the experimentally measured jet flow demonstrates the reduction in uncertainty even with measurement noise. The net force from all results compares within ±10 N (≈ 3% of net force value). The net force is decomposed into core and bypass contributions using the maximum of the gradient of axial velocity, static temperature, and static density. All three gradient approaches compare within ±8 N (≈ 2% of net force value), and the bypass force contribution uncertainty can be reduced by a factor of two relative to previous non-intrusive force decomposition techniques. This demonstrates an advancement of the filtered Rayleigh scattering force decomposition approach through its increased capacity to handle noisy data, enabling its deployment in ground test and on-wing applications.

It is shown that the precision of multi-flow field parameter estimation with frequency-scanning filtered Rayleigh scattering (FSM-FRS) degrades at high temperature due to the Doppler broadening of the Rayleigh scattering signal. This is shown in FRS measurements acquired on a high-temperature reacting flow facility with total temperatures up to 1300 K and Mach 0.3, validated against a thermocouple. Through singular value decomposition of the FRS model as well as examination of the Fisher information matrix, error covariance, and parameter uncertainty estimates, the signal broadening is linked to a reduction in parameter sensitivity and parameter cross-sensitivity. The resulting loss in precision manifests as an error of up to 33% when compared against the probe in the high signal broadening regime. Therefore, it is recommended to increase the number of frequency scanning points FRS measurements are taken at to increase Fisher information and thus improve parameter estimation, or avoid measurement setups with high amounts of Doppler broadening.

The origin and characteristics of large-scale pressure fluctuations have been attributed to the large-scale coherent motions in turbulent boundary layers, often termed "superstructures." Existing studies lack a few key aspects, including limitations to lower Reynolds numbers, significant spatiotemporal aliasing, and limited measurement domains. This study uses synchronous wall-parallel stereoscopic particle image velocimetry (PIV) and wall-pressure measurements to quantify the characteristics of large-scale flow structures and their relationship with wall pressure fluctuations in a smooth-wall turbulent boundary-layer flow. These measurements are acquired at a friction Reynolds number Re tau approximate to 3300 in the upper part of the logarithmic layer (x2+approximate to 250) under the influence of a small pressure gradient (beta=-0.31). Space-time correlations reveal the presence of a correlation between wall pressure and velocity fluctuations. The wavenumber-frequency cross-spectra between wall pressure and velocity components are presented, showing the direct contribution of the streamwise and wall-normal velocity components in generating wall-pressure fluctuations, while the spanwise velocity component is considered insignificant.

This Letter addresses Rayleigh-Brillouin scattering (RBS) modeling methodology for binary mixtures of nitrogen and gaseous hydrocarbon fuels in the kinetic scattering regime, with its extension to filtered Rayleigh scattering (FRS). Modeled mixture FRS signals were generated by summing mole-weighted FRS signals of mixture constituents, with the required RBS lineshapes for the pure species modeled using the widely accepted Tenti S6 gas kinetics model. The modeled FRS mixture signals were then compared with measurements, and noticeable discrepancies up to 21% were observed. Such deviations demonstrate that the conventional FRS modeling framework for the mixtures of complex hydrocarbon fuels in the kinetic scattering regime requires refinement, with a specific treatment of diffusive effects.

This study applies an innovative control volume analysis method combined with filtered Rayleigh scattering (FRS) to perform force measurements in the exhaust flow of a Honeywell TFE731-2 turbofan engine at Virginia Tech's TurboLab. A particle swarm optimization technique was employed to locate the best fit between the modeled and measured FRS spectra, resulting in a total force measurement of 8995.56 N with a 0.13% deviation from the force balance measurement of 9007 N. To separate the core and bypass flow streams, the Townsend approximation was applied to identify the dividing streamline as the location of the peak shear stress within the mixing layer. A turbulent shear stress mixing length model was used to approximate the maximum shear stress location at the peak of the squared axial velocity gradient. This approach resulted in a core force of 4085.22 N and a bypass force of 4910.34 N. The bypass ratio was experimentally determined to be 2.53, with an error of 2.3% compared to the manufacturer's value of 2.60. Uncertainty quantification, which incorporates random, bias, and control volume sources, resulted in uncertainties of approximately 0.25% for total force, 11% for core force, and 9% for bypass force, with higher uncertainties of the decomposed force attributed to boundary definition errors. In general, this work demonstrates the successful integration of FRS and control volume analysis for accurate force decomposition in complex exhaust environments and laid the ground work for in-flight measurement development.

