-9-
onerspeed ev1dence of the klnds that were noted durlng development tests. The first eV1dence of overSpeed from tests, perceptlble turblne and compressor tip diameter growth and resultant compressor tlp rub, acours at 20 percent overspeed £15,600 engine r.p.m.). At 1ncrea51ngly hlgher overspeeds, compressor tlp rub is more pronounced, turblne blade tlp rub and some bearlng dlstreSS becomes evldent. No measurable growth of turblne or compressor diameters or bearlng distress of the klnd assoolated w1th overspeed was noted. Based on propeller development work, the flrst ev1dence would be brlnelllng of the blade bearlng races and 1t would occur at about 53 percent (21,120 englne r.p.m ) overspeed. Forty-one percent (19,500 englne r.p.m.) overspeed tests showed no brlnelllng. Io brlnelllng of the klnd that would result from_overspeed1ng was noted on any of
the propeller bearlng races.
Attentlon was dlrected to the No. 3 powerplant by unusual marklngs on the safety coupllng, the 50 percent closed p081t10n of the electrlcally operated 011 shutoff valve and_the totally closed posrtlon of the actuator of the electrlcally operated fuel shutoff valve located Wlthln the fuel Control These shutoff valves are operated.by the cockplt powerplant emergency control whloh among other functlons feathers the propeller. Operating tlmes from "open" to "closed" of these valves are; fuel, .3 to .h seconds and 011, .5 to .97 seconds
The safety coupllng functlons to dlsconnect the propeller from the englne 1n the event other protective dev1ces have felled to functlon and the propeller lS furnishlng energy (negathe torque) by wrndm1lllng actlon to drlve the englne Thls actlon by the safety coupllng 1s generally termed "decoupling" and occurs when negative torque reaches approxlmately 1,700 shaft horsepower. Comparlson of the marks on the 1nner and 1ntermed1ate members of the do. 3 coupllng w1th llke marks on coupllngs known to have Operatlonally decoupled and ratcheted revealed a dlsslmllar pattern. MetallOgraphlc and Vlsual study revealed that hlgh negatlve torque loads Were applled whlle the 1ntermed1ate member was out of alignment w1th the outer member. Impact loads between the 1nner and 1ntermed1ate members were
applled 1n both the pOS1t1ve torque and ax1al dlrectlon.
Saparatlon of the No. l englne at the a1r 1nlet to compresaor caSe Split llne occurred early in the sequence of events as eV1denced by the parts forward of the separatlon llne belng the flrst magor component along the fllght path. Except for a sectlon of the air 1nlet casting flange between 5 00 and?:30 o‘clock locatlon whlch broke away and remalned With the compresaor flange, the l/h—BB cap screws separated by tens1on fallures and the 5/16~2h cap screws pulled the 1nserts from the alr 1nlet castlngs. Cap screw 1nserts whlch pulled out at the 10 00 to 11:00 o‘clock locatlon w1ped metal from the face flange. The dlrectlon of thls Wiplng actlon 1nd1cates the 1nlst hous1ng rotated wltn respect to the compressor about a polnt measured radlallyoutwardat 11:00 o'clock and flve to 31x 1nches out51de of the bolt c1rcle. Dlrectlon of rotatlon was ClOCleSe relative to the compressor case and looklng forward. The a1r 1nlet hou51ng flange showed compress1ve loadlng between the 10-00 to 11:00 o‘clock locatlon A VlSual and metallographlc examlna— tlon,1nd1oated that most of the scrape marks at the holes where the *ushlngs pulled not had been made by the external threads of the bushings and there was no ev1dence found of a reversal of the scraplng dlrectlon or repetltlve movement of tze bushlngs across the Scraped areas. Hardness tests of the stud.and casa materlals were
satisfactory.
larks were made by contact of the leadlng edge of the fvrst stage compressor blades w1th the surface of the shelf just rearward of the 1nlet gulde vanes .
