Performance Evaluation, Emulation, and Control of Cross-Flow Hydrokinetic Turbines
Cavagnaro, Robert John
MetadataShow full item record
Cross-flow hydrokinetic turbines are a promising option for effectively harvesting energy from fast-flowing streams or currents. This work describes the dynamics of such turbines, analyzes techniques used to scale turbine properties for prototyping, determines and demonstrates the limits of stability for cross-flow rotors, and discusses means and objectives of turbine control. This involves a progression from the analysis of a laboratory-scale prototype turbine to the emulation of a field-scale commercial turbine under realistic control. Understanding of turbine and system component dynamics and performance is leveraged at each phase, with the ultimate goal of enhancing the efficacy of prototype testing and enabling safer, more advanced control techniques. Novel control strategies are under development to utilize low-speed operation (slower than at maximum power point) as a means of shedding power under rated conditions. However, operation in this regime may be unstable. An experiment designed to characterize the stability of a laboratory-scale cross-flow turbine operating near a critically low speed yields evidence that system stall (complete loss of ability to rotate) occurs due, in part, to interactions with turbulent decreases in flow speed. The turbine is capable of maintaining 'stable' operation at critical speed for short duration (typically less than 10 s), as described by exponential decay. The presence of accelerated 'bypass' flow around the rotor and decelerated 'induction' region directly upstream of the rotor, both predicted by linear momentum theory, are observed and quantified with particle image velocimetry (PIV) measurements conducted upstream of the turbine. Additionally, general agreement is seen between PIV inflow measurements and those obtained by an advection-corrected acoustic Doppler velocimeter (ADV) further upstream. Definitive evidence linking observable flow events to the onset of system stall is not found. However, a link between turbulent kinetic energy of the flow, the system time constant, and the turbine's dynamic response to turbulence indicates changes in the flow occurring over a horizon of several seconds create the conditions under which system stall is likely. Performance of a turbine at small (prototype) geometric scale may be prone to undesirable effects due to operation at low Reynolds number and in the presence of high channel blockage. Therefore, testing at larger scale, in open water is desirable. A cross-flow hydrokinetic turbine with a projected area (product of blade span and rotor diameter) of 0.7 m^2 is evaluated in open-water tow trials at three inflow speeds ranging from 1.0 m/s to 2.1 m/s. Measurements of the inflow velocity, the rotor mechanical power, and electrical power output of a complete power take-off (PTO) system are utilized to determine the rotor hydrodynamic efficiency (maximum of 17%) and total system efficiency (maximum of 9%). A lab-based dynamometry method yields individual component and total PTO efficiencies, shown to have high variability and strong influence on total system efficiency. The method of tow-testing is found effective, and when combined with PTO characterization, steady-state performance can be inferred solely from inflow velocity and turbine rotation rate. Dynamic efficiencies of PTO components can effect the overall efficiency of a turbine system, a result from field characterization. Thus, the ability to evaluate such components and their potential effects on turbine performance prior to field deployment is desirable. Before attempting control experiments with actual turbines, hardware-in-the-loop testing on controllable motor-generator sets or electromechanical emulation machines (EEMs) are explored to better understand power take-off response. The emulator control dynamic equations are presented, methods for scaling turbine parameters are developed and evaluated, and experimental results are presented from three EEMs programmed to emulate the same cross-flow turbine. Although hardware platforms and control implementations varied, results show that each EEM is successful in emulating the turbine model at different power levels, thus demonstrating the general feasibility of the approach. However, performance of motor control under torque command, current command, or speed command differed; torque methods required accurate characterization of the motors while speed methods utilized encoder feedback and more accurately tracked turbine dynamics. In a demonstration of an EEM for evaluating a hydrokinetic turbine implementation, a controller is used to track the maximum power-point of the turbine in response to turbulence. Utilizing realistic inflow conditions and control laws, the emulator dynamic speed response is shown to agree well at low frequencies with simulation but to deviate at high frequencies. The efficacy of an electromechanical emulator as an accurate representation of a fielded turbine is evaluated. A commercial horizontally-oriented cross-flow turbine is dynamically emulated on hardware to investigate control strategies and grid integration. A representative inflow time-series with a mean of 2 m/s is generated from high-resolution flow measurements of a riverine site and is used to drive emulation. Power output during emulation under similar input and loading conditions yields agreement with field measurements to within 3% at high power, near-optimal levels. Constant tip-speed ratio and constant speed proportional plus integral control schemes are compared to optimal nonlinear control and constant resistance regulation. All controllers yield similar results in terms of overall system efficiency. The emulated turbine is more responsive to turbulent inflow than the field turbine, as the model utilized to drive emulation does not account for a smoothing effect of turbulent fluctuations over the span of the fielded turbine's rotors. The turbine has a lower inertia than the demand of an isolated grid, indicating a secondary source of power with a similar frequency response is necessary if a single turbine cannot meet the entire demand. Major contributions of this work include exploration of the system time constant as an indicator of turbine dynamic response, evidence a turbine experiences system stall probabilistically, a reduced-complexity field performance characterization methodology, and demonstration of the effectiveness of electromechanical emulators at replicating turbine dynamics.
- Mechanical engineering