HOMER - Holistic Optical Metrology for Aero-Elastic Research

26 June 2018

HOMER (Holistic Optical Metrology for Aero-Elastic Research) is aiming at deploying and further developing state-of-the-art non-intrusive experimental flow diagnostic methods to include and combine both the aero-dynamic and aero-elastic analysis. As a result fluid-structure-interactions (FSI) experiments realised in wind tunnels and other specific test facilities will deliver an unprecedented combination of knowledge and data from the fluid and the structure domain, with increased coherence and interconnection. The related Technology Readiness Levels to be tackled and achieved within HOMER are spanning from TRL 3 to TRL 6.

Fig. 1: Sketch of dynamics within the Collar triangle of forces

The forces acting on an aircraft in the aeroelastic regime are linked through the equations of dynamics. More schematically, the collar triangle depicts the above interconnection (see Fig. 1).

The particular objective of the project is to develop a previously unattained combined non-intrusive diagnostic approach achieved by simultaneous optical measurements of time-resolved volumetric flow fields (including fluid flow accelerometry) by using 3D PIV/LPT methods, of load distributions on surfaces by pressure from PIV/LPT approaches and by Pressure Sensitive Paint (PSP) and of the dynamics of surface movements resp. deformations by using DIC and advanced point or marker tracking techniques.

Aerodynamic forces (Pressure from PIV, LPT and PSP)

Non-intrusive surface pressure measurement using Pressure Sensitive Paint (PSP) delivers distributed information of the aerodynamic force, which is one of the balancing forces (inertia-, elastic- and aerodynamic forces) acting on the object. The same force can be obtained experimentally from flow-field information, using the PIV- and LPT-based pressure measurement approach. This method essentially relies on invoking the flow governing equations for the exchange of momentum, which allows the pressure gradient to be expressed in terms of the measurement of velocity and acceleration (material derivative).

Fig. 2: Large volume (~48 liters) impinging turbulent jet experiment

High spatial and temporal resolution with a direct measurement of the material derivative (particle acceleration) via STB (see Fig. 2 for a large scale volumetric experiment) or Tomo PTV has been achieved in past experiments that demonstrate the suitability of this measurement principle to tackle both steady and unsteady aerodynamic loads. Moreover, a new multi-pulse STB approach is available which is suited for investigation of high-speed flows and delivers the particle velocity and acceleration along short 4-pulse particle tracks. This new method has been assessed for pressure from LPT methods within the frame of the predecessor project NIOPLEX (see Fig. 3). In the scope of this project a numerical simulation of a transonic base flow was used to provide the basis for synthetic experimental data sets with the purpose of testing and validating pressure-reconstuction techniques. Several different techniques were assessed and compared, ranging from relatively standard two-frame PIV methods, to more advanced procedures based on multi-frame and Lagrangian Particle Tracking approaches. Either time-resolved time sequences were used, or multi-pulse bursts of limited duration, the latter being deemed more realistic to be reached under true high-speed flow applications.

Fig. 3: The NIOPLEX comparative test case

This assessment revealed that under these conditions most methods could satisfactorily reconstruct instantaneous pressure fields, with the LPT-based techniques producing the more accurate results (up to r.m.s. error in pressure coefficient below 0.01) than the PIV-based approaches, which is attributed to a combination of higher spatial resolution of the input data and better use of time information in the data sets. Also, for PIV-based methods, the use of longer series of time-resolved input data allows more accurate reconstructed pressure fields. Therefore, the multi-pulse STB technique is opening a new field for aerodynamic and -elastic flow investigations at high Reynolds numbers and up to supersonic speeds e.g. for Shock-Wave-Boundary-Layer-Interaction (SWBLI) and buffeting with unprecedented accuracies. At the same time the (multi-pulse) STB method enables a very high resolution of mean statistical flow properties as the individual tracks can be averaged in bins of subpixel size which allows for a precise and highly resolved pressure integration e.g. along the whole wing with nose and trailing edge (or any model with strong curvatures) based on the mean or phase-locked velocity and acceleration fields.

