Active Flow and Combustion Control 2014

The trend in jet engine development to reduce airfoil count is still ongoing in order to reduce weight and cost without sacrificing performance. In this context highly loaded airfoils have successfully been developed in the last decade. A better understanding of boundary layer transition and separation phenomena on turbine blades in combination with periodic incoming wakes was the key for this success. The Institute of Jet Propulsion of the Universität der Bundeswehr in Munich, Germany was heavily involved in the research effort from the very beginning. Since the flow gets prone to massive flow separation, associated with drastic increase of total pressure losses, further increased aerodynamic loading can only be achieved when applying passive or active methods to control the boundary layer on the blade surface. Reviewing shortly activities with passive devices this paper gives an overview of the research on active boundary layer control with fluidic oscillators on turbine blades.

This paper discusses the effect of a periodical disturbance in the wake of a highly loaded axial compressor on the pressure distribution and the secondary flow characteristics of the turbo machinery stator blades. A large scale axial compressor cascade, consisting of six two dimensional passages, has been used for this investigation. The test rig is equipped with an active flow control system to enhance the operating range of the compressor. Results that have been achieved by means of side wall actuation are also presented.

This contribution deals with learning control algorithms applied as closed-loop flow controllers. The system to be controlled is an experimental linear compressor stator cascade with sidewall actuation. A damper flap device located downstream of the trailing edges of the blades generates periodic disturbances of the individual passage flows. By learning from period to period, the controllers are able to successfully decrease the effect of the disturbance on the pressure coefficient distribution of the blades’ suction surface.

Unsteady flow separation was partially controlled on a double-bladed H-Rotor vertical axis wind turbine model using pulsed dielectric barrier discharge plasma actuators at wind speeds between 4.4m/s and 7.1m/s. With pulsations applied in an open-loop manner on the inboard side of the blades, flowfield measurements showed that the development and shedding of the dynamic stall vortex could be modified. Pulsations were then applied in a feed-forward manner by predetermining the plasma-pulsation initiation and termination azimuth angles. These angles were selected on the basis of wind speed and turbine rotational speed, with the objective of maximizing the net turbine output power. Remarkably, a net turbine power increase of more than 10% was measured. For the purposes of power regulation, a hysteresis controller was applied to the turbine subjected to a fluctuating wind profile. Control produced a 7% increase in net power and a reduction from ±6.5% to ±1.3% in power fluctuations.

Over the last decade numerous experimental and numerical studies have been conducted by DLR together with partners regarding the maturation of active flow separation control technologies at low speed high-lift conditions. The DLR-F15 high-lift airfoil has been established as a common test bed for such kind of investigations within the German national framework and in the scope of European Research. The present paper highlights major results of this research regarding flow separation suppression by means of pulsed blowing.

Using the Coanda-effect in active high lift flaps opens the way to achieve large lift coefficients, needed for the design of aircraft with short take-off and landing capabilities. In transport aircraft applications overall aircraft design requirements also call for low power consumption of the active high-lift system. Moreover, these aircraft need a reasonable stall angle for their operations. The present contribution describes recent research that aims at high-lift augmentation to be achieved with low rates of active blowing and with suited angle-of-attack ranges. The characteristics of well designed internally blown flaps are addressed. Of similar importance is the design of the leading edge where droop noses and slats may be introduced to avoid locally large boundary layer losses. Moreover, blowing of a Coanda wall jet over a flap can be combined with boundary layer suction in order to yield even higher efficiencies of the overall high-lift system.

Actuator and sensor placement can be just as consequential for the performance of localized feedback flow control as controller design. Yet, effective placement is not well understood, and the use of suboptimal placements is common. This manuscript reports descriptions and characteristics of effective actuator and sensor placements for optimal flow control. We review \(\mathcal{H}_{2}\) optimal placements in the linearized Ginzburg–Landau and Orr–Sommerfeld/Squire models of fluid flow. We then analyze the feedback control of these models by relating physical observations with mathematical tools. Although these tools do not fully predict optimal placements, they do reveal patterns that most or all effective placements share. Most notably, effective actuator–sensor placements provide good authority over unstable modes and transient growth, and avoid large time lags between inputs and outputs.

