• Energy Research

Abstract The design of a three-bladed horizontal axis research wind turbine for wake investigations is described. The turbine has a rotor diameter of 1.3 m and will be operated in a large water towing tank at a submergence depth of 2.5 m, yielding 3.3 % blockage. The test rig is configured to allow for underwater-stereo-particle-image-velocimetry measurements of the near and the far wake at chordwise Reynolds numbers approaching 700 000. The low-Reynolds number airfoil SG6040 is selected to constitute the rotor blades. Cavitation-free operation is assured by assigning a rated angle of attack of 1°. The blades are designed to maintain a constant circulation along the blade span and to minimize tip deflections. The blade pitch angles are adjustable. Various blade designs, with and without passive flow control devices, can be tested by replacing individual blade sections. This eliminates the need for multiple sets of blades. Sensors for acquiring the turbine’s thrust, torque, rotational speed, the azimuthal positions of the blades, and the imposed blade root bending moments are incorporated into the design. The wind turbine simulation suite QBlade is utilized to simulate the turbine characteristics. A model of the entire test rig is derived, representing its structural properties. Results from the analytical verification of the structural model are provided. QBlade is utilized to analyze the test rig design by imposing design loads and calculating the modal properties.

Thermoacoustic instabilities are a serious problem for lean premixed combustion systems. Due to different time and length scales associated with the flow field, combustion, and acoustics, numerical computations of thermoacoustic phenomena are conceptually challenging. This work presents a coupled method for the simulation of thermoacoustic instabilities in low Mach number reacting flows. The acoustics are represented by a reduced order model that can be obtained from network techniques or finite element computations. A detailed chemistry finite-difference zero Mach number solver is used for the small scale flame dynamics. Under the assumption that the pressure is continuous across the flame, the acoustic model can be reduced to a time-domain relation mapping the velocity perturbation downstream of the flame to that upstream. Closure is obtained by the flame code, which delivers the jump in velocity across the combustion zone. The method is applied to an experimental laminar premixed burner-stabilized flat flame Rijke tube, that exhibits strong thermoacoustic oscillations associated with the 5k=4 mode of the geometrical set-up. In addition to the fundamental oscillation, a significant subharmonic response of the flame is observed. Results from the coupled simulation are compared to the experimental data. Good qualitative and quantitative agreement is found.

Combustion instabilities crucially affect the operational range of modern lean premixed gas turbine combustors and must be avoided or kept at low amplitudes. The main uncertainty of current prediction models is the flame describing function (FDF) that characterizes the flame response to high amplitude acoustic forcing. In this work, we present a new FDF model based on linear hydrodynamic stability analysis. This work is in continuation of an earlier study, where the frequency dependence and saturation of the FDF gain of a perfectly premixed flame was linked to the growth rates of the Kelvin–Helmholtz (KH) instability. In this work, we report on FDF measurements in a newly designed swirl-stabilized combustor. We identify two independent mechanisms that determine the flame response. The first stems from swirl-fluctuations that are generated in the swirler and the second stems from the KH instability. The swirl-fluctuations are approximated by a convective time lag model. The KH instability is predicted from linear hydrodynamic stability analysis based on the time-mean flow measured via PIV. A combination of both models leads to a good quantitative agreement with the measured FDF. Besides the practical advantages of predicting the FDF from stationary flow data, the model reveals the mechanisms driving the saturation of the FDF and guides the way out from the black-box treatment of the nonlinear flame response.

The passive Gurney flap has proven being a good mean to alter the static aerodynamic lift on an airfoil. The Gurney flap can be applied to either the suction side for lift reduction or to the pressure side for lift enhancement. Measurements on a FX 63 -137 airfoil with an active Gurney flap were conducted in a wind tunnel to investigated the time dependent behavior of two-dimensional as well as finite flaps. The time-dependent pressure response of the deploying flaps were evaluated locally over the airfoils surface and the spanwise time-dependent lift was evaluated. Results have shown, that the lift response behaves differently for a flap deployment than for a flap retraction, where the convergence generally takes longer. Furthermore, the adjacent flap sections of the finite flap have shown to also experience a dynamic lift change. This response is different than the response in front of the flap, which is due to the influence of to the surface pressure expansion around the finite flap as well as the flap tip vortices. The measurements have shown, that the timedependent lift in the adjacent sections do play a relevant part in the overall lift response of the wing.

