General
Teaching
  -Current term
  -Past terms
Study Abroad
Research
Textbook
Contact

Research

Current Active Projects

Active Control of Flexible Rockets

Rocket

Feedback control systems are critical to maintaining the attitude of a rocket, but such control systems traditionally consider that the rocket is a rigid body. New launch vehicle designs, such as the Ares I with a slender first stage consisting of a single, extended Shuttle solid fuel booster and an upper liquid fueled upper stages, cannot be treated as rigid bodies because the fundamental bending modes are very close to the attitude controller bandwidth. Traditional designs using notch filtering to reject such modes may not provide acceptable robustness. Our approach is to design an adaptive controller that utilizes a model reference based neural network scheme to provide the required robustness without sacrificing performance. The neural network is self-training and compensates for differences between the assumed vehicle model and the actual dynamics.

Seismic Fragility of Nonstructural Building Subsystems

Failure of nonstructural subsystems may constitute the majority of damage from moderate to severe earthquakes. However, due to the wide variety of nonstructural subsystems and the broad range of designs for individual subsystems, it is very hard to esitmate the seismic fragility of installed components.Our research focuses on developing methods for rapid seismic fragility estimation for electrical subsystems in buildings. The work was initiated with NSF funding through the Mid-America Earthquake Center and is currently being carried out in collaboration with Schneider Electric/Square D.

Selected Previous Projects

Seismic Response Across Regions

metamodel

This research developed a method to rapidly assess the fragility of structures and geostructures over a specified region. The earthquake response of a structure is affected by the uncertainty in not only the earthquake but also the uncertainty in the characteristics of the structure itself. For an individual structure or geostructure, the uncertainty arises largely from material properties and construction methods, but for a regional collection of structures whose individual characteristics are not known, additional uncertainty arises from macro-level parameters such as structural type, base planform, orientation, as well as vertical and planform irregularities, and the applicable design codes.

Since detailed analysis of each structure or geostructure in the collection is impractical, this research addressed the problem by developing a methodology based on the use of computationally efficient metamodels to represent the overall structural behavior of the collection. In particular, response surface metamodels were developed using a Design of Experiments approach to select the most influential parameters. Monte Carlo simulation was carried out using probability distributions for the parameters that are characteristic of the target collection of structures or geostructures, and the fragility of the collection was estimated from the computed responses.

The basic methodology was applioed to several applications that are relevant to structures in Mid America. These included a single building (an unreinforced masonry firehouse typical of an essential facility) as a reference case, and the effect of seismic rehabilitation on the computed fragility was also investigated. The method was next applied to a class of lowrise steel moment frame structures using a 2D structural model only, and the fragilities of a collection of such structures was estimated. This was followed by an extension involving the use of 3D structural models for a class of lowrise L-shaped steel moment frame structures. Finally, the methodology was applied to two important classes of geostructures that are common to Mid America: (1) earthen and rockfill dams, and (2) levees.

The proposed methodology is suitable for incorporation into advanced GIS-based risk assessment and management software systems for practical applications.

Interdependent Response of Networked Systems

power network

Infrastructure systems such as electric energy, potable water, oil and gas, telecommunications, and the internet are essential for the continuous functionality of modern societies. Different topologies underline the structure of these networked systems. The different topologies (i.e., physical layout) of these networks determines the way the networks transmit and distribute their content. Also, their ability to absorb unforeseen natural or intentional disruptions depends on complex relations between network topology and optimal flow patterns.

Most network studies concentrate on specific networks, but this approach ignores a fundamental fact that most networks are interdependent with other networks and systems.  Thus the performance and functionality of a given network depends on the state of interacting networks. This study focused instead on network topology, flow patterns within the networks, and optimal interdependent system performance.  This approach also allows for probabilistic response characterization of networked systems when subjected to internal or external disturbances.  The response of real and idealized interacting systems is nonlinear with respect to the intensity of the disruptions due to the increased complexity and intractability introduced by their coupling.  Methods developed in this research can identify the role that each network element has in maintaining network connectivity and optimal flow.  This information can be used in the selection of effective pre-disaster mitigation and post-disaster recovery actions.  Results of this research can also provide guidance for resilient network growth and can reveal new areas for research on interdependent dynamics.  Finally, the algorithmic structure of the developed methods suggests straightforward implementation of interdependent analysis in advanced computer software applications for multi-hazard loss estimation.

Passive Damping for Unreinforced Masonry Low-rise Buildings

URM building

Essential facilities are defined as buildings that support functions related to post-earthquake emergency response and disaster management. For such buildings simply insuring life safety and preventing collapse are not sufficient, and the buildings must remain operational during or suitable for immediate occupancy after a major earthquake. In the central U.S., unreinforced masonry (URM) is the most common type of construction for essential facilities, and such material is well known to be highly vulnerable to strong earthquakes. Passive response modification for this type of building, and particularly for low-rise firehouses, was the focus of this research project.