This paper presents the application of an interconnected systems approach to the flow transition control problem. The control system consists of a network of local controllers with dedicated actuators and sensors. The measured feedback signals used by the controllers are the local changes in wall shear force, and the generated control action is a local change in fluid wall normal velocities. The approach presented here does not require periodicity of the channel as was required by most earlier approaches. Thus, in contrast to previously proposed control schemes which operate in Fourier domain, this approach works in physical domain. Secondly it does not restrict the number of interacting units, thus allowing the application of MEMS arrays. In order to synthesize such controllers, first the dynamics of the fluid is converted into interconnected system form. The model is validated using a non-linear simulation environment developed in FLUENT, and by analysing the growth in the transient energy. The model is then used to synthesize an interconnected controller. The simulated closed-loop response shows that the controller can delay the transition by reducing the transient energy.

Worldwide efforts in combustion development focus on reducing carbon dioxide emissions. Besides improving conventional combustion systems, also alternative systems have to be considered as a measure to achieve that goal. The present work aims to increase the usability of gasoline controlled auto-ignition (GCAI) with its potential benefit regarding fuel consumption of up to 30 %.

The operating range of low temperature gasoline combustion is limited by instability and high sensibility regarding the thermodynamic state. In a first step, experimental investigations on a 1-cylinder research engine with Electromechanical Valve Train (EMVT) were carried out to identify measures that extend the GCAI operating range. In particular the injection strategy appears to be an important control value. So, a distinct amount of fuel injected during the intermediate compression by using combustion chamber exhaust gas recirculation helps to decrease the feasible load by up to 62.5 % (at 2000 min^− 1). Increased charge dilution realized by charging or external exhaust gas recirculation influences the maximum cylinder pressure gradient and allows raising the examined load point from IMEP = 4.6 bar to IMEP = 6.7 bar.

The experiments reveal that the auto-ignition process is affected by comparable high combustion variability with negative influence on stability and controllability. Hence, numerical investigations were carried out to gain a deeper understanding of the combustion fluctuations. Using an 1D gas exchange model in combination with computational fluid dynamics (CFD) the origin of those fluctuations can be determined. On the one hand auto-correlations between consecutive combustion cycles are exposed. This demonstrates the high sensibility of the GCAI process for the thermodynamic state. Furthermore, a superimposed stochastic fluctuation caused by a distinct intake pressure noise explains the combustion variations.

The precise process model is used to develop a model-based predictive controller. To that end the complexity of the model is reduced based on 1D gas exchange calculations and the Watson approach. The reduced process model is capable to reproduce the above mentioned fluctuations.

Moderate and Intense Low Oxygen Dilution (MILD) combustion is characterized by substantial reduction of high temperature regions and thereby reduction of thermal NO _x emissions. Beside the application in furnaces, the MILD combustion seems to be also an auspicious concept in stationary gas turbines for simultaneous reduction of NO _x and CO emissions. Nonetheless, the maintenance of MILD combustion for gas turbine relevant conditions, as the high temperature, at different operating points poses a challenge. The reason is the necessary fast mixing of recirculated burnt gases with fresh air and fuel in the combustion chamber. In this work the application of control for dealing with this task is investigated. As research approach the pulsation of the fresh gas is examined. On the one side the potential of using control for reducing the emissions level is evaluated. On the other side it is analyzed which challenges have to be solved in future concerning the control algorithm, if control shall be applied.

Due to its huge complexity, progress in understanding and prediction of turbulent combustion is extremely challenging. In principle, progress is possible without improved understanding through direct numerical solution (DNS) of the exact governing equations, but the wide range of spatial and temporal scales often renders it unaffordable, so coarse-grained 3D numerical simulations with subgrid parameterization of the unresolved scales are often used. This is especially problematic for multi-physics regimes such as reacting flows because much of the complexity is thus relegated to the unresolved small scales. One-Dimensional Turbulence (ODT) is an alternative stochastic model for turbulent flow simulation. It operates on a 1D spatial domain via time advancing individual flow realizations rather than ensemble-averaged quantities. The lack of spatial and temporal filtering on this 1D domain enables a physically sound multiscale treatment which is especially useful for combustion applications where, e.g., sharp interfaces or small chemical time scales have to be resolved. Lignell et al. recently introduced an efficient ODT implementation using an adaptive mesh. As all existing ODT versions it operates in the incompressible regime and thus cannot handle compressibility effects and their interactions with turbulence and chemistry which complicate the physical picture even further. In this paper we make a first step toward an extension of the ODT methodology towards an efficient compressible implementation. The necessary algorithmic changes are highlighted and preliminary results for a standard non-reactive shock tube problem as well as for a turbulent reactive case illustrate the potential of the extended approach.

