DIRECT-DRIVE ROTARY GENERATOR DESIGN BY GENETIC ALGORITHM FOR OCEAN WAVE ENERGY APPLICATION

Monday October 5, 2009 - Rm. 50, Everitt Lab by Christopher Haller (OSU Graduate Student 2009).

Supplying the world’s future energy needs through the means of present bulk nonrenewable energy production methods is known to be unsustainable in the future. To solve this energy problem, a variety of renewable energy sources are being investigated, one of which is ocean energy. The world’s oceans have both the energy density and capacity necessary to sustain energy needs in the foreseeable future. The research presented herein focuses on the design of a rotating direct-drive permanent-magnet alternating current (PMAC) generator to be used in a low-speed, high-torque ocean environment. A genetic algorithm was utilized for optimization of the time domain model of the PMAC generator. The final time domain optimized generator was then evaluated in a finite element analysis tool to verify results. As a result of this analysis, a target generation capacity of four hundred kilowatts was found to be possible for the PMAC generator with appropriately applied space and weight constraints. The genetic algorithm identified a tradeoff between machine diameter, length, and power capacity while also identifying weight constraint limitations. These results lead to the conclusion that ocean wave energy generators are a feasible means of power production, but that the size and weight costs will be prohibitive at low generator operating speeds.

Control of power converters: A transition towards geometric control.

Conventional voltage mode control (single-loop) suffers from poor line regulation, bandwidth and phase margin. This is usually overcome using peak current mode control (multi-loop). However, it suffers from poor load regulation and high output-impedance. This presentation illustrates a method of improving load regulation using inductor current filtering. However, for a non-minimum phase converter, like a boost converter, the elimination of right-half-plane zero is of prime interest. This can be accomplished by designing a proper current-control scheme, incorporation a freewheeling switch. During light-load conditions, any PWM control results in reduced efficiency because of domination in the switching loss. A new pulse skipping modulation scheme is discussed, which retains improved efficiency for a wide range of input voltage and (light) load current conditions. It results in a monotonic variation in spectral composition, thereby reduced EMI problems. However, most of the control techniques can not ensure time-optimal control because of a lack of adaptation mechanism using analog circuitry.

A geometric control can be effectively used in guiding the switching logic in order to achieve dynamic performance beyond the physical limit of a system. This presentation also discusses how to design a time-optimal switching logic if all the system parameters are known. Its extension towards a stochastic system where one or more parameters are unknown is discussed from a system identification stand-point. Finally, the issues relating to large and small signal stability are discussed.

COMPARATIVE PERFORMANCE ANALYSIS OF DRIVES FOR INDUCTION MOTORS

Monday, March 30, 2009, 4:00 - 5:00 p.m., Room 50 Everitt Lab ------------ Abstract: Motor control will be reviewed and then the presentation will focus on the comparison between two major types of induction motor drives: direct torque control (DTC) and indirect field oriented control (IFOC). The presentation will give an introduction to both DTC and IFOC and then explore the similarities and differences between the two. It will be shown that the drives themselves can be separated from their switching schemes. A parameter sensitivity analysis of the two drives will be given along with simulation results using MATLAB- Simulink and some experimental results using our in-house modular inverter that verifies the theory.

GPU GENERATED FLOW ARROWS

Monday, February 9, 2009, 4:00 – 5:00 p.m., Room 50 Everitt Lab ------------ Abstract: Graphical processing units (GPUs) have been experiencing enormous gains in performance in recent years. This presentation discusses the application of a new GPU feature, the geometry shader, for power system visualization. In particular, an algorithm for efficiently generating flow arrows is presented. The motivation for this work is achieving real time rendering rates for large (wide area) one line diagrams. In large one line diagrams, there tend to be a large number of flow arrows as well as small objects like circuit breakers and buses. Generating these numerous small objects has been found to take a large portion of the time needed to render a one line diagram. The geometry shader allows for the parallel generation of these objects on the GPU, and the timing results indicate that on modern hardware the rendering rates are significantly faster than traditional methods.

The role of self-validated computing in power flow analysis in the presence of data uncertainty

Monday, January 26, 2009, 4:00 – 5:00 p.m., Room 50 Everitt Lab ------------ Abstract: Power flow analysis serves to determine the steady state of the power system for a specified set of load and generation values. It is one of the most intensely used tools in various power engineering applications, including network optimization, voltage control, state estimation and market studies. The most common formulation of the power flow problem – the deterministic power flow – has all input data specified from the snapshot corresponding to a point in time or from a proper set of “crisp” values that the analyst constructs under the assumptions for the system under study, such as the expected generation/load profiles for a certain peak demand condition. The solution for the steady is deemed representative for a limited set of system conditions. However, when the input conditions are uncertain, the analyst fails to know the precise actual conditions in the system and therefore numerous scenarios need to be analyzed to cover the range of uncertainty. Under such conditions, reliable solution algorithms, incorporating the effect of data uncertainty into the power flow analysis, are therefore required. Reliable power flow solution algorithms allow the analyst to estimate both the uncertainty in the input data and in the solution tolerance. In this way, the uncertainty propagation effect is explicitly represented and the level of confidence of power flow studies can be assessed. We discuss the role of self-validated computation based on interval and affine arithmetic. We present illustrative examples of the evaluations that can be obtained.

Low-power energy harvesting for wireless sensors

Monday, December 8, 2008, 4:00 – 5:00 p.m., Room 50 Everitt Lab ------------ Abstract: There is an exponentially growing demand for real-time information in all aspects of engineering systems, from intelligent building controls to health care systems, environmental control systems, structural integrity monitoring, and entertainment and marketing systems. The demand is fueling proliferation of wireless sensors and interactive devices that communicate via low power wireless networks. This seminar introduces some of the key challenges and potential solutions for providing a fundamental resource to wireless devices: Energy. The focus is on ultra-low power applications that require a few microwatts of average power to perform infrequent monitoring and wireless data transmission functions. It is desirable to harvest energy from the environment to maximize lifetime and minimize the profile, volume and weight associated with energy storage. Solutions are being developed to efficiently collect energy from low power solar, RF, thermal, and mechanical fluid flow and vibration sources and transfer it to low leakage batteries and capacitors. Emphasis is on methods to match the converter input to the power source output for maximum energy harvesting at low power levels. Recent results using custom IC solutions demonstrate 50% to 90% converter efficiency in the 5to 500 W input power range. A brief overview of research programs at the Colorado Power Electronics Center (CoPEC) at the University of Colorado will also be presented.

Doubly Excited Brushless Reluctance Machine For Advanced Wind Power Generation

Monday, December 1, 2008, 3:00 – 4:00 p.m., Room 50 Everitt Lab ------------ Abstract:Technologies in wind power generation have made tremendous progress in the last 20 years and wind energy utilization is becoming one of the most viable solutions to support our modern life styles. In this presentation, existing technologies of wind power generation are surveyed and pros and cons of various wind generation technologies are compared. A variable speed system based on Doubly Excited Brushless Reluctance Machine (DEBRM) is described to demonstrate its uniqueness and potentials in wind power generation. Special issues in designing DEBRM for wind power generation are discussed. A prototype DEBRM design is presented to verify the theoretical investigation.

Archive to be expanded over the next few weeks.