ArticlePDF Available

Harvesting and Managing Energy While Re-entering an Atmosphere Using a Shuttlecock Design

Authors:

Abstract

Abstract / Pre-Summary This sub-page to www.rocketslinger.com is a companion or supplementary document to another document found at https://www.researchgate.net/publication/354612024_Methods_of_Decelerating_a_Spacecraft_Through_Atmospheric_Re-entry_Using_a_Shuttlecock-like_Design and at http://www.rocketslinger.com/BadMinton/ (same document, 2 locations). In BOMs (Bills Of Materials) lingo, the “parent” BOM calls a “child” BOM, where the child is a sub-assembly of the parent. This document deals with sub-assemblies of the designs described in the “parent document”, and so, it will henceforward be referred to as the parent document for brevity. Temperatures endured during re-entry (both on the windward and lee sides) are briefly examined here, as are temperature tolerances of materials and subassemblies. Then a “heat wall” design is described, which can be used to protect subassemblies from heat damage. Much of this is focused on hydraulic machinery, which will need to be used to control the “petals” (or grid fins) as were described in the parent document. Many variations of a powered hinge are described here. One of the best versions consists of a curved (arc-shaped) friction plate attached to each petal-becoming-a-grid-fin (AKA, fairing segment). The friction plate is gripped between two powered, slowly rolling funnel shapes, to create a torque-limiting device, to absorb sharp peak loads without damage. This hinge system is called a “Bifrost Hinge”, and the name given to it, is explained. To harvest energy (or power used to perform work) in a spacecraft entering an atmosphere, we could use roller chains similar to (but larger and stronger than) the chains used in bicycles and chain saws. In a chain saw, some of the plates (in the chains) are extended outwards with teeth, for cutting wood or other materials. Here, instead, some of the plates are extended outwards to “catch the breeze” during re-entry, the same as windmill vanes catch energy from the wind. The overall device, then, here, will be called a chain windmill. The vanes in such a device could possibly take many forms, but two are described here. One is a single plate per each set of chain-plates that is so equipped, which is stopped (by a hard-stop) from folding past 90 degrees, with the right angle formed to “catch wind”, and not fold further towards the lee side. However, this vane is allowed (during the chain segment’s return journey towards the windward side) to fold out of the way of the prevailing wind. The alternate (here-described) vane is a “butterfly vane set”, with twin segments that “flap” to catch wind in one direction, and fold out of the way in another direction. They can each perform a partial rotation around pins protruding outwards from each pair of roller-chain plates that is so equipped. Each of the two twin vanes (“butterfly wings”) provides the hard-stop for the other, in this design. In either form, a chain windmill is not at all an optimal solution here, for us. An energy-harvesting device could also resemble a paddle wheel on a riverboat. This approach will be far less disadvantaged by lack of lubrication than a chain windmill would be. And as will be shown, lubrication WILL be a major problem for the chain windmill (unlike the paddle wheel). Also described here is an airflow spoiler that could be placed to the windward side of the energy-harvesting device, possibly sized and located in a compromise between partially protecting the energy-harvesting device (best choice being a paddle wheel) from too much heat and strong airflow, and not getting enough airflow to harvest enough power. The spoiler is composed of a hinged heatshield-covered plate that can be thrust out into (or retracted from) the airstream. It could be activated by a (heat-walled) “Bifrost hinge”, which could also double up as a “hydraulic battery” used to store extra power. These are sometimes called “hydropneumatic accumulators”. In the design context here, the variably pressurized nitrogen in such a device could be partially or entirely replaced by the variable pressure of passing air in the airflow spoiler (which pushes on the plate). Such a design choice may have to be balanced with the other possible purpose of the spoiler, which is to partially protect the paddle wheel. The body of this document contain far more details, and sometimes-implausible variations. However, the above covers the most important basics.
From (by) RocketSlinger@SBCGlobal.net (email me there please)… This is a
sub-site to main site at www.rocketslinger.com
This web page last updated 03 Oct 2021
Harvesting and Managing Energy While Re-entering an
Atmosphere Using a Shuttlecock Design
Abstract / Pre-Summary
This sub-page to www.rocketslinger.com is a companion or
supplementary document to another document found at
https://www.researchgate.net/publication/354612024_Methods_of_Decelerating_
a_Spacecraft_Through_Atmospheric_Re-entry_Using_a_Shuttlecock-
like_Design and at http://www.rocketslinger.com/BadMinton/ (same document, 2
locations). In BOMs (Bills Of Materials) lingo, the “parent” BOM calls a “child”
BOM, where the child is a sub-assembly of the parent. This document deals with
sub-assemblies of the designs described in the “parent document”, and so, it will
henceforward be referred to as the parent document for brevity.
