https://ntrs.nasa.gov/search.jsp?R=20120010444 2019-08-30T20:40:58+00:00Z View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by NASA Technical Reports Server

entrance pupil points of interest. These field or aperture, depending on the allo- nates of interest are obtained for every are chosen to be the edge of each space, cations. Once this minimum set of coor- plane of the propagation, the data is for- so that these rays produce the bounding dinates on the pupil and field is ob- matted into an xyz file suitable for FRED volume for the beam. The x and y global tained, a new set of rays is generated optical analysis software to import and coordinate data is collected on the sur- between the field plane and aperture create a STEP file of the data. This re- face planes of interest, typically an image plane (or vice-versa). sults in a spiral-like structure that is easily of the field and entrance pupil internal These rays are then evaluated at imported by mechanical CAD users of the optical system. This x and y coor- planes between the aperture and field, at can then use an automated algorithm to dinate data is then evaluated using a a desired number of steps perceived nec- wrap a skin around it and create a solid convex hull algorithm, which removes essary to build up the bounding volume that represents the beam. any internal points, which are unneces- or cone shape. At each plane, the ray co- This work was done by Joseph Howard and sary to produce the bounding volume of ordinates are again evaluated using the Lenward Seals of Goddard Space Flight Center. interest. At this point, tolerances can be convex hull algorithm to reduce the data Further information is contained in a TSP applied to the size of either the to a minimal set. When all of the coordi- (see page 1). GSC-16176-1

High-Performance, Multi-Node File Copies and Checksums for Clustered File Systems Ames Research Center, Moffett Field, California Modern parallel file systems achieve performance of clustered file systems. Mcp and msum provide significant per- high performance using a variety of Mcp and msum are drop-in replace- formance improvements over standard techniques, such as striping files across ments for the standard cp and md5sum and md5sum using multiple types of par- multiple disks to increase aggregate I/O programs that utilize multiple types of allelism and other optimizations. The bandwidth and spreading disks across parallelism and other optimizations to total speed-ups from all improvements are multiple servers to increase aggregate achieve maximum copy and checksum significant. Mcp improves cp perform- interconnect bandwidth. To achieve performance on clustered file systems. ance over 27×, msum improves md5sum peak performance from such systems, it Multi-threading is used to ensure that performance almost 19×, and the combi- is typically necessary to utilize multiple nodes are kept as busy as possible. nation of mcp and msum improves veri- concurrent readers/writers from multi- Read/write parallelism allows individual fied copies via cp and md5sum by almost ple systems to overcome various single- operations of a single copy to be over- 22×. These improvements come in the system limitations, such as number of lapped using asynchronous I/O. Multi- form of drop-in replacements for cp and processors and network bandwidth. The node cooperation allows different nodes md5sum, so are easily used and are avail- standard cp and md5sum tools of GNU to take part in the same copy/checksum. able for download as open source soft- coreutils found on every modern -file processing allows multiple ware at http://mutil.sourceforge.net. Unix/Linux system, however, utilize a threads to operate concurrently on the This work was done by Paul Z. Kolano and single execution thread on a single CPU same file. Finally, hash trees allow inher- Robert B. Ciotti of Ames Research Center. Fur- core of a single system, and hence can- ently serial checksums to be performed ther information is contained in a TSP (see not take full advantage of the increased in parallel. page 1). ARC-16494-1

Stiffness and Damping Coefficient Estimation of Compliant Surface Gas Bearings for Oil-Free Turbomachinery Initial applications include design of turbochargers, blowers, compressors, pumps, and turbine engines. John H. Glenn Research Center, Cleveland, Ohio

Foil gas bearings are a key technology valuable tool in system development. of sheet-metal foil layers from which they in many commercial and emerging oil- The current work extends this tool con- derive their name. They are compliant free turbomachinery systems. These cept to include rules for stiffness and bearings that offer high-speed rotor sup- bearings are nonlinear and have been damping coefficient estimation. It is ex- port while accommodating shaft mis- difficult to analytically model in terms of pected that these rules will further accel- alignment and distortion often encoun- performance characteristics such as load erate the development and deployment tered in turbomachinery. Lightly loaded, capacity, power loss, stiffness, and damp- of advanced oil-free machines operating low-temperature foil gas bearings are ing. Previous investigations led to an em- on foil gas bearings. commodities that predominate in the pirically derived method to estimate Foil gas bearings are self-acting hydro- rotor support for aircraft air cycle ma- load capacity. This method has been a dynamic bearings comprised of a series chines (ACMs). More highly loaded foil

32 NASA Tech Briefs, June 2012