|
|
|
When I was a young 2685 (USAF code for mathematician) working as a ballistics engineer we got hold of some of the first TI-59 calculators with the nifty little magnetic card readers for inputting and storing programs. I had been playing with ballistic missile codes (both for live missiles and for a program called STRIKE, Strategic Targeting Resource Integrator and Kill Evaluator) and suggested to the powers that be that it might be possible to get a pretty good fuel usage estimate for use in retargeting the Minuteman III (3 warhead system). Given the go-ahead, I pressed on.
The first problem was the development and testing cycle times. I was used to the IBM 360 turnaround times for punch decks of 2-4 hours, often overnight because of the classification of the programs (Secret, Top Secret, SCI). The TI-59 was slow, so I developed a sort of emulator on the HP9825 desktop calculator (see www.hpmuseum.org) so I could get faster turnarounds. In those days computers belonged to "Data Automation" (AD), so we only got the HP9825 because the case said "Calculator". The system had a tape drive, card reader, plotter, printer, full keyboard, 32-character display and 4K of memory with its own programming language, but hey, the box said "Calculator" so, by the book, it wasn't a computer. By emulating the TI59 on the HP9825 I was able to do real-time run/debug/correct style programming that was really fun compared with the batch style programming we were using on the IBM 360.
By using some really tight code I was able to simulate the flight and deployment of the booster and the three reentry vehicles, computing the cross range and downrange perturbations and the resulting fuel costs, and estimate the total fuel used to deliver the RVs to the three targets as input by the user. This all had to fit in the 960-step/100 memory register of the TI59. Considering that amongst other things I had to invert a 3*3 matrix, space was pretty tight.
Using data from the actual targeting programs used in the Minuteman III, I built some curve fitting functions for fuel usage, used simple spherical trig range and azimuth calculations and managed to get within 5% of the actual fuel usage (the program's estimate was always close to the projection from the mainframe, and was usually in the .95 to 1.05 range about the true value).
The final program was to be carried on the ABCP (KC-135 "Looking Glass", www.theaviationzone.com) to permit quick evaluation of proposed retargeting of MMIII missiles.