Inertial forces in the structure domain

Determining the contribution of the structure inertia forces requires the accurate measurement of its distributed motion and deformation. This is often pursued with a combination of installed sensors (accelerometers) and optical techniques that track the structure as a whole or in parts. Connecting the resulting data with flow measurements is a challenging task as two separate systems need to be jointly operated and mutually calibrated. The unified approach followed in HOMER for cases where the model can be fully immersed into the flow measurement volume is in principle simpler as the structure motion analysis can be performed with the same imaging system used for the flow diagnostics so that the procedures for the analysis of fluid and structure motion will deliver integrated and coherent information. The development of this simplified and faster approach within HOMER (only one system of cameras need to be calibrated) will be based on the advanced LPT technique Shake-The-Box (STB), which will track the fluid tracer particles and distributed surface point markers (advanced point tracking) during the same evaluation step. An uncertainty assessment comparing the different existing and novel surface deformation measurement techniques based on synthetic and experimental test cases will be performed within the scope HOMER.

Fig. 4: Combined PIV and PSP measurement of a generic backward-facing-step configuration at transonic flow speeds (left) and 3D surface reconstruction of a flying bird using optical metrology (right) (UNIBWM)

A thread throughout the project is the description of the motion of wall panel flutter induced by a turbulent boundary layer flow over them. The response of flexible panels to the flow over them is a challenging topic, which requires simultaneous flow and surface measurements. These measurements are typically restricted to small volumes and areas. The goal within the current project is to create and validate experimental tools that will allow to capture both the surface deformation and the flow in a large volume simultaneously. Combining techniques that can use LPT to improve the description of the flow, and obtain pressure fields with surface measurements and including both in data assimilation approaches will allow us to describe panel flutter better and in more detail.

Elastic forces and the vibration regimes

Digital Image- and Digital Volume Correlation and advanced point tracking provides the surface or volume distribution of the displacement and strains of a structure under loading and/or in motion, respectively. The so-called inverse elasticity problem then consists in obtaining the stress distribution as well as identifying the material properties based on the knowledge of the structure's kinematics. For simple systems like a plate or a membrane of known mechanical properties, this information will be sufficient to infer not only the inertia forces but also the elastic forces. For more complex structures, the experimental observations need to be corroborated by structural mechanics modelling and elastic simulations to fill the gap between the surface measurement data and a reliable calculation of inertia- and elastic-forces by building appropriate dynamic structural models based on FEM, such as the FE updating, virtual field method, admissible displacements, or reciprocity gap formulation, can be used to solve the inverse problem. Within HOMER, the connection between the flow and surface kinematic data and the structure dynamical behavior will be pursued by both solving the inverse elasticity problem and by applying data assimilation (DA) procedures in the fluid domain. When needed, the more complex material properties will be identified offline by solving the full inverse problem using pre-defined loadings. The structural designs will therefore be modeled according to their respective constitutive material. DA procedures have already been applied to novel pressure and loads determination schemes based on 2D- /3D-PIV or dense Lagrangian particle tracking (LPT) (by e.g. Shake-The-Box or time-resolved tomo PTV/PIV data) and Navier-Stokes-regularized interpolation by synthetic experiments within the project NIOPLEX and already show very promising results. Innovative DA approaches like VIC+, FlowFit, adjoint methods and 4Dvar are under development in several groups of the consortium and have shown a significant increase of measurement spatial resolution and accuracy compared to the state-of-the-art represented by spatial cross-correlation algorithms.

Fig. 5a: 3D vortical structures and reconstructed pressure fields of three various flexible revolving wings at low Reynolds number

DIC measurement uncertainties have been assessed in a variety of manners, and taking account of multiple experimental and numerical parameters, including the experimental design, pattern deformation, interpolation schemes, or structural discontinuities. Such approaches will be combined to those that are in development for fluid measurement to provide a comprehensive error assessment methodology for the fluid-structure problems adressed in HOMER.

