SRch6

NASA Investigation Board Report On The INITIAL FLIGHT ANOMALIES OF SKYLAB 1

CHAPTER VI : ANALYSIS OF POSSIBLE FAILURE MODES OF THE METEOROID SHIELD

During the course of this investigation, a total of ten possible failure modes of the meteoroid shield were postulated and examined by the Board. These failure modes are presented and discussed, in increasing order of probability, in the following.

1.	Premature Firing of the Meteoroid Shield Separation Ordnance
	The flight data shows that the three ordnance break wire event sensors on the ordnance strap were still intact after the first indication of the MS anomaly. The in-flight firing command was therefore not issued prematurely. The ground firing command to fire the ordnance was not given. It was therefore concluded that the failure of the MS was not a result of a premature firing of the separation ordnance.

2.	Failure of Butterfly Hinges
	The flight load on the butterfly hinges at 63 seconds after lift-off was calculated to be 44 lbs/inch. This load is created by pretensioning of the auxiliary tunnel frames prior to lift-off and by the circumferential growth of the OWS due to internal pressurization. Assuming uniform circumferential loading, the factor of safety of the butterfly hinge on the auxiliary tunnel side was approximately 5 and on the SAS-1 side was 11. Because of the friction between the MS and OWS that must be overcome in rigging the shield for flight, the ground loads on these hinges was undoubtedly higher than the flight loads. This fact, and the high factor of safety of the hinges, leads to the conclusion that they did not initiate the failure.


3.	Failure of the Trunnion Bolts or Straps
	Because of the friction between the MS and OWS, loads greater than the flight load were probably imposed m the trunnion system in torquing up the bolts for ground rigging. The calculated flight load, without friction, on the trunnion bolts and straps at the 63 second event was 44 lbs/inch (see fig. 4-4). This load was arrived at by assuming that during the boost phase there was sufficient acoustic and vibration energies to distribute the load circumferentially even in the MS. With a factor of safety of 6 in the trunnion bolts, 4 in the trunnion straps and 3. 5 in the trunnion strap rivets, the Board concluded that the trunnion system did not initiate the failure.

4.	Failure of MS due to Thrust Block Slippage
	There are twelve thrust block supports approximately equally spaced circumferentially between the aft end of the MS and the forward end of the aft OWS skirt to take the MS boost loads. The fourteen tension straps under the main tunnel are bonded to the OWS wall and attach to the MS through the butterfly hinges. They also take a portion of MS boost loads. The mating support blocks are machined at a 10 degree angle so that during a normal deployment there will be minimum separation interference. A simple test showed that this 10 degree slope could produce a radial displacement of the MS away from the OWS wall when subjected to vibratory loads. This radial movement is, however, resisted by the pretensioning of the shield by the auxiliary tunnel frames, along with the additional tension buildup due to the thrust blocks moving radially outward under tank pressurization. The bonded tension straps that also support the MS introduces a binding-friction cockedload between the OWS wall and shield when the shield is subjected to boost loads. From photographs taken of the swing links in orbit by the astronauts (see figs. 6-1 and 6-2) it is evident that none of them were bent downward to an extent which would indicate that the shield slid down over the aft skirt. Consideration of the above facts leads to the conclusion that a slippage over the thrust blocks did not initiate the failure.


5.	Buckling Failure at Support Blocks
	Two other potential failure modes exist in relation to the thrust blocks. Conceivably, a failure might result from buckling and cripling of the longitudinal panel joints due to high concentrated loads at the block support points. This could take two forms:

	a. A barrell-like mode of buckling failure where the shield deflects into a circumferentially oriented corrugated or bellow like shape and slides down the OWS wall.
	b. A crippling mode of failure where the flanges of the longitudinal panel joints deflect laterally (circumiferentially) and collapse.
	Each of these modes is hindered by friction between the meteoroid shield and the workshop wall. In addition, the butterfly hinge would force asymmetric docking of the MS, resulting in increased friction and diagonal forces which would tend to prevent buckling.

	Elementary analyses have been made by both MDAC and LRC, including conservative estimates of radial support from hoop tension and disregarding friction, the butterfly hinge restraint, and asymmetric cocking. These elementary computations reinforced the intuitive analyses that buckling at the thrust block was not the initiating failure mode.

6.	Crushing Loads on Auxiliary Tunnel Frames
	Aerodynamic crush pressures on the auxiliary tunnel tend to flatten the tunnel frame springs and thus reduce the tension of the MS around the OWS. If the crush pressure is very high, the possibility exists that the preloading of the shield could be lost and the spring frame could collapse (buckle).


	Data from analyses made at both MDAC and LRC are shown in figure 6-3 and indicate that the required loss in preload would not occur until the crush pressures are more than about 6 times the expected maximum crush pressure occurring at the aft end of the tunnel. The expected maximum crush pressure will reduce the preload at 63 seconds by approximately 10 percent.