Temperatures endured during re-entry (both on the windward and lee
sides) are briefly examined here, as are temperature tolerances of materials and
subassemblies. Then a “heat wall” design is described, which can be used to
protect subassemblies from heat damage. Much of this is focused on hydraulic
machinery, which will need to be used to control the “petals” (or grid fins) as were
described in the parent document.
Many variations of a powered hinge are described here. One of the best
versions consists of a curved (arc-shaped) friction plate attached to each petal-
becoming-a-grid-fin (AKA, fairing segment). The friction plate is gripped between
two powered, slowly rolling funnel shapes, to create a torque-limiting device, to
absorb sharp peak loads without damage. This hinge system is called a “Bifrost
Hinge”, and the name given to it, is explained.
To harvest energy (or power used to perform work) in a spacecraft
entering an atmosphere, we could use roller chains similar to (but larger and
stronger than) the chains used in bicycles and chain saws. In a chain saw, some
of the plates (in the chains) are extended outwards with teeth, for cutting wood or
other materials. Here, instead, some of the plates are extended outwards to
“catch the breeze” during re-entry, the same as windmill vanes catch energy from
the wind. The overall device, then, here, will be called a chain windmill.
The vanes in such a device could possibly take many forms, but two are
described here. One is a single plate per each set of chain-plates that is so
equipped, which is stopped (by a hard-stop) from folding past 90 degrees, with
the right angle formed to “catch wind”, and not fold further towards the lee side.
However, this vane is allowed (during the chain segment’s return journey towards
the windward side) to fold out of the way of the prevailing wind.
The alternate (here-described) vane is a “butterfly vane set”, with twin
segments that “flap” to catch wind in one direction, and fold out of the way in
another direction. They can each perform a partial rotation around pins
protruding outwards from each pair of roller-chain plates that is so equipped.
Each of the two twin vanes (“butterfly wings”) provides the hard-stop for the
other, in this design. In either form, a chain windmill is not at all an optimal
solution here, for us.
An energy-harvesting device could also resemble a paddle wheel on a
riverboat. This approach will be far less disadvantaged by lack of lubrication than
a chain windmill would be. And as will be shown, lubrication WILL be a major
problem for the chain windmill (unlike the paddle wheel).
Also described here is an airflow spoiler that could be placed to the
windward side of the energy-harvesting device, possibly sized and located in a
compromise between partially protecting the energy-harvesting device (best
choice being a paddle wheel) from too much heat and strong airflow, and not
getting enough airflow to harvest enough power. The spoiler is composed of a
hinged heatshield-covered plate that can be thrust out into (or retracted from) the
airstream. It could be activated by a (heat-walled) “Bifrost hinge”, which could
also double up as a “hydraulic battery” used to store extra power. These are
sometimes called “hydropneumatic accumulators”. In the design context here,
the variably pressurized nitrogen in such a device could be partially or entirely
replaced by the variable pressure of passing air in the airflow spoiler (which
pushes on the plate). Such a design choice may have to be balanced with the
other possible purpose of the spoiler, which is to partially protect the paddle
wheel.
The body of this document contain far more details, and sometimes-
implausible variations. However, the above covers the most important basics.
As with other sub-pages of www.rocketslinger.com, the intent here is to
“defensively publish” propulsion-related (and “misc.”) ideas, to make them
available to everyone “for free” (sometimes called “throwing it into the public
domain”), and to prevent “patent trolling” of (mostly) simple, basic ideas.
Accordingly, currently-highly-implausible design ideas (usually marked as such)
are sometimes included, just in case they ever become plausible, through radical
new technology developments (often in materials sciences).
Introduction / Basics
Please read the abstract above… Some of those basics may not be
thoroughly (completely) repeated here below.
Dear Reader, excuse me as I will often slip out of stilted formal modes of
writing here. I have no boss or bosses to please with these writings, so I’ll do it
my way! I’ll often use a more informal style from here on in, using “I”, “we”, “you”,
etc. “We” is you and me. “You” are an engineer, manager, or other party
interested in what’s described here.