The severity of flow-induced vibration issues not uncommonly arises only in a late stage of the aircraft design process, or even during final flight tests. The counter-measures that have to be applied subsequently can be expected to introduce substantial time and cost burden. The phenomena range from local small-amplitude vibrations that interfere with comfort or fatigue to large-scale LCOs or full-grown flutter that pose the airframe as a whole at serious risk.

In practice, especially in the case of local flow-induced vibrations it is often difficult to differentiate between a genuine flutter instability and a dynamic response problem in which the flow acts as a driver for the structure. In the first case the flow would immediately return to steady state if the involved structure was notionally brought to rest. In the second case the flow would keep oscillating with a certain natural frequency. The latter phenomenon can be masked by the fact that the frequency of the nonlinear fluid-dynamic oscillator can jump to the structural natural frequency once a certain structural response amplitude is exceeded (“lock-in”). An inherent flow unsteadiness can be traced back quite often to the occurrence of shock wave structures interacting with separated boundary layers for instance in the vicinity of wing-body junctions, wing-pylon-nacelle regions or flap track fairings. Other sources of flow unsteadiness are vortical structures shed from upstream or free shear layers for instance originating from engine jets.

Fig. 5b: Results of simultaneous flow, load, and deformation measurements on a membrane wing

Today’s numerical simulation methods are potentially able to cover all the mentioned types of flow unsteadiness, be it motion-induced (relevant for flutter/LCO) or due to large-scale flow instabilities (relevant for dynamic response). The current state-of-the-art is represented by the solution of the (unsteady) RANS equations that can be obtained routinely even for complex flow situation but come at the cost of usually inadequate turbulence modeling. This can be partially overcome by the concept of scale-resolving flow simulation like LES and derived so-called hybrid LES-RANS methods. All methods share a strong need for calibration and validation data for which, as an essential aim of the HOMER project, novel suited optical metrology has to be developed, tested, assessed and demonstrated up to industrial wind tunnel speeds and scales.

Shock buffet and buffeting

The shock buffet phenomenon (SB) is a well-known large-scale flow instability occurring at high angles of attack and moderate transonic Mach numbers around airfoils and wings. It is characterized by the interaction of a suction-side lambda shock system with the turbulent boundary layer. Beyond a critical angle of attack the shock starts to oscillate and large fluctuations of the air loads due to cyclic flow separation and re-attachment are the result. The SB instability manifests itself already far below the actual SB onset by the emerging of a typical resonance peaks in the linearized frequency response functions of the unsteady airloads. These resonance peaks can result in non-classical 1-degree-of-freedom aileron and torsion flutter depending on the location of the flexural axis and corresponding structural eigenfrequencies.

The SB stability boundary can be reliably reproduced with numerical RANS models. The same does not hold for the prediction of the shock oscillation amplitudes itself once the buffet boundary is exceeded. Different RANS and RANS-LES models will produce a large variety of periodic solutions. The situation gets worse when the structural response that cannot be avoided in most practical situations (traditionally termed “buffeting”) is taken into account and the task is, for instance, to predict response amplitudes with and without lock-in. A large portion of that uncertainty originates from the limited understanding of the flow physics behind SB oscillations. Still no testable theory exists to explain the phenomenon. Early attempts that proclaimed a crucial role of a hypothetical acoustic feed-back loop between the shock and the trailing edge collide with the observation of SB over wall-bounded bumps utterly without trailing edges.

Although being a pure 2-D flow phenomenon SB and buffeting are of high relevance for the outer-envelope behavior of modern 3-D transport aircraft wings (cf. Figure 6). Recently, a growing interest of the aircraft industry can be stated in more reliable modelling of the flow physics at high-speed off-design conditions.