Passive energy dissipation has not been considered for response modification of URM structures because of the very high initial elastic in-plane stiffnesses of URM walls. However, a large number of URM essential facilities commonly include flexible wooden floor and roof diaphragms that couple the URM walls into structures that are more flexible (and complex) than the URM walls alone.
Three response modification concepts were investigated. Type 1 is activated by relative displacements between the in-plane walls and the diaphragm. Type 2 is activated by relative displacements between the diaphragm and the ground, and Type 3 is activated by the relative in-plane displacements within a wall. Only Type 1 and 2 configurations were studied in this project. Tapered metallic flexures were examined as passive damping devices because metallic hysteresis offers good energy dissipation with excellent long-term stability and reliability, which is particularly well-suited for scenarios with infrequent earthquakes.

The passive energy dissipation devices were designed using an energy-based criterion with the objective being to maximize the ratio of the energy dissipated in the devices to the total input seismic energy subject to selected constraints on peak forces and dynamic ductility.

The response modification concepts were evaluated using a highly idealized laboratory test model and a full-scale test structure. The results verified that incorporation of properly designed passive energy dissipators in the rehabilitation of certain kinds of URM structures lowered energy dissipation demand in the main structure, reduced seismic response, and lessened the possibility of damage. Given the inherently stiff nature of URM structures, this approach can only be applied in cases where sufficient deformation can be developed to fully activate typical metallic hysteretic dampers considered in this study. The primary benefit was a significant reduction in out-of-plane wall deformation induced by a flexible floor diaphragm. The effectiveness of this approach was reduced when the diaphragm stiffness for shear mode deformation was more than about 8% of the in-plane wall stiffness, but this is well above the stiffnesses of typical floor and roof diaphragms. Thus the use of a passive energy dissipator as outlined in this study could be considered as a part of a rehabilitation scheme that aims to reduce the deleterious effects of an overly flexible diaphragm system.

Finally, the probabilistic nature of the problem, specifically the variation in building configuration for a regional class of buildings such as firehouses, was taken into account. Fragility curves and damage probability matrices were used to describe the likelihood of damages under different ground shaking intensities. This research made use of meta-modeling techniques through a combined response surface methodology and Monte Carlo simulation approach to include randomness in configurations over a class of buildings in a region. Improvement in the seismic responses when a passive energy dissipation system is implemented were shown to lead to a lowered probability of damage in the buildings..

In-plane Response Models for Perforated Unreinforced Masonry Walls

URM wall

An efficient and simple approach was developed for modeling the in-plane response of perforated URM walls. The wall is decomposed into characteristic regions such as piers and spandrels, and the in-plane behavior of each region is modeled by a single nonlinear spring with empirically determined properties. These nonlinear springs are then assembled into a 2-D parallel-series composite spring model for the complete wall. The boundary conditions for each of these components are modeled by using either an “effective height” or an “effective end stiffness.” An additional factor based on overall wall aspect ratio is applied to reflect overall in-plane bending. The model was calibrated against plane stress solutions and was verified by comparison with experimental results. Good agreement was found for the wall stiffness and strength. Finally, a simple 2-D lumped-parameter building model was developed in DRAIN-2DX and ABAQUS using these in-plane wall models with springs to model the shear deformation modes for flexible floor/roof diaphragms. No consideration was given to out-of-plane wall stiffness but the masses were considered.

Ductile Cladding Connection Systems for Seismic Response Reduction

text fixture

Analytical and experimental studies led by Profs. J. Craig and B. Goodno since the late 1980's have shown that promising levels of seismic response attenuation can be designed into new and existing building structures through use of “advanced” connections for precast cladding systems. Such connectors must be designed not only to attach the cladding panel to the building structure but the can also be designed to provide ductility and energy dissipation by permitting relative displacements between the structure and the cladding panels. A special test fixture was designed by doctoral student J-P. Pinelli who used it to test a family of flexural connectors he developed. Other students designed and tested connectors that used ductile torsional deformation and connectors using composite laminated neoprene and steel isolation pads enclosed within steel flexures for bearing applications.

Nonlinear analytical models developed for the various connector designs were incorporated into existing nonlinear software for time history dynamic analysis of planar structural systems. Optimization of the connector properties for an actual 20 story building application resulted in controlled energy dissipation in the cladding connectors, and reduced demands on the supporting structural framework. Results for this building showed that either up to 41% reduction in peak displacement response could be achieved from the baseline (as-built) configuration by retrofitting advanced cladding connectors, or else as much as a 17% reduction in structural weight (in the longitudinal direction) could be achieved for the same baseline response level. This suggests that use of an energy dissipating cladding system could lead to either improved serviceability (reduced drift) or else a savings in structural steel, or some combination of both.