The parent document describes what I now briefly summarize: Payload
fairings can be built with embedded grid fins buried inside other materials
(plastics, etc.) that will burn, ablate, or fall away in the heat of re-entry. The
fairings segments (4 of them being a good number here) can also be arranged
on hinges that join them to the base of the payload section. The only sensible
option here is for those hinges to reside outside of the circular profile of the
cylindrical rocket body, and of the base of the fairing. This, then, I call a “partial
hammerhead design”, where, in fairings lingo, the “hammerhead” design refers to
the girth of the fairings (at the base of the payload cavity) exceeding the girth of
the rocket body.
The hinged fairings splay out from their base like the petals of a flower
open up, so here they are called “petals”. The payload is deployed, and the de-
orbit burn is performed. The petals stay deployed or partially deployed till the
plastic or other “sacrificial materials” ablate or fall away; now the petals become
“grid fins”. Via “powered hinges”, the grid fins (which were formerly petals) can
now be adjusted more-open or more-closed. When this is done in a rolling-
around-the-circle pattern, we can induce a heat-spreading “rotisserie roll” in the
body of the rocket, as we descend.
No details were given (in the parent document) about how the “powered
hinges” should be designed and built. Here, I will present my moderately-
educated ideas about how to do so. OK, more honestly, here come my best
SWAGs at doing this, and associated things! Further below, of course.
The parent document also briefly describes what were there called screw-
propellers” or “screw turbines” as energy-harvesting devices. There are a few
mentions here of associated ideas. But, presumably at the base of these screw
turbines, we would want to add hydraulic pumps (far more likely) or electrical
generators (far less likely) to harvest energy. Either way, the pumps or
generators would have to be shielded from ambient heat during re-entry. Heat-
shielding will be addressed in some detail, further below. Heat-shielding
discussions are new here, compared to the parent document. Anyway, the
screw-propellers or screw-turbines are dismissed here as not being very suitable
for our use, especially as compared to a paddle-wheel design.
I’ll describe “chain windmills” here (see a brief summary in the abstract
above), as an alternative to “screw turbines”. Their best energy-harvesting
modes would also be to drive hydraulic pumps. The other choice is electrical
generators. Why do we chose hydraulic pumps? See
https://sciencing.com/differences-hydraulic-motors-electric-motors-7351549.html
Differences Between Hydraulic Motors & Electric Motorsand
http://info.texasfinaldrive.com/shop-talk-blog/hydraulic-motors-vs-electrical-
motors-why-hydraulic-wins Hydraulic Motors vs Electrical Motors:
Why Hydraulic Wins”.
Motors v/s generators considerations are VERY highly similar here! Also,
the more times we convert one form of energy to another (oil pressure to
electricity or vice versa), the more energy we lose to inefficiencies. So we’ll stay
mostly hydraulic here, per my preferences, in these design notes.
Airflow spoilers and “hydraulic batteries” (“hydropneumatic accumulators”)
are also mentioned in the abstract. They are also discussed in much more detail
below. Let’s move on!
Internet Research on Relevant Facts About Re-Entry Heat and
High-Temperature Materials and Subassemblies
Here, we assume re-entry is from low Earth orbit. Any higher-speed re-
entry seems to be out of reach with current materials and technologies, for any
kind of “shuttlecock” re-entry style, as described here.
The below is some fairly cursory internet research on associated matters,
to be used to guide our design work here. My general intent here is to conduct
“amateur public-domain rocket science” (AKA “open source”), so, as usual, your
comments questions, and contributions are welcome at
RocketSlinger@SBCGlobal.net. So, in accordance with my intent, research
matters (links) that are NOT used here (but are highly relevant, and could be
used by others) are included here. We generally start with the most-relevant,
and conclude with the least-relevant.
Temperature Tolerances Needed… Assuming re-entry from low=Earth
orbit, of course. https://en.wikipedia.org/wiki/Atmospheric_entry is of general
interest. https://www.airspacemag.com/how-things-work/shuttle-tiles-12580671/
Shuttle Tilesis subtitled “Why the space shuttle can withstand reentry
temperatures up to 2,300 degrees.”
https://www.nasa.gov/centers/ames/research/2007/faq-shuttleglass.html says
(about the lee as opposed to windward side)…”This side experiences lower high
heat transfer compared with the windward side, and so it does not reach as high
a temperature. This is because the pressure is much lower, at least two orders of
magnitude lower (1/100 or less pressure) on the back side. The hot gas on the
windward side expands to the leeward side, which means the pressure drops
quickly, and so does the gas temperature. (Deletions by me here). So windows
must still be able to withstand high temperatures, say about 1000 C. So Shuttle
windows are made from a high-temperature quartz glass that can withstand
heating and cooling without cracking.”
https://www.daviddarling.info/encyclopedia/R/reentry_thermal_protection.h
tml says almost the exact same thing as above.