Fig. 6: Relation of wing loads and buffet boundaries

In order to make substantial progress in reducing the environmental impact of aircraft, one of the key research axes is the reduction of aircraft weight to contribute to providing a step-change in fuel consumption levels. This challenge is based on the development of innovative configurations or disruptive technologies. It requires especially the development and the assessment of new technologies and methodologies for both structural design and load control. In the specific domain of fluid structure interaction, aeroelastic wind tunnel testing and numerical simulations take as their main objective to improve the understanding of the classical physical phenomena involved in fluid structure interaction such as Flutter, Gust Response, Buffet, Limit Cycle Oscillation etc. The acquisition of comprehensive and relevant experimental databases allows to validate numerical capabilities and tools such as high fidelity tools of Computational Fluid Dynamic (CFD) and Computational Structure Mechanics (CSM). In the process, precious new insights can be gained into innovative configurations involving complex or nonlinear aeroelastic phenomena (High Aspect ratio wing, Truss Braced Wing, etc.), as well as in the efficiency assessment of control strategies of aeroelastic phenomena: Buffet control by active flow control, Gust Load alleviation based on advanced control functions, Flutter margin increase through closed loop approaches etc.

To date, on the numerical and modelling side of these studies, the standard approach classically considers a linear behavior for the structure. The structural model is in this case restricted to a small- displacements small- strains formulation. The structural model may then be handled in the aeroelastic simulation directly using a finite element model (FEM), or via a modal projection of the static/dynamic structural equations. Regarding the fluid model, large efforts has been made during the last decades to move from a linear description of the fluid behavior in aeroelastic simulations (via integral formulations, such the one used in the so-called Doublet Lattice method) to a non-linear one, gradually including compressibility, viscous and turbulent effects, in order to reach a high level of maturity of aeroelastic static and dynamic simulation implementing RANS and URANS fluid models. Attempts are now made to include even higher fidelity aerodynamic models in aeroelastic models using LES or DES formulations, which is still out of reach for complex configurations.

However, only few efforts have yet been made to implement fully non-linear models for both fluid and structure in the aeroelastic modelling. In the static case, accounting of large displacements in the formulation is now possible via the coupling of non-linear fluid solvers and structural solvers such as NASTRAN. However this leads to rather high computing costs, but may still be necessary in the case of non-linear large displacements structural behaviors, such those observed for high aspect ratio flexible wings or rotating blading’s of turbomachines. In the dynamic case, the challenge is even more severe, because of the need of a robust fluid-structure CFD/CSM coupling algorithm in the time-domain, and that of both individually efficient non-linear fluid and structural solvers. Reduced order models may also be used to couple aerodynamic solvers to a non-linear structural model.

Fig. 7: Aeroelastic gust response of a 2DoFs suspended airfoil in transonic flow – (CleanSky SFWA - CleanSky2 NACOR)

At ONERA and DLR, numerical activities have been conducted for years in the development of numerical solution strategies for the modelling of aeroelastic linear and non-linear stability and response problems. These activities are now mainly oriented towards the development of aeroelastic simulation environments complementing the CFD solvers elsA and TAU, widely used in the European aeronautics industry. The purpose of these simulations is the prediction of the in-flight static or dynamic behavior of flexible aerodynamic structures and their aeroelastic stability. The available simulations include non-linear and linearized harmonic forced motion computations, static coupling and consistent dynamic coupling simulations in the time-domain. Harmonic balance method is also implemented for periodic forced motion simulations. A linear behavior of the structure is assumed, but external coupling capabilities have been developed in the last few years, which allow for the coupling of the elsA and TAU solver with a structural solver, potentially implementing non-linear capabilities. These aeroelastic numerical tools have been implemented in a number of internally, nationally or European funded projects (e.g. CleanSky SFWA, CleanSky GRA, CleanSky2 NACOR), see Fig. 7. In the scope of the HOMER project, one may also cite the ONERA-DLR collaborations in the NLAS, NLAS2 and HIFAS common research projects, in particular dealing with LCO aeroelastic phenomena numerical prediction.