A control volume framework is derived to decompose forces produced by a turbofan with special considerations for application to experimental measurements. The rich history of aerodynamic force decomposition is reviewed. The parallel expanded streamtube (PES) method of Drela was chosen as the best fit for application to experimental data because it allows for realistic measurement types and locations, especially for optical methods. The PES method is extended here using a combination of the Townsend criterion and Prandtl's turbulent mixing length model to decompose the exhaust stream into the core and bypass flow of a turbofan engine. The methodology uncertainty including random, bias, and control volume error is derived using the propagation of uncertainty technique. Finally, a simple analytically simulated data set is used to test the extended PES (EPES) method and compute the uncertainty. The total error is then presented along with the error influence coefficients assuming uncertain variables such as axial velocity, specific heat ratio, specific gas constant, static temperature, static density, freestream static pressure, freestream velocity, and measurement area. The largest sources of uncertainty were found to be the static temperature and density. In general, the EPES method showed approximately a 1% error in total force, a 2.5% error in core force, and a 2% error in bypass force. This work is applied to real data in Part II of this paper using the filtered Rayleigh scattering technique to gather inlet and exhaust values to measure ground force on a Honeywell TFE-731 turbofan at Virginia Tech's TurboLab.

The observed stochasticity in rebounding trajectories for particle-surface interactions in turbomachinery is often attributed to irregularities in particle shape. However, at higher speeds, plastic deformation of the surface can significantly influence the rebounding particle trajectory. This work explores the interaction between particle shape and surface plastic deformation at high speeds and the effect that it has on the observed spread in rebounding trajectories. Two parameters often used to characterize particle-surface interactions in turbomachinery are the coefficient of restitution (COR) and rebounding angle. A fully time-resolved two-dimensional Particle Tracking Velocimetry (2DPTV) system is employed on a high-speedfree jet rig to study particle-surface interactions for 150 mu m 250 mu m crushed quartz incident on Grade 4 (Commercial) Titanium. Particle speeds range between 50 m/s and 130 m/s and nominal incidence angles of 30 degrees and 90 degrees are studied. In this work, a novel approach that couples the study of both the particle and particle surface interactions is discussed. In addition to a full description of the experimental procedure, a source of random uncertainty stemming from particle shape coupled with rotation and interpixel effects is presented for both velocity and rebounding angle measurements. Moreover, the sensitivity of particle rebounding trajectories to particle shape is explored. A model from the literature is modified to include the effects that particle shape has on the normal coefficient of restitution. The sensitivity study of particle shape is then expanded to total COR and rebounding angle. It is shown that increases in plastic deformation decrease the sensitivity of the COR (both total and normal) to particle shape, reducing the spread of COR values. In contrast, the spread in rebounding angles is dominated by particle shape irregularities and incidence angle with little dependence on plastic deformation.

Sub-convective wall pressure fluctuations are compared for a smooth and rough surface with homogeneously distributed roughness elements. Limitations in measurement techniques make it challenging to measure the sub-convective pressure spectrum over smooth surfaces, and there are no general techniques for rough walls. The study presented here uses a recently developed approach to accurately measure the sub-convective pressure fluctuations, particularly at low-wavenumbers for a single flow condition for two surfaces. Correlations between different sensors suggest smaller scales in rough walls than in smooth walls. A significant expansion of the convective ridge is noticed. Normalizations of the wavenumber-frequency spectrum as phi(pp)(k(1)delta*, omega delta*/U-c)U-c/tau(2)(omega)delta*(2) shows a near-exact collapse at the convective peak. The normalized spectral levels show a difference of about 25 dB in the sub-convective domain due to rough wall convective ridge behavior. The absolute sub-convective pressure spectrum levels relative to 20 mu Pa are about 30 dB below the convective pressure fluctuations for smooth walls. Rough wall measurements show absolute levels 15-20 dB higher than the smooth wall. A wavenumber-white behavior is observed at higher frequencies for smooth and rough walls. Both smooth and rough walls show a convection velocity dependent on frequency. The rough wall convective ridge width shows a frequency dependency, whereas the smooth wall convective ridge width collapses at all frequencies.

This study investigates how particle separation efficiency behaves differently in a single central vortex tube separator (VTS) tube operating in isolation compared to when it functions as part of a multi-tube array. Specifically, it aims to establish the sensitivity of separation performance to array-induced aerodynamic effects by examining the mechanisms responsible for observed differences between the canonical single-tube and seven-tube configurations. Using a custom-designed particle separation rig, a wide range of test dusts with standardized military and industrial specifications were tested for separation efficiencies. For ISO 12103-1 A4 Coarse Dust, the 7VTS configuration exhibited a significantly lower mean separation efficiency (74.73 +/- 1.23%) compared to the 1CVTS (83.36 +/- 1.62%). This trend was consistent across other fine-particle distributions: AFRL 03 and ISO-A3 Medium showed 66.60 +/- 2.14% and 68.70 +/- 3.11% efficiency, respectively, in 7VTS. In contrast, larger particles such as MIL-E-5007C and Sieved Quartz achieved 88.13 +/- 1.39% and 86.85 +/- 3.15%, respectively, with smaller differences between the two configurations, where for the 1CVTS case, MIL-E-5007C achieved 89.46 +/- 0.87%. Results show that the single VTS achieves higher separation efficiencies than the array. The study provides compelling evidence for substantial loss of particle separation efficiency when tubes are integrated into arrays, begging further detailed study to understand and optimize array configurations. To better predict separation behavior across diverse particle types, an effective Stokes number framework was additionally developed, incorporating nonlinear drag behavior and shape corrections through empirically derived drag models.