Note that 1,000 C = 1,832 F. There’s a handy C to F converter at
https://www.google.com/search?q=Centigrade+convert+to+fahrenheit&oq=Centi
grade+convert+to+fahrenheit&aqs=chrome.0.69i59.225990j0j4&sourceid=chrom
e&ie=UTF-8. Also (importantly!) note that the “lee” side is about 100 times lower
in pressure than the “windward” side.
https://www.nasa.gov/mission_pages/constellation/orion/orionheatshield.html
says The shuttle enters the atmosphere at lesser speeds, 4.7 miles (7.5
kilometers) per second, generating a lower maximum temperature of 2,900
degrees Fahrenheit”.
So we can conclude that our temperature-tolerance targets are about
2,300 F to 3,000 F or so for the windward side then, and 1,832 F for the lee side.
These are our temperature-tolerance targets. Also importantly, note that
windward side pressures can be about 100 times as much as lee side pressures.
Metals to be used? General knowledge (trivial to look up) is that the
melting point of titanium 3,034°F. Alloys of stainless steel also work.
https://www.businessinsider.in/bird-poop-and-dust-could-seriously-complicate-
elon-musk-and-spacexs-latest-plan-to-reach-mars/articleshow/68085712.cms is
relevant… From there…
Musk has said the nose of Starship may be exposed to temperatures of
about 2,700 degrees Fahrenheit. The type of steel alloy SpaceX may use on
Starship's outer skin, called 310S, melts at about 2,400 degrees.” (My comment;
so they need heat shielding there at the nose).
In passing, let me add, ceramics are of interest as well. They’re generally
good in “compression” modes, but not in “tension” modes. Bending and torsional
stress modes are therefore dicey for ceramics! I still think metals are better for
most of what high-temperature and high-stress structural uses are described
here. But you might want to look into the materials used here:
https://www.imperial.ac.uk/news/176628/new-record-worlds-most-heat-resistant/
“New record set for world's most heat resistant material”, which says that
Tantalum carbide (TaC) and hafnium carbide (HfC) are refractory ceramics…”,
and a ceramic made of them is rated at up to nearly 4,000 C, or 7,232 F!
We want to use hydraulics a LOT here! An obvious place to start is the
hydraulic fluid itself. The following web site discusses new high-temperature
TARGETS for the fluids, so this is best-case!
https://onlinelibrary.wiley.com/doi/abs/10.1002/jsl.3000090203 The rationale is
presented for the development of 315 °C bulk oil high-temperature hydraulic
fluids…” 315 C = 599 F, so we must conclude that any hydraulic gear (to include
pipes or hoses) MUST be heat-shielded!
I am NOT (here, much) recommending the use of hydraulic cylinders, but
they’re certainly an option, if properly heat-shielded. Keep in mind that, if you
use these, the fluid isn’t the only heat concern.
https://www.hydraulicspneumatics.com/technologies/cylinders-
actuators/article/21885384/hydraulic-cylinders-keep-their-cool says…
“…premature seal degradation can usually be avoided by specifying a seal
material that (is) designed to operate in higher temperatures. For example,
ethylene propylene and Viton seals usually accommodate temperatures
exceeding 350° F.”
Lubrication is a concern! How about high-temperature lubes?
https://www.machinerylubrication.com/Read/30674/high-temperature-lubricants
says 220 C = 428 F is perhaps the best that we can do! If we can heat-shield a
“well” of oil or grease (lube), and have our otherwise-exposed moving parts dip
into there now and then, we are doing extremely well! Enclosing the lubricated
bearings in heat-shielded areas is even better. In the worst cases, we’ll have to
make do with un-lubricated parts.
Mechanical or structural cables aren’t used in the here-described designs,
but heat-resistant cables are strong candidates for use in design variations here.
https://lapplimited.lappgroup.com/products/high-temperature-cable.html provides
cables tolerating up to 1,565 C = 2,849 F, and the lee side max is about 1,832 F,
so we could use these, very cautiously, if we wanted to.