Flame stabilization in a swirl-stabilized combustor occurs in an aerodynamically generated recirculation region which is a result of vortex breakdown. The characteristics of the recirculating flow are dependent on the swirl number and on axial pressure gradients. Coupling to downstream pressure pulsations is also possible. In order to fix the position of the recirculation zone, an extended fuel lance was inserted into the burner. An additional benefit of the extended lance was to enable secondary fuel injection directly into the recirculation zone where the flame is stabilized. Tests were conducted with and without secondary fuel injection. The measurements included optimization of the location of the extended lance in the mixing chamber and variation of the amount of secondary fuel injection at different equivalence ratios and output powers. Flow visualizations showed that stabilization of the recirculation zone was achieved. The effect of the extended lance on pressure and heat release oscillations and on emissions of NOx, UHC and CO was investigated. The results were confirmed in high pressure single burner pressure tests and in a full scale land-based test gas-turbine. The lance has been successfully implemented in engines with sufficient stability margins and good operational flexibility. This paper shows the careful development process from lab scale tests to full scale engine tests until the implementation into the field engines.

Linear stability analysis by means of low-order network models is widely spread in industry and academia to predict the thermoacoustic characteristics of combustion systems. Even though a vast amount of publications on this topic exist, much less is reported on the predictive capabilities of such stability analyses with respect to real system behavior. In this sense, little effort has been made on investigating if predicted critical parameter values, for which the combustion system switches from stability to instability, agree with experimental observations. Here, this lack of a comprehensive experimental validation is addressed by using a model-based control scheme. This scheme is able to actively manipulate the acoustic field of a combustion test rig by imposing quasi-arbitrary reflection coefficients. It is employed to continuously vary the downstream reflection coefficient of an atmospheric swirl-stabilized combustion test rig from fully reflecting to anechoic. By doing so, the transient behavior of the system can be studied. In addition to that, an extension of the common procedure, where the stability of an operating point is classified solely based on the presence of high amplitude pressure pulsations and their frequency, is given. Generally, the predicted growth rates are only compared with measurements with respect to their sign, which obviously lacks a quantitative component. In contrast to that, in this paper, validation of linear stability analysis is conducted by comparing calculated and experimentally determined linear growth rates of unstable modes. Besides this, experimental results and model predictions are also compared in terms of frequency of the least stable mode. Excellent agreement between computations from the model and experiments is found. The concept is also used for active control of combustion instabilities. By tuning the downstream reflectivity of the combustion test rig, thermoacoustic instabilities can be suppressed. The underlying mechanism is an increase in the acoustic energy losses across the system boundary.

Unstable thermoacoustic modes were investigated and controlled in an experimental low-emission swirl stabilized combustor, in which the acoustic boundary conditions were modified to obtain combustion instability. Several axisymmetric and helical unstable modes were identified for fully premixed and diffusion flame combustion. These unstable modes were associated with flow instabilities related to the recirculation wake-like region on the combustor axis and shear layer instabilities at the sudden expansion (dump plane). The combustion structure associated with the different unstable modes was visualized by phase locked images of OH chemiluminescence. The axisymmetric mode showed large variation of the heat release during one cycle, while the helical modes showed variations in the radial location of maximal heat release. Closed loop active control system was employed to suppress the thermoacoustic pressure oscillations and to reduce NOx emissions. Microphone and OH emission detection sensors were utilized to monitor the combustion process and provide input to the control system. An acoustic actuation was utilized to modulate the airflow and thus affecting the mixing process and the combustion. Suppression levels of up to 5 dB in the pressure oscillations and a concomitant reduction of NOx emissions were obtained using an acoustic power of less than 0.002% of the combustion power. At the optimal control conditions it was shown that the major effect of the control system was to reduce the coherence of the vortical structures which gave rise to the thermoacoustic instability.