Heat release fluctuations generated by equivalence ratio fluctuations may interact with the acoustics of a gas turbine combustion chamber leading to unwanted combustion instabilities, which remains a critical issue in the development of low emission, lean premixed gas turbine combustors. The present article addresses this topic by numerical investigations of one-dimensional lean premixed methane/air flames subject to prescribed sinusoidal equivalence ratio fluctuations. Compared to previous investigations, we focus on turbulent conditions and emission predictions using the one-dimensional linear-eddy model (LEM) and detailed chemistry. Within the limitations of the one-dimensional LEM the approach allows to investigate the fully non-linear regime of flame response to equivalence ratio fluctuations under turbulent conditions. Results for different forcing amplitudes and turbulence levels indicate a strongly non-linear behavior for high forcing amplitudes.

If a detonation fails in a tube of a pulsed detonation engine, special measures are needed to refill this tube with fresh air. To initiate these measures, misfirings have to be detected reliably. In this contribution, the acoustic effects of burner tubes on a down-stream plenum are exploited to determine which of multiple tubes did not produce the expected acoustic signature. This is done with a bank of Kalman filters in combination with the Bayes’ rule. To develop and test these methods, a surrogate, non reacting experimental set-up is considered.

Enhanced mixing and conditioning of a transverse jet in Mach-2 cross flow was investigated through the application of a staged pulse detonation injector. NO planar laser-induced fluorescence and high-frame-rate shadowgraph imaging provided measurements for comparison to CFD modeling of the interaction. The large momentum flux of the pulse detonator led to significant redistribution of the plume from an upstream injectant for several milliseconds while simultaneously elevating its temperatures. The result was a conditioning of the transverse jet plume that could be related to increased reactivity if the upstream injectant was fuel. Typical PD exhaust times would enable potential quasi-steady conditioning of the transverse jet plume if pulsation frequencies on the order of 100 Hz are used.

Harnessing detonations for energy conversion and transport applications requires methods for efficient deflagration-to-detonation transition (DDT) over short distances. The results of three different experiments, characterizing different types of obstacles for flame acceleration and DDT are reported in this work. Flame acceleration by obstacles with identical blockage ratio but different geometric details is investigated using light-sheet tomography. Small but distinct differences in propagation speeds are identified, which correspond to the various obstacle geometries. DDT experiments are carried out to investigate these configurations beyond initial flame acceleration observable with high-speed imagery. A strong effect of obstacle spacing on DDT success is observed, indicating an optimal spacing of slightly larger than two tube diameters. A so-called pseudo-orifice is considered in order to recreate the flow behind a mechanical orifice with the same blockage ratio considered in the previous experiments (0.43). The pseudo-orifice injects fluid perpendicular to the flow, creating a circumferential jet-in-crossflow configuration. Particle image velocimetry is conducted in an acrylic water test-rig in order to measure the flow field in several planes in the acrylic combustion chamber model to assess the effect of the pseudo-orifice on the flow.

Fuel-air mixing is a crucial process in low emission combustion systems. A higher mixing quality leads to lower emissions and higher combustion efficiencies. Especially for the innovative constant volume combustion processes ”Shockless Explosion Combustion” (SEC) the mixing of fuel and air is an important parameter, since the whole combustion process is triggered and controlled via the equivalence ratio. To enhance the passive scalar mixing, fluidic oscillators are investigated and compared to the standard jet in crossflow fuel injection configurations. The mixing quality of the different geometries is assessed in a water test-rig by making use of planar laser induced fluorescence. After a short introduction to the SEC-process, the test-rig and the different injection configurations are introduced. To verify whether the mixing quality is sufficient for the SEC-process, a numerical investigation using the experimentally determined unmixedness is conducted. It is not only shown that the fluidic oscillators are able to enhance the mixing quality and create an independence of the mixing quality from the jet in crossflow momentum, but it is also verified in a first numerical calculation that the achieved mixing quality might be good enough for the Shockless Explosion Combustion process.