Exposure of propulsion gas turbines to inlet flow contaminated with dust, sand, or ash particulates can lead to a myriad of complex and interrelated damage modes that reduce engine operational life, increase maintenance costs, and pose a safety risk to passengers and hardware assets. Experimental and computational research is ongoing to better understand the fundamental physics underlying this phenomenon, but data from full-scale engine tests with particles are needed for anchoring and validation under fully representative conditions. In this study, compressor blade/particle interactions are investigated at field-relevant conditions using Rolls-Royce/Allison M250-C20C turboshaft engines in an instrumented engine test cell. A novel experimental dataset was produced, yielding a qualitative visualization of particle impact regions on blades and vanes of an on-engine full six-stage axial compressor at transonic tip speeds for two particle compositions and two inlet particle delivery configurations. This investigation contributes the first experimental dataset of its kind for a rotating frame at transonic blade tip speeds (nominal Mach 1.0). By comparing the resulting impact patterns produced in this work to those of fielded hardware, it is shown that for field-relevant high-Stokes number particle conditions at the first-stage rotor, particle/engine dynamics simplify significantly due to ballistic inertial particle behavior. In addition, the spatial distribution of particle concentration and particle velocities across the compressor inlet plane was found to have only minor effects on the resulting particle/blade impact patterns for the two dust injection configurations tested.

This work presents the evaluation of a vorticity transport based reduced order model (ROM) in comparison to similar data created through Ansys CFX simulation for three canonical swirl flow profiles: Bulk Swirl, Twin Swirl, and Quad Swirl. Modal decomposition through singular value decomposition (SVD) is performed on the resulting in-plane velocity profiles for both the ROM and high-fidelity RANS simulations. The accuracy of the ROM is assessed through both conventional error analysis and a novel energy-weighted modal assurance criterion (MAC) methodology. The MAC based error methodology is proposed to distinguish ROM performance across dominant and minimally contributing mode content. The results and conclusions for this research effort show through conventional L1 error analyses a growth of absolute error with respect to propagation distance. However, the effectiveness of the ROM across dominant low wavenumber flow features with near perfect MAC agreement (MAC = 1.0) is simultaneously observed. Further MAC analysis of the high wavenumber modes shows a reduction in model matching as wave number increases suggesting a sensitively to viscous effects.

Producing a screen for use in generating an accurate, complex total pressure distortion, involves solving two distinct problems. First, the basic loss characteristics of wire mesh screens must be established. Second, the interaction between multiple regions with different loss coefficients (as seen in a complex screen) must be accurately accounted for. Both tasks must be solved effectively to design an accurate screen. For the first goal, we present a comprehensive data set of correlations for the losses generated by a wire mesh, stacked on top of a “backer” support screen. While some publicly available data exists for the losses induced by a single mesh on its own, the losses in a real total pressure loss screen are dependent on the interaction between the backer and the loss meshes. As a result, the data presented here includes the effects of the backer. For the second task, as flow encounters a screen composed of many meshes, it redistributes itself such that the flow rate is higher through the low-loss regions and lower through the high-loss regions. A simple parallel flow model, with appropriate physically-justified assumptions, can accurately predict this redistribution, and we demonstrate its effectiveness via comparison with CFD and experiment.

The frequency-modulating filtered Rayleigh scattering (FM-FRS) technique has been developed and applied for the boundary layer measurements. The FM-FRS technique effectively extracts Rayleigh scattering information from noisy low SNR signals caused by intense wall glare. The boundary layer velocity profiles measured by FM-FRS show excellent agreement with an independent pressure probe measurement and the law-of-the-wall, including approximately 100 μ m above the wall. This technique is desirable for the practical applications of the Rayleigh scattering technique for flow diagnostics to regions that were limited due to strong background and low signal-to-noise ratio.

