Rising energy prices and carbon emission standards are driving a fundamental shift from fossil fuels to alternative
sources of energy such as biofuel, solar, wind, clean coal and nuclear. In 2008, the U.S. installed 8,358 MW of new
wind capacity increasing the total installed wind power by 50% to 25,170 MW. A key technology to improve the
efficiency of wind turbines is smart rotor blades that can monitor the physical loads being applied by the wind and then
adapt the airfoil for increased energy capture. For extreme wind and gust events, the airfoil could be changed to reduce
the loads to prevent excessive fatigue or catastrophic failure. Knowledge of the actual loading to the turbine is also
useful for maintenance planning and design improvements. In this work, an array of uniaxial and triaxial accelerometers
was integrally manufactured into a 9m smart rotor blade. DC type accelerometers were utilized in order to estimate the
loading and deflection from both quasi-steady-state and dynamic events. A method is presented that designs an
estimator of the rotor blade static deflection and loading and then optimizes the placement of the sensor(s). Example
results show that the method can identify the optimal location for the sensor for both simple example cases and realistic
complex loading. The optimal location of a single sensor shifts towards the tip as the curvature of the blade deflection
increases with increasingly complex wind loading. The framework developed is practical for the expansion of sensor
optimization in more complex blade models and for higher numbers of sensors.
Filament-wound rocket motor casings are being considered by the United States Army for use in future lightweight
missile systems. As part of the design process, a real-time, minimal-sensing, quasi-active health-monitoring system is
being investigated. The health-monitoring scheme is quasi-active because abnormal loads acting on the structure are
identified passively, the input force is not measured directly, and the curve-fit estimate of the impact force is used to
update the frequency response functions (FRFs) that are functions of the system properties. This task traditionally
requires an active-interrogation technique for which the input force is known. The updated FRFs and the estimated
impact force can then be used in model-based damage-quantification methods. The proposed quasi-active approach to
health monitoring is validated both analytically with a lumped-parameter model and experimentally with a composite
missile casing. Minimal sensing is used in both models in order to reduce the complexity and cost of the system, but the
small number of measurement channels causes the system of equations used in the inverse problem for load
identification to be under-determined. However, a novel algorithm locates and quantifies over 3000 impacts at various
locations around the casing with over 98% success, and the FRF-correction process is successfully demonstrated.
This research demonstrates two methodologies for detecting cracks in a metal spindle housed deep within a vehicle wheel end assembly. First, modal impacts are imposed on the hub of the wheel in the longitudinal direction to
produce broadband elastic wave excitation spectra out to 7000 Hz. The response data on the flange is collected using
3000 Hz bandwidth accelerometers. It is shown using frequency response analysis that the crack produces a filter, which
amplifies the elastic response of the surrounding components of the wheel assembly. Experiments on wheel assemblies
mounted on the vehicle with the vehicle lifted off the ground are performed to demonstrate that the modal impact method
can be used to nondestructively evaluate cracks of varying depths despite sources of variability such as the half shaft
angular position relative to the non-rotating spindle.
Second, an automatic piezo-stack actuator is utilized to excite the wheel hub with a swept sine signal extending
from 20 kHz. Accelerometers are then utilized to measure the response on the flange. It is demonstrated using
frequency response analysis that the crack filters waves traveling from the hub to the flange.
A simple finite element model is used to interpret the experimental results. Challenges discussed include
variability from assembly to assembly, the variability in each assembly, and the high amount of damping present in each
assembly due to the transmission gearing, lubricant, and other components in the wheel end. A two-channel
measurement system with a graphical user interface for detecting cracks was also developed and a procedure was created
to ensure that operators properly perform the test.