Supernova Stadyshell Shutter
Nearby, damaging supernovae will occur only a few times per gigayear, but without some adaptive response, the stadyshell might be irrevocably damaged.
Presume that as the stadyshell is constructed, our knowledge of the behavior of supernovae is improved as well. Supernovae occur about twice per century in our galaxy - and that we want to study candidates to high resolution to predict the precise time they explode. Presume we need to image a candidate supernova to "0.01 sunspot" pixels - a resolution of 20 km. Supernovae in our galaxy will be closer than 40,000 parsecs - 8 billion AU. Presume we need to image them to infrared wavelengths of 2.5 μm. The aperture size for an imager (sparse, with nulls computed and added to blank out other stars) would be 1 AU diameter - mindbogglingly large for a telescope, but 25 ppm of a completed stadyshell. The knowledge gained will help predict and accomodate supernovae.
Supernovae ramp exponentially to maximum luminosity in 7 to 20 earth days, and endure for up to 200 days. The full blast from a nearby supernova might melt or damage a stadyshell thinsat perpendicular to the energy flux; however, turned edgewise to the distant supernovae, only one thin edge is exposed.
Assume a meter-square thinsat, with "armored" hinges that allow it to fold in half. Instead of exposing the whole meter-squared thinsat to the supernova, the thinsat turns, then folds behind the hinge. The folding process conserves angle, but we can use light pressure from the sun to modify the resulting angle. Thinsats directly online between the sun and the supernovae must "turn" the most, but have the most solar power (500 mW) and light sail thrust (1.6 nN) to work with. A turn and stop can be accomplished in less than 1 earth day.
So, the most affected thinsats will be edge-on to the sun and supernova, and lose most light-pressure maneuvering (some light can be reflected at them from distant parts of the shell), and "plummet" towards the sun for 200 days. The gravitational acceleration at 50 AU is 2.4 μm/s² - in 200 days (17 megaseconds), they have fallen 3.6 million kilometers, 0.024 AU, 500 ppm of total distance, and are descending at 40 m/s . They turn back flat to the sun, "spread their wings", and change to nearly full reflectivity, doubling the light pressure. After another 200 days, they have stopped at 7.2 million kilometers down, and accelerate back upwards for another 200 days. Finally, they turn edgewise to let solar gravity slow them to a stop at their original position. Downtime, about 2 years, but better than destruction. With perhaps 5 such events per gigayear, that is 99.999999 % average availability.
Alternatively, they can go only partly reflective, and partly edgewise, for their descent and ascent - perhaps 10x deeper, and 10x longer to return to the original position, but 90% of full power during this 20 year process. THe biosphere on the earth below will need much longer to recover from a major supernova.
If the stadyshell can predict the time and magnitude of an oncoming supernovae with 10% accuracy, it can prepare for it by thrusting the effected thinsats outwards, spreading the the accomodation time further while reducing the geometric perturbation of the shell.