Longshot City

This is me trying to find a way around (some of) the complexity that doomed the Napier Nomad of 1947.  The Nomad, now a museum piece, is still the most fuel-efficient aircraft engine ever built (in terms of simple mechanical output per fuel input), but it didn’t work very well and came out right before big dumb lightweight turbojets and cheap post-war oil:

(below is a video, not a picture)

Like ye ole Nomad, it’s basically a gas turbine with a very-compressed diesel engine for a combustor.  Since a diesel can handle significantly higher instantaneous maximum temperatures than a turbine can, better fuel efficiency results.

Plus, the gas turbine also acts as a bitchin’-powerful turbocharger for the diesel and keeps its size=weight down when compared to a regular diesel running by itself.

And finally, it’s apparently a better deal to let both the diesel and turbine produce power, as opposed to a diesel with a simple turbocharger that only drives itself.

But, instead of using scary gears to join the gas turbine and diesel together onto a single output shaft, the Parallel Compound Cycle engine cheats.  No gears.  Each side, the diesel and turbine, drives its own propulsion fan.

So the diesel and turbine sides are connected only by tubes (compressed air and exhaust to/from the diesel), not gearing.  They’re both propelling the same airplane, so their power outputs are combined that way.

Further, since there are two “actuator disks” per engine (one for the diesel’s fan, and another for the turbine’s) instead of one, propulsive efficiency (thrust per horsepower) is upped a little bit to boot.

This was probably written about and shot down decades ago but dangit, I just haven’t found anything about it.  Just in case this is a new idea, I offer it up here.  Thank you.

This one hasn’t let me go for months.  The idea is to zealously over-engineer the few parts of a computer where almost all failures happen: the power suppy, disks and thermal management components.

Not “hot-swapping” bad parts, but rather just building in the spares it’ll need down the line and forgetting about it.

The point is being able to deploy a bit of software to a site and not be constantly biting your nails about it failing because the computer broke.   There are $2k computers out there controlling $2M machines with a some-hundred a month service contract, which just doesn’t make a lot of sense.  I’m interested in whether there’s a niche out there for $3k computers re-packaged into $15k bunkers that do the exact same thing, but for much longer, unattended.

There’s “cost-effective” and then there’s “trust-effective.”

(A point I forgot to make is that this thing has a slow-clock low-power supervisor computer, with its own mil-spec Vicor power supply and little UPS, that handles the temperature sensing, fan control and power supply switching.  That’s custom, and is invisible to the big “other” computer.)