	Although crush pressures do slightly relax the MS preload, these results indicate that buckling or complete loss of preload is not a probable failure mode.

7.	Meteoroid Shield Flutter
	The possibility of panel flutter was considered in detail in two separate analyses by MSFC and MDAC. Both concluded that brief periods-of low amplitude. flutter might have occurred, but that high stresses and structural fatigue were highly unlikely. No reason to question this finding has been found and all remainig aerodynamic discussion will center on steady or slowly varying pressures.

8.	Small Volumes of Entrapped Air Under the MS
	Although the meteoroid shield was intended to conform tightly to the OWS (except under the auxiliary tunnel) there were many small volumes of enclosed air between the MS and the OWS which had to be vented in order to prevent development of burst pressures during ascent through the atmosphere. Chief among these are the spaces outside the wardroom window and scientific airlocks. Smaller volumes were the channels formed along the longitudinal joints between shield panels, unavoidable slight wrinkles, gaps under hinges and between the panels at the pyrotechnic foldout assembly, and at the butterfly hinges. Venting was provided by many holes drilled along the panel joints as shown in figure 6-4. In addition, the inherent construction of many details of the shield provided additional venting such as through the butterfly hinges, etc. Capacity for outflow from these enclosed volumes was considered adequate.


9.	Lifting of the Forward Edge of the Meteoroid Shield
	The possibility that the forward edge of. the MS projected far enough into the slip stream to experience high pressures was examined at length. Several design features of the ordnance fold-out panel make it by far the most likely candidate among all portions of the forward edge as indicated in figures 6-5 and 6-6. In this region, the total height of the MS edge above the OWS surface is greater than elsewhere because: (1) there are three layers of 0.025 inch skin instead of only one; (2) these layers are separated by stiffeners as shown; (3) the two hinges add to the bulk; and (4) a torsion link is exerting an outward force of 18 pounds on me side of this panel. In addition to these features of the fold-out panel, it must be expected that an additional standoff of about 0.12 inch resulted during flight from the swelling of the OWS due to internal pressure. When pressurized, the tank grows about 0.12 inch less in radius at the flanges than over the middle sections, and the lightly loaded shield does not conform to these new contours. The sum of all the above features results in the forward edge of the MS extending to at least within 0.11 inch of the outer surface of the forward skirt fairing. Postflight wind tunnel measurements conducted at MSFC and MDAC indicated that the design was such that high bursting pressures could have been produced during transonic flight in this region under the MS. This situation must therefore be classified as a possible failure mode, although the actual flight data indicate that another failure mode was more probable.

10.	Auxiliary Tunnel Venting
	When the 63 second anomaly occurred, the flight dynamic pressure was near its maximum value as shown in figure 6-7 and the velocity was near Mach 1. A diagram of the general distribution of pressures predicted to exist over the Skylab spacecraft at Mach 1 is shown in figure 6-8. These pressures are deduced from wind tunnel data and represent the pressure which would occur if the vehicle were to fly steadily at Mach 1 and there were no flow fluctuations.


	The local changes in pressure produced by the auxiliary tunnel are shown schematically in figure 6-9. Dips in pressure result from expansion of the flow around the blunt base of the auxiliary tunnel forward fairing and around the base of the auxiliary tunnel.

	The design intent for the meteoroid shield and its auxiliary tunnel was that the aft end of the tunnel be sealed for "no leakage" and that the forward end be vented into the base region of the forward fairing so as to discharge air into the forward low-pressure region. The 9-to-10 square inch vent area shown in figure 6-10 was intended to provide a crushing pressure (an external pressure exceeding the internal pressure) over the entire tunnel, as shown schematically on figure 6-11.

	Post-flight investigation revealed that the aft end of the tunnel was, however, not completely sealed because of:

	a. An unexplained omission to seal or cap two hollow structural stringers on the aft skirt which extended into the aft fairing of the auxiliary tunnel (see fig. 6-12). These stringers yielded about 2.2 square inches of leakage area.

	b. An inadequate metal-to-metal fit between the aft fairing of the auxiliary tunnel and the two stringers to which it was secured resulted in approximately 2 square inches of additional leakage area. This vent area is shown in figure 6-12, and occurs at the flare of the aft fairing where the flare mates and seals with the aft end of the auidliary tunnel.

	c. An unplanned venting resulting from leakage past a molded rubber boot used to seal the movable joint and rearward facing end of the auxiliary tunnel. This boot and adjacent details are shown in figure 6-12. A metal yoke provided a positive clamp to a molded flange on the bottom of the boot over the rigid aft fairing. A bonded seal was achieved between the upper molded flange on the boot and the auxiliary tunnel.


