Harry R. Kolar

Metallurgical re-examination of wear modes II: adhesive and abrasive

Abstract Adhesive and (three-body) abrasive wear modes in the interface between a 303 steel shaft and a 52100 steel ball-bearing inner race were studied. Scanning electron micrographs showed significant differences in the different stages of adhesive and abrasive wear modes. Further scanning electron microscopy examinations of the cross section of the worn samples showed severely strained layers on the shafts from adhesive wear and heat-affected transformed surface layers on the bearing races from both adhesive and abrasive wear tests. Transmission electron microscopy examinations also showed a severely strained layer in the case of the adhesively worn shaft. A significant increase in hardness was measured on shafts from both wear modes. Auger analysis of worn surfaces detected the existence of modified layers and material removal. The metallurgical techniques described in this paper can be successfully used to diagnose wear.

Complex real-time environmental monitoring of the Hudson river and estuary system

Multiparameter and multiscale real-time environmental monitoring of a river and estuary system will be realized through the River and Estuary Observatory Network (REON) for the Hudson River in New York. In this paper, we describe a system under development that provides a holistic view of this complex and dynamic natural environment for scientific research, education, management, and environmental policy-related applications. The system incorporates a complex array of sensor technologies encompassing the physical, chemical, and biological measurement domains. REON supports Lagrangian, Eulerian, and autonomous robot sensor deployments, as well as flexible telemetry options through an open and consistent middleware architecture with advanced device management capabilities. Multimodal data streams are ingested and analyzed by an intelligent distributed streaming data analysis system known as System S. The challenges of managing high volumes of complex heterogeneous data are addressed via a distributed network of intelligent computational nodes that incorporate both autonomic algorithms and active knowledge management including a temporal component. REON provides an adaptive computing environment that provides isotropy in terms of data access and collaborative computation in contrast to traditional hierarchical control systems for sensor environments. Also presented is the underlying information infrastructure that supports a robust and integrated data modeling, simulation, and visualization manifold.

Embedding intelligent decision making within complex dynamic environments

Decision-making is a complex and demanding process often constrained in a number of possibly conflicting dimensions including quality, responsiveness and cost. This paper considers in situ decision making whereby decisions are effected based upon inferences made from both locally sensed data and data aggregated from a sensor network. Such sensing devices that comprise a sensor network are often computationally challenged and present an additional constraint upon the reasoning process. This paper describes a hybrid reasoning approach to deliver in situ decision making which combines stream based computing with multi-agent system techniques. This approach is illustrated and exercised through an environmental demonstrator project entitled SmartBay which seeks to deliver in situ real time environmental monitoring.

Collaboration platforms in smarter water management

Water resource management, delivery, and research are inhibited by fragmented data sources. It is nearly impossible for public officials to make informed planning decisions that benefit water wholesalers, retailers, and consumers or to efficiently operate water systems beyond their physical and organizational boundaries. Organizations operate water systems within their service areas in ways that may be suboptimal for the sustainable management of the resource as a whole. The organizations make decisions on the basis of available data, which may be incomplete or in the wrong spatial or temporal scale. Data is not shared with other organizations whose decisions and conclusions could be improved with more complete information. This can lead to more complex and fragmented water management decision-making processes that do not address the entire water resource. Sonoma County Water Agency (SCWA) in California and the SmartBay project in Ireland use advanced information technology to create collaboration platforms enabling multi-organizational management of water resources, based on information availability and sharing. While SCWA is deploying such a platform for pilot testing in early 2011 and SmartBay has been operating since late 2008, both projects provide the opportunity to overview the core components and technologies of collaboration platforms and the qualitative benefits (environmental, economic, financial, and political) that can result for water resource management.

A real-time stream storage and analysis platform for underwater acoustic monitoring

We describe a distributed, real-time system for the collection and analysis of underwater acoustic data. The system uses a number of preprocessing steps to classify and detect acoustic events and to identify and compensate for gaps in the data stream. Different event-detection techniques are applied in a distributed manner on the incoming data stream from each sensor to aid in the indexing and storage of the data. Other event-detection techniques process multiple simultaneous streams to identify and classify events of interest. Building upon the deployed system, a stream analytical platform provides data handling, preprocessing, and analytics in real time. These analytics identify and classify anthropogenic, environmental, and animal noise (a significant amount of which occurs outside the audible range of human hearing) and ascertain the direction of the noise source.

A vector-sensor based approach to noise monitoring in Galway Bay

The goal of this analysis is the separation of artificial sound from the ambient or background soundscape in real-time via singular value decomposition of a 3×3 covariance matrix obtained from the set of vector measurements sourced from a particle velocity sensor. The summary noise power is then computed for each octave band, and this is used to detect the presence of artificial, polluting sound sources.