Unstable thermoacoustic modes were investigated and controlled in an experimental low-emission swirl stabilized combustor, in which the acoustic boundary conditions were modified to obtain combustion instability. Several axisymmetric and helical unstable modes were identified for fully premixed conditions. The combustion structure associated with the different unstable modes was visualized by phase locked images of OH chemiluminescence. The axisymmetric mode showed large variation of the heat release during one cycle, while the helical mode showed variation in the radial location of maximal heat release. The helical and axisymmetric unstable modes were associated with flow instabilities related to the recirculating flow in the wakelike legion on the combustor axis and shear layer instabilities at the sudden expansion (dump plane), respectively. A closed loop active control system was employed to suppress the thermoacoustic pressure oscillations and to reduce undesired emissions of pollutants during premixed combustion. Microphone and OH emission detection sensors were utilized to monitor the combustion process and provide input to the control system. High frequency valves were employed to modulate the fuel injection. The specific design of the investigated experimental burner allowed testing the effect of different modulated fuel injection concepts on the different combustion instability modes. Symmetric and antisymmetric fuel injection schemes were tested. Suppression levels of up to 12 dB in the pressure oscillations were observed. In some cases a concomitant reductions of NOx and CO emissions were obtained, however, in other instances increased emissions were recorded at reduced pressure oscillations. The effect of the various pulsed fuel injection methods on the combustion structure was investigated.

The results of stereo particle-image-velocimetry (PIV) measurements are presented in this paper to gain further insight into the wake of a finite width Gurney flap. It is attached to an FX 63-137 airfoil which is known for a very good performance at low Reynolds numbers and is therefore used for small wind turbines and is most appropriate for tests in the low speed wind tunnel presented in this study. The Gurney flaps are a promising concept for load control on wind turbines but can have adverse side effects, e.g., shedding of additional vortices. The investigation focuses on frequencies and velocity distributions in the wake as well as on the structure of the induced tip vortices. Phase-averaged velocity fields are derived of a proper-orthogonal-decomposition (POD) based on the stereo PIV measurements. Additional hot-wire measurements were conducted to analyze the fluctuations downstream of the finite width Gurney flaps. Experiments indicate a general tip vortex structure that is independent from flap length but altered by the periodic shedding downstream of the flap. The influence of Gurney flaps on a small wind turbine is investigated by simulating a small 40 kW turbine in QBlade. They can serve as power control without the need of an active pitch system and the starting performance is additionally improved. The application of Gurney flaps implies tonal frequencies in the wake of the blade. Simulation results are used to estimate the resulting frequencies. However, the solution of Gurney flaps is a good candidate for large-scale wind turbine implementation as well. A FAST simulation of the NREL 5 MW turbine is used to generate realistic time series of the lift. The estimations of control capabilities predict a reduction in the standard deviation of the lift of up to 65%. Therefore, finite width Gurney flaps are promising to extend the lifetime of future wind turbines.

The aerodynamic loads on a two-dimensional airfoil can be altered with an active Gurney flap placed at the trailing edge. The flap can be moved either to the suction side for lift reduction or to the pressure side for lift enhancement. Measurements on a FX 63 -137 airfoil with a finite as well as a full span active Gurney flap were conducted in the wind tunnel. The dynamic wake during the flaps movement was measured with a High Speed Mono PIV system at center-span location. Furthermore, the time-dependent stream-wise pressure distribution was measured with a wake rake at various spanwise positions. The time-dependent pressure measurements as well as the PIV snapshots were further analyzed with Fast Fourier Transformations as well as the Proper Orthogonal Decomposition technique. Results for the static non-moving Gurney flap have shown, that the position of the vortex street is altered depending on the position of the Gurney flap. The average flow-field showed the decamber mechanism which is responsible for altering the static loads. The frequency of the vortex street was found higher for the flap deflected to the suction side than to the pressure side. The wake development during the active flap deployment and retraction process was observed. Lift and drag development could be associated with a trigger to the wake development. It was found, that the change of lift was accompanied by a change of the vortex street in frequency as well as in vortex strength over the flap movement.