	Because the auxiliary tunnel was required to lift freely away from the OWS and move circumferentially upon deployment in orbit into the position shown in figure 4-15, only a wiping butt seal could be achieved along the bottom edges of this boot "seal". When a differential pressure is applied to the boot, the butt seal deflects away from the OWS surface and creates two orifices of "semi-oval" cross section whose size depends upon the applied pressure differential.

	A full-scale test was performed at MDAC to determine the leak rate at the bottom edge of the rubber boot. These data indicated that a pressure dependent leakage area of 1.8 square inches would occur under the flight environment existing at 63 seconds.

	The above three sources of unplanned leakage resulted in a differential pressure distribution over the auxiliary tunnel significantly different from the design distribution shown in figure 6-11. Post-flight calculations performed at MDAC, MSFC, and LRC indicated that, for the total 6 square inches of leakage area into the aft end of the tunnel, the pressure distribution along the tunnel at Mach 1.0 would be approximately as shown in figure 6-13. These deduced pressures produce large lifting forces on much of the tunnel and part of the adjacent shield areas near the forward edge of the MS.

	With this leakage into the aft end of the tunnel, the effect is to lift the forward end of the auxiliary tunnel and the adjacent shield until a critical position is reached where high velocity ram air rushes under the shield and tears it outward from its mountings on the OWS.

	Another means of producing bursting pressures under the forward edge of the MS in the region of the auxiliary tunnel is illustrated in figure 6-14. This is an illustrative view of the wave pattern produced by the flared portion of the auxiliary tunnel forward fairing. At low supersonic flight speeds, the high and low pressure regions (the compression and expansion from the flare) extend to considerable distances away from the tunnel itself. - High pressure over the MS forward edge and lower pressures aft tend to lift the overall structure. Lifting due to this mechanism would be indistinguishable from that due to auxiliary tunnel leakage described above.


	Because lifting of the auxiliary tunnel as a result of internal pressure is a prime failure mode, both analytical and experimental post-flight studies have been performed to determine the elastic behavior of the MS in the region of the tunnel. Analytical studies were performed by MDAC using a finite element model of an area of the meteoroid shield as shown on figure 6-15. These analyses indicate that the deflection of the auxiliary tunnel and shield away from the OWS tends to become divergent. That is, as an area of the shield becomes exposed to a differential burst pressure, the shield lifts up exposing additional area, which results in further lifting.

	Experimental studies were done by MSFC using the air bladder test rig shown on figure 6-16 to generate burst pressures over the forward edge of the auxiliary tunnel area and crush pressures over the rear. The conclusion of both of these efforts is that the MS and auxiliary tunnel are quite limp and easily lifted from the OWS tank into the slip stream. A burst pressure of about 0.5 psi was found to be sufficient to effect this failure mode.

	The above postulated onset of failure is judged to be the most probable means by which the MS was lifted into the airstream.

Summary and Recommendations

The preceding analysis and discussion of possible failure modes of the meteoroid shield have identified at least two ways that it could fail in flight. Although the most probable cause of the present failure was the lifting of the shield from the OWS tank by excessive pressures in the auxiliary tunnel, other failure modes could have occurred in other regions of flight or under more severe flight environments than were encountered by Skylab 1. Among these other modes of potential failure, which could combine in various ways under varying conditions of flight, are excessive pressures under the forward edge of the shield, or inadequate venting of the folded ordnance panel. The inherently light spring force of the auxiliary tunnel frames, the crushing loads on these frames in flight, the inherent longitudinal flexibility of the shield assembly, the forces applied by the swing links to deploy the shield, the possible "breathing" of the shield panels as cavities are vented, the non-cylindrical nature of the underlying pressurized tank, and the uncertain tension loads applied to the shield in rigging for flight all contribute to a lack of rigidity of the shield and a weakness of its structural integrity with the underlying tank structure.

A simple and straightforward solution to these inherent problems of the present shield design is therefore not likely. A fundamentally different design concept seems in order. One solution is, of course, to simply omit the meteoroid shield, suitably coat the OWS for thermal control and accept the meteoroid protection afforded by the OWS tank walls. Although the Board has not conducted an analysis, meteoroid flux levels are now know to be considerably lower than those used in the original calculations. A new analysis, based on these flux levels, may show acceptable protection.

Should some additional meteoroid protection be required, the Board is attracted to the concept of a fixed, non-deployable shield. Although the inherent weight advantages of a separable bumper are not available in this approach, the mission of Skylab could probably be satisfied in this manner. One concept would be to bond an additional layer of metal skin to the surface of the tank with a layer of non-venting foam between the OWS tank and the external skin. The problem being statistical in nature, the entire shell of the OWS would not have to be covered.

Figure 6-1. - View of Kapton surface of the OWS showing forward torsion rod swing link.