Now we transition to parts that we can’t or don’t use, that are of less
interest. They’re included here (more or less) on an “FYI” basis. High-
temperature belts (think V-belts or similar)
https://www.durabelt.com/hightempresistantbelts.php says Works up to 230oF
(110oC). Belts are out! High-temperature hoses
https://www.ducting.com/high-temperature-hoses says… “There is one specific
variation of the stainless steel hose that uses a titanium alloy. It can be used with
temperatures up to 1650° F…” Hoses are HIGHLY questionable, would have to
be used VERY cautiously!
Between belts and hoses, we can make two very important conclusions:
‘1) ANY use of flexible or pliable materials for heat-exposed covers or shields, is
a VERY bad idea! Hydraulic parts (with support hoses), then, can’t move around
through heat-exposed areas. In a heat-exposed area, their positions MUST
remain fixed, and heat-shielded!
‘2) Any ideas about using “heat walls” that use a cooling fluid (like water or liquid
nitrogen), that therefor need cooling-fluid-supplying flexible hoses, as the heat-
wall-protected part moves through heated areas, is off limits as well. Any such-
cooled devices will need to remain in fixed positions as well. They will need to be
supplied with cooling fluids by insulated high-temperature (inflexible) metal pipes
or ducts.
High-temperature cloth or fabric used for insulating hoses (or other
devices) is off limits as well. Fiberglass can withstand 1,000 F, and we need
1,832 F or higher. Asbestos is ALMOST good enough, at a melting point of
1,600 F (or so), but is also hazardous as well.
If you’re really set on trying to use moving, flexible hoses, and insulating
them with fabrics, then see https://nutec.com/products/ceramic-fiber-blanket/
MAXWOOL CERAMIC FIBER BLANKET”, rated 2,600°F (1425°C). However,
notice that the recommended uses here are as “blankets”, not wrappings on
hoses, and so, this fabric might not be suited for uses involving being flexed a lot.
And as you can see, ceramics can sometimes be turned into flexible fibers. You
might, then, look into the materials used here:
https://www.imperial.ac.uk/news/176628/new-record-worlds-most-heat-resistant/
“New record set for world's most heat resistant material”, which says that
Tantalum carbide (TaC) and hafnium carbide (HfC) are refractory ceramics,
meaning they are extraordinarily resistant to heat. Their ability to withstand
extremely harsh environments means that refractory ceramics could be used in
thermal protection systems on high-speed vehicles…”, and is rated at up to
nearly 4,000 C, or 7,232 F!
As previously mention, we’re primarily interested in hydraulics here.
However, electrical generators and motors are still of some interest. Here,
https://netl.doe.gov/node/3754 says (in an oil-well context) “None of the
commercially available generator systems developed by major service
companies are capable of operation at 250°C. Their target (at this web site) is
250 C or 482 F. Electrical motors self-heat and so they would be even worse.
We will have to robustly cool ALL parts of power management here, in our design
context!
Not used here at all, but of general interest, are thermoelectric device
and Peltier “heat pump” devices. For thermoelectric devices,
https://www.energy.gov/sites/prod/files/2014/03/f13/acep_04_elsner.pdf says
Large increases in thermoelectric conversion efficiency (>3 times) appear
feasible from T H’s of 150°C to 1000°C.”… TH meaning temp-hot and 1832 F =
1000 C… This is perhaps the upper limit of what may happen some sunny day…
Which is the same as our lee side… Probably, this is not of much practical
interest for now.
Now on a Peltier device, we’d want the cold side in our cooled-down-
devices boxes, and the hot side out in the ambient hot air… Ambient is best-
case 1832 F; can we beat that? A high temperature Peltier device here:
https://tetech.com/peltier-thermoelectric-cooler-modules/high-temperature/ says
Very High Temperature (VT) modules are rated for intermittent use to
temperatures up to 200°C whereas all other modules we sell are rated for use up
to 80°C.” 200°C = 392 F, and so we write this off, except possibly for uses
entirely within the avionics circuits, which is outside of our scope here.
On a practical (non-academic, non-developmental) level, Digikey is an
excellent place to see what is available “off the shelf”. For Peltier devices, there,
https://www.digikey.com/en/products/filter/thermal-thermoelectric-peltier-
modules/222 lists maximum Th (Temperature-Hot) of 50 C, which is even
worse than above… For practical “off the shelf” coolers that is.
Moving on to some other assorted matters, “hydraulic batteries” are of
much interest here. See
https://www.hydraulicspneumatics.com/technologies/accumulators/article/218829
59/hydraulic-batteries-save-fuel “Hydraulic ‘Batteries’ Save Fuel”. From there,
Hydropneumatic accumulators are widely used in hydraulic systems because
they provide auxiliary power during peak periods. This lets designers select
smaller pumps, motors, and reservoirs in the main system.” We’ll refer to this
again.