Particle breakage is of great interest within multiple engineering disciplines, as applications exist in powder production, industrial pipe flow, and aircraft engines. While important, not many measurement techniques exist to directly measure particle breakage. This work introduces an image-based measurement technique and algorithm that detects breakage and tracks resulting fragments for high-speed particles. This allows for the breakage probability (or selection function) to be measured directly by comparing the number of particles that bounce to the number of particles that break for a given impact condition. The presented algorithm is also shown to have an uncertainty of no more than +/- 6%. Moreover, resulting fragment velocities and angles are related to initial impact conditions and compared with corresponding particle bounce results. The breakage algorithm is demonstrated on experimental data consisting of 150 mu m - 250 mu m sieved quartz incident on Grade 4 (commercially pure) titanium with speeds between 60 m/s and 100 m/s and a nominal angle of incidence of 30 degrees . It is demonstrated that the technique yields good results as groups of fragments resulting from particle breakage are shown to have the same average rebounding angle and a lower average rebounding velocity compared to particles that bounce. These findings underscore the improved particle breakage measurements that this new technique achieves.

This work focuses on data mining a filtered Rayleigh scattering (FRS) database to diagnose the source(s) of measurement biases in these applied aerodynamic measurements. A qualitative and quantitative bias analysis provides empirical evidence that there is a subset of FRS measurement configurations where the signal model does not agree with measured signals at known aerothermodynamic states of the gas. A root cause assessment of signal bias sources suggests that the key contributing factor is uncertainty in the modeled laser Rayleigh scattering (LRS) signal over the full range of applicable experiment configurations for this optical technique.

This paper investigates depolarization influences on detected filtered Rayleigh scattering (FRS) signals to understand, in general, when it is necessary to model both the polarized and depolarized portion of the signal, and when it is sufficient to just model the polarized portion. The effectiveness and robustness of spectroscopic FRS (i.e. the frequency scanning method) is driven by the ability to accurately model these signals across the full range of operating conditions for the instrument. Due to the molecular Rayleigh scattering process, the polarization of the scattered laser light will be slightly depolarized with respect to the incident light. The analysis in this work shows that for a subset of operating conditions (e.g. lower angles between laser propagation direction and observation direction, and near-ambient thermodynamic state), the molecular vapor filter for the FRS measurement suppresses most of the polarized signal, while allowing the depolarized signal to pass through. As such, there is a need to model this depolarized portion of the signal to obtain accurate measurements across the full FRS operating space. Applying the proposed signal model in this work to an extensive FRS database demonstrates vast improvement in residuals between measured and modeled FRS signals.

In this paper, we investigate large-scale superstructures over a k-type homogeneous rough wall (k(s)(+) = 209) at a speed of 1.3 x 10(6) Rem(-1), under varying pressure gradient conditions characterized by Clauser parameters ranging from beta = -0.6 (Favorable Pressure Gradient) to ss = 0.98 (Adverse Pressure Gradient). Time-resolved stereoscopic wall-parallel PIV measurements, combined with two-point spatial and temporal correlation analyses, reveal the presence of predominantly streamwise-aligned large-scale structures. While these correlations qualitatively capture the spatial and temporal coherence of the flow, only subtle variations are observed across different pressure gradient conditions. To gain deeper insights, kernelized Residual Dynamic Mode Decomposition (kResDMD) was employed to extract coherent modes that are not superstructures themselves but served as the skeletal representation of these structures. This modal characterization enabled us to quantify pressure gradient effects on structure distribution, organization, orientation, and frequency content, providing a more comprehensive description of superstructure dynamics in rough-wall turbulent boundary layers.

In the current study, the applicability of Filtered Rayleigh Scattering (FRS) technique to the fuel mixture fraction evaluation is investigated at elevated pressures. FRS is one of these techniques that has been used by the combustion community and it offers non-intrusive velocity, temperature, and also concentration field measurements, where the latter is crucial to understand fuel mixing through mixture fraction. In this regard, the Rayleigh-Brillouin scattering (RBS) lineshapes of fuel and diluent (nitrogen) are required to extract the fuel mixture fraction information from the FRS images. While it is relatively straightforward to measure the RBS lineshape for pure nitrogen, this is not quite the case for complex hydrocarbon fuels limiting the use of FRS technique at combustor conditions. Therefore, an indirect approach is performed in this study based on frequency-scanning FRS measurements of the combined lineshapes of fuel/nitrogen mixtures. For this purpose, a benchtop experimental rig is configured which enables Rayleigh scattering measurements at elevated pressures for various hydrocarbon fuel/nitrogen mixtures. With different pre-adjusted amounts of propane (C3) and n-butane (n-C4) fuels used, FRS intensities of their mixtures with nitrogen are calibrated for a sweep of fuel mole fraction values at these pressures. In the next step, these measured mixture FRS signals were compared with the Tenti S6 gas kinetic model simulations to assess its validity for the hydrocarbon fuels of interest at high pressures. It was observed that the modeled and measured FRS spectra normalized with reference conditions do not follow each other over the frequency scanning range for both fuels and their mixtures with nitrogen. With that being said, our work proposes a practical evaluation of mixture FRS measurements by replicating combustor-relevant conditions, which is believed to be a starting point for mixture fraction measurements with more complex hydrocarbon fuels in realistic combustor environments.