Shockless isochoric explosion has been proposed as an alternative to conventional constant-pressure combustion in gas turbines to improve the thermal efficiency. Fully homogeneous auto-ignition and fast burnout facilitate quasi-isochoric combustion, but the ignition delay times depend on local temperatures and mixture composition. This implies that homogeneous auto-ignition is difficult to achieve in the presence of local inhomogeneities. Therefore, in this study, the ignition delay time, its temperature dependence and the excitation time of the potential fuels are investigated. An automatic optimization technique is used to tailor fuels such that the temperature dependence of ignition delay times is minimized to realize homogeneous isochoric auto-ignition in realistic systems even in the presence of inhomogeneities and also to avoid the formation of detonation waves by considering the kinetic properties. Reaction kinetic mechanisms are developed for the chemical modeling of the relevant fuel species and tailor-made compositions of the fuel mixtures are determined by means of simulation.

A correct description of unsteady, transient combustion processes controlled by chemical kinetics requires knowledge of the detailed chemical reaction mechanisms for reproducing combustion parameters in a wide range of pressures and temperatures. While models with fairly simplified gas-dynamics and a one-step Arrhenius kinetics in many cases makes possible to solve the problem in question in explicit analytical form, many important features of combustion can not be explained without account of the reactions chain nature, describing qualitatively a few major properties of the phenomena in question with some poor accuracy if any, often rendering misinterpretation of a verity of combustion phenomena. However, for modeling real three-dimensional and turbulent flows we have to use reduced chemical kinetic schemes, since the use of detailed reaction mechanisms consisting up to several hundreds species and thousands reactions is difficult or practically impossible to implement. In this lecture we consider the option of a reliable reduced chemical kinetic model for the proper understanding and interpretation of the unsteady combustion processes using hydrogen-oxygen combustion as a quintessential example of chain mechanisms in chemical kinetics. Specific topics covered several of the most fundamental unsteady combustion phenomena including: the regimes of combustion wave initiated by initial temperature non-uniformity; ignition of combustion regimes by the localized transient energy deposition; the spontaneous flame acceleration in tubes with no-slip walls; and the transition from slow combustion to detonation.

The Richtmyer-Meshkov instability (RMI) arises from the interaction of a shock wave with a density gradient in a fluid. The density gradient can be caused by temperature and/or species concentration. The RMI leads to strong mixing and in the case of reactive flows an existing flame will be accelerated. Such acceleration processes can lead to a transition to detonation. This work presents first results in the simulation of these instabilities driven by a shock induced acceleration. It examines the case of single-mode interface perturbations and compares its growth and behaviour for reactive and non-reactive flows. The chemistry is described by a multiple-species one-step Arrhenius based kinetics model. The skew-symmetric finite difference formulation of the Navier-Stokes equations is used to simulate the flow. Ideal gases of stoichiometric premixed methane-air, i.e. CH ₄ + 2 O ₂ + 7.52 N ₂, are considered. The starting point is the validation of the physical model. Subsequently the role of the RMI in the transition to detonation is analysed by testing the influence of the shock Mach number on the onset of detonations.

Direct numerical simulation of dynamical systems is of fundamental importance in studying a wide range of complex physical phenomena. However, the ever-increasing need for accuracy leads to extremely large-scale dynamical systems whose simulations impose huge computational demands. Model reduction offers one remedy to this problem by producing simpler reduced models that are both easier to analyze and faster to simulate while accurately replicating the original behavior. Interpolatory model reduction methods have emerged as effective candidates for very large-scale problems due to their ability to produce high-fidelity (optimal in some cases) reduced models for linear and bilinear dynamical systems with modest computational cost. In this paper, we will briefly review the interpolation framework for model reduction and describe a well studied flow control problem that requires model reduction of a large scale system of differential algebraic equations. We show that interpolatory model reduction produces a feedback control strategy that matches the structure of much more expensive control design methodologies.

Interpolation based model reduction is applied to a reactive process. A zero-dimensional, perfectly stirred, constant pressure reactor with complex chemistry, modeled by the GRI3.0 scheme, is considered. The aim of this work is to analyze how interpolatory model reduction performs for combustion processes, where the solution is very sensitive to the choice of input parameters. In this study the initial temperature is chosen as varying parameter.