Of general interest are the following:
https://scholarworks.iupui.edu/bitstream/handle/1805/7396/Vaezi_2014_energy.p
df;jsessionid=B7BC2AB03D619A47177D51FF43D852BF?sequence=1 is
“Energy Storage Techniques for Hydraulic Wind Power Systems”. Also see
https://www.sandia.gov/ess-ssl/EESAT/2005_papers/Lemofouet.pdf Principle of
Hybrid Energy Storage Systems Based on Hydro-Pneumatics and
Supercapacitors for Distributed Generation and Renewable Energy Sources
Support”. Then https://www.mobilehydraulictips.com/energy-efficient-hydro-
accumulators-for-energy-storage-or-conversion/ is a commercial. At my casual
glance, it seems to involve mostly dampeners to dampen out hydraulic pressure
oscillations.
Let me briefly add that hydraulic accumulators need not be heavy and
thick-walled, which we wouldn’t want for aerospace applications. The can have
COPV (Composite Overwrapped Pressure Vessel)-style lightweight yet strong
walls. For samples of providers of aerospace-type accumulators,
https://ph.parker.com/us/en/aerospace-accumulators and
https://www.valcor.com/missiles-and-aerospace/aerospace-accumulators/.
Also of general interest for actuators, see
https://en.wikipedia.org/wiki/Worm_drive for possible use (not used here). A
linear actuator may also be of interest: See
https://en.wikipedia.org/wiki/Linear_actuator for actuators. A sub-type being a
“screw actuator” is what we may sometimes want, along with a hydraulic motor.
OK, now I think that it’s high time to move on!
Building a “Heat Wall” (or Ice-Box) for Subassemblies
High-temperature shielding tiles are well enough known from the Space
Shuttle days, and now from Space X’s “Starship”. These are easily enough
researched, and I have no special comments to add here, about those. Under
the tiles, one can locate a blanket. Search here in this document for “Maxwool”
search-string for a good candidate material. One or both of these could be on
the outer layers of a “heat wall” for components that need to be kept cool. This is
also true for fluid-carrying pipes (don’t forget, fixed locations, not moving hoses!)
that carry hydraulic fluids, or water, or liquid or gaseous nitrogen, for cooling.
Liquid nitrogen (or helium or other inert coolant) may gasify during its journey, but
that will be OK, as long as it is kept as cold as possible.
Now please see a document that is kept at duplicated locations at
https://www.researchgate.net/publication/331556573_Designs_for_Passively_Th
ermally_Gated_Fluid_Flow_Switches and at
http://www.rocketslinger.com/Psv_Tgt_Fsw/, titled Designs for Passively,
Thermally Gated Fluid Flow Switches”. This was written when Space X was still
considering the use of “transpiration cooling” for “Starship”. So it is written
assuming that the cooling gas (liquid nitrogen being an excellent choice) would
be kept in a “sandwich layer” between two metal outer rocket-walls, and that tiny
simple weep holes and-or passively thermally gated switches would allow cooling
fluids to flow OUTWARDS, through the outer skin of the rocket body. Here, for
this application, we could let the cooling fluids weep INWARDS, into the “ice
boxes” that keep our most sensitive gear (some of which self-heat when used)
cold. Passive, simple, tiny weep holes, thermally gated passive switches, and
active electrically gated, temperature-sensors-equipped fluid switches could be
used, any one alone, or in any desired mix. In the below drawing, let’s just label
them all as “pores”. The coolant flows to the “ice-box”, and is injected (or flows)
inwards as is needed. At some “quiet spot” where there’s not much external
airflow (ambient air flows undesirably, excessively pulling out our cooling gasses
via the “chimney effect”), we place a small outlet hole to allow cooling-gasses
outflows to ambient pressure, preventing our “ice-box” from becoming a balloon,
and blowing up!
See the below drawing of an “ice-box”, and keep in mind that this is the
“deluxe” version. Not all ice-boxes will need all of the layers shown.
Figure #1
A good-quality “ice-box” will be needed for all devices that are located in
hot ambient air, in the (during ascent, fairings-enclosed) payload bay, during
descent (after the “flower petals open” and the payload is deployed, and the de-
orbit burn is performed). For many devices, we will have no sensible choices,
other than to deploy them there, in the otherwise-now-emptied payload bay.
However, let’s assume we’ll want to keep things modular. The payload
bay will often contain a satellite (more often satellites plural) that need clean-
room conditions, unlike the rocket itself. The payload bay will be positively
pressurized with clean air, on the way up. This isn’t true of the inside of the
rocket body, where ambient air pressure (outside of fuel and oxidizer tanks of
course) is acceptable. So there will be a payload bay “floor” (that needs to be
airtight) and a rocket-body “roof” (not airtight), both of which will need to be
bolted (or otherwise fastened) together. We could call this a “deck” of the
payload bay.
We can leave voids (holes) or indentations (pockets) in the roof of the
rocket body. Now we can put SOME of our heat-sensitive components “below
the deck” instead of “above the deck”. Below deck, using the shelter of a colder,
LARGE rocket body (a large heat sink), we can use less-bulky and less-
expensive “heat walls” to shelter some of our heat-sensitive components. These
might include avionics, batteries, hydraulic-fluid tanks, hydraulic accumulators,
liquid nitrogen tanks, and (perhaps) a hydraulic motor driving an electrical
generator, to keep our electrical batteries topped off.
To keep things modular and not too messy, we’ll want to keep the
connections and disconnections between the rocket body and the payload
compartment minimized. Since many components MUST be above deck, that
means we’ll want to cluster most of the associated parts together, associated
with the payload bay. On the other hand, that makes the payload bay (which
must move from a clean room where the payload is assembled) heavier (harder
to move). For that reason, heavy (and dirty?) items might want to stay with the
rocket body. Some balance (of conflicting objectives) must be struck here.
Fleshing Out the Details of the “Powered Hinge”
The parent document mentions a “powered hinge”, but never describes it
in detail. The time is now to describe a good design for such a device, in this
application. The “powered hinge” opens the “petals” of the bud (closed payload
bay) to become petals instead of fairings-segments. Petals later become “grid
fins” as sacrificial materials fall, burn, or ablate away.
Many of us are familiar with “gas shocks” for partially lifting or counter-
balancing a (vertically opening) heavy door, or for softening the opening and
closing of many kinds of doors. A “gas shock” (pneumatic cylinder) here is highly
similar to a hydraulic cylinder. Envision it spanning from the payload bay “floor”
(deck) towards a central strong spar on the “petal”, in a right triangle (in a petal-
closed condition), with a triangle side on the petal, and a triangle side on the
“deck”. The hypotenuse is the hydraulic cylinder. At way-wide-opened positions,
the door-opening force exerted is WAY sub-optimal! Far worse, the hydraulic
cylinder would sweep through different positions for us, and we can’t use hoses
(hydraulic or otherwise), or flexible “heat walls”, here. I won’t bother to provide a
special drawing here (unless asked, at RocketSlinger@SBCGlobal.net), so here
are some illustrations of such a device: Amazon Lifter.
So a good solution that I could devise, is to add (rigidly, with no
deliberately added ability to flex) an almost-1/4th of a circle (an arc, like a pie
crust on a pie slice) of a strong heat-resistant metal, firmly bonded to the central,
strong structural spar on a petal. The arc can be called a “pusher arc”. It will
need to VERY strongly push on the petals, during descent, against prevailing
winds. If the “pusher arc” is ever “asked to” pull, instead, it will be only very
weakly pulling inwards. The “pusher arc” will be cogged on both sides, phased
cog-to-cog at 180 degrees from one side to another, so that hydraulic-motor-
driven cog wheels (qty 2) can grip it firmly at all times. From a side profile, it will
look like this:
H
C
C
C
Petal is Powered
Open and Shut = Fixed-Location
Powered Cogwheel
Heat Walls
Pipes &
Wires
Pusher
Arc
Petal /
Grid Fin
With
Strong
Central
Spar
Hinge
“Deck” or Floor
Powered Hinge Basics Diagram
Figure #2
The metallic pusher arc will need to be built large and strong, to handle all
of the forces involved. The petals-becoming-grid-fins will try to twist, flutter,
shake, and bend. More on that later.
The heat wall encloses a curved metal sheath (a slot; think of a scabbard).
The scabbard can dispense lubrication (grease) onto the heat-walled cogwheels
and (cogged) pusher arc. When the pusher arc becomes exposed to hot
ambient air (as it must, when extended), the grease will burn or be blown away,
to be replenished when it re-enters the “scabbard”.
