GCCS and Rebreather

Diagnosis and Prognosis of Scrubber Faults for Underwater Rebreathers based on Stochastic Event Models

Imperfect CO2 removal mechanisms of CO2 scrubbers often lead to the existence of CO2 in gas inhaled by a diver from underwater rebreathers. This may cause CO2 related rebreather faults and subsequently would increase the risk of human injuries. We introduce a stochastic model for three CO2 related rebreather faults: CO2 bypass, scrubber exhaustion, and scrubber breakthrough. We establish the concept of CO2 channeling that describes the cause of the faults and present a CO2 channeling model based on a stochastic process driven by a Poisson counter. This helps us to investigate how CO2 flow inside the rebreather is affected by CO2 related faults. Fault diagnosis/prognosis algorithms are developed based on the stochastic model and are tested in simulation.

Recent Status of the GCCS implementation for the Long Bay Project

1. Direct interface between the GCCS and the dock server.


A direct interface between the GCCS and the dock server has been added to the GCCS. When a glider is in operation, glider terminal, which is an application provided by Webb to help glider users interact with gliders, keeps a record of all text output on the screen in a log file. The GCCS now directly accesses the dock server to e xtract necessary information such as surfacing position and time, and flow velocity to generate optimal waypoints from log files on the dock server without assistance of third party software. All the files of the collected data by a glider during the dive also contain necessary glider surfacing information for the GCCS to generate a list of waypoints, but it takes time to transmit the data files to the dock server via Iridium. In the new interface in the GCCS, this necessary information is taken from glider terminal log files as soon as it appears on the screen of the glider terminal, so the GCCS can generate a set of optimal waypoints while the collected data files are being transmitted from gliders to the dock server and can send this waypoints list as soon as the glider is ready after the file transmission .


2 . Incorporation of HYCOM in the GCCS


To analyze the behavior of the glider in the Long Bay, we first developed a simple tidal and Gulf Stream current model based on M2 tide and sinusoidal meandering motion of Gulf Stream, and then simulated gliders under the simple current model using the GCCS. Later, t o obtain more realistic simulation results, HYCOM (HYbrid Coordinate Ocean Model, http://www.hycom.org ) was used for glider simulations. In the Long Bay project, the GCCS generates a set of waypoints for a glider over the time horizon of at least 12 hours. In addition to the fact that Gulf Stream meanders and has unpredictable eddies, the ocean flow by Gulf Stream is often too strong for gliders to navigate through to get to the target. To deal with this the GCCS takes into account current or predicted flow velocities over the time horizon to generate optimal path. However, ocean model forecast files contain forecast data over the limited forecast time period, so for multiple days of implementation the GCCS needs to switch between ocean model forecast files corresponding to the date, the flow prediction data of which the GCCS needs. T he GCCS is now able to run for a multiple day experiment, which needs one or more sequential ocean model forecast data files.


3. Simple glider navigation algorithm for virtual mooring (Station keeping)


In the Long Bay experiment, one glider will be trying to maintain its fixed surfacing position , being operated as a quasi-fixed position profiler . In the GCCS, glider s are modeled as particles traveling at constant speed relative to flow. Since the flow is not constant in general, the relative glider speed is not constant due to the influence of flow. If we want a glider to reach a given position, we need a navigation algorithm to create a list of w aypoints for the glider between its current position and the desired position. A simple glider navigation controller is designed to perform this behavior. Figure 1 shows the domain of the Long Bay with two gliders. The one at (-78.1°, 32.95°) near the red dot will perform virtual mooring.


4 . Planned future events


(a) Using HYCOM ocean model forecast data, a hindcast simulation is planned for the time period over which a real glider was deployed by UNC in the Georgia Bight in August 2006, trying to hold station at 30.8 ° N, 81.316 ° W. (b) SABGOM ocean model forecast data, which would have more frequent forecast intervals than HYCOM, will be also provided for a better emulation of the flow field.


Figure 1 . In (a), outer dotted box represents t he domain of the Long Bay project, inner dotted box is used for the m ain focus region for a virtual mooring glider . The colored contours represent SST field in the domain, and the arrows, which cover the entire field, show the flow field in the domain. In (b), it shows the implementation of the simple glider navigation algorithm for the virtual mooring glider in the GCCS. Colored rectangles represent glider surfacing positions with corresponding time in the color bar.