The cog wheels (driven by hydraulic motors) MUST be VERY firmly held
into place, to handle the tremendous forces involved! Structurally, they can be
held up on a strong “post”, reinforced by “braces”. These structures may or may
not be encased by heat walls (or partially enclosed), as is convenient. The below
drawing should clarify and emphasize this point. Note that these drawings can
become quite cluttered, so they will sometimes be repeated, adding and
subtracting different elements.
H
Strong Structural Support
Structural
Post Strong
Axles
And Drivetrains
Support
Braces
Figure #3
The floor here (and the roof of the rocket body, together perhaps best
called the “deck”) will need to be built stronger than it would otherwise have to
be, with all of the forces exerted by the petals-becoming-grid-fins. The petals will
act like a manual, simple can opener with a prying handle (not the kind that you
turn). We don’t want to “open our can”, here, we want to recycle it without
excessive damage! Another way to think about it is to compare the “deck” here
with what Space X calls their “thrust puck”. We’ll likewise need to “beef up” this
structure at the top lip of the “can” of the rocket body, similar to a “thrust puck”.
But this is outside of our scope to say more. I have no detailed mechanical or
structural expertise here!
Another thing that can be done is to double up the pusher arcs. Think of
another one behind the one shown, in the above drawings. One can be located
at each end of the hinges. If the hinges are long enough, perhaps we can line up
3 or 4 of these structures (pusher arcs, scabbards, etc.) on each hinge. If we do
so, adjustments will need to be made on the petals. A strong central spar
becomes 2 or more such spars, perhaps with holes in the grid fins in between
multiple spars. We don’t want to go “solid planar” out there, or we’ll collect too
much force!
Then yet another thing that we can do to “beef up” the powered hinge is to
nest larger and smaller pusher arcs together. Below, the arrangement is shown,
conceptually, with the cogs stripped off of the pusher arcs, and the cogwheels
not show, to simplify, and to make room.
H
Nested Pusher Arcs
Figure #4
The above could have 4 separate “scabbards” for the 4 separate pusher
arcs. Or, we could go further yet, and smear all 4 arcs into a single solid plate
(like a pie shape with the innermost tip nibbled off), and place arcs of tracks of
holes in the plate, for cogwheels to run through these tracks (cog wheels rotate
90 degrees from what is shown in above drawings). Now we would have just
one scabbard, not 4, of course.
In Norse mythology, the “Bifrost Bridge” is a rainbow bridge between
otherworldly realms. Our “otherworldly realms” being bridged by our rainbow
bridge here are space, the exosphere, thermosphere, mesosphere, etc., so, for
several reasons, we will call this entire design of a powered hinge (with
variations) the “Bifrost Hinge” (please don’t think of me as being unhinged, for
that).
Now I have yet another idea that I consider worthy of describing, in terms
of the “Bifrost Hinge”, then. My mechanical intuition tells me that this is one of
the best versions, actually, for whatever it is worth. Picture in your mind’s eye,
each of these cogged-on-both-edges pusher-arcs being replaced by TWO solid
arc-plates, with rungs (like ladder-rungs) placed between the two (identically
sized and shaped) pusher-arcs. This should be easier to manufacture than the
cogged-on-both-edges pusher-arcs. The cogwheels then grab the “ladder” rungs
(be sure to phase the cogs-pushing in an interleaved pattern, from one cogwheel
to the next, for evenly modulated force). It is NOT mandatory to place the
cogwheels on both sides of the “ladder”. If the outermost edges of the ladder-
sides are VERY firmly supported and lubricated by the “scabbard”, then all
cogwheels could reside on the inner edge. The above-described solid “pusher
plate” with tracks of holes, option, is OUT, for this design variation, because such
a plate resides where the cogwheels want to go, here, and in the exact same
plane. Other options (nested rainbow style and one behind the other, on the
same hinge) do remain viable, for the “ladder” approach.
An illustration is easy enough to provide, so here it is:
H
Ladder
Rungs
Ladder
Sides,
2; 1
Hidden
Ladder-Style Arc Pusher
Figure #5
There’s one more mechanical aspect of “Bifrost Hinges” that I’d like to
discus, but first, let’s briefly detour, and then return. You’ll see why (assuming
you’re a linear reader) in a moment.
Operational (Flight) Modes
The “flower bud” will remain tightly closed, enclosing clean, pressurized air
(for most applications) surrounding the payload, all of the way up to orbit. The
“petals” will then open fully, to get well out of the way for payload deployment.
Petal-opening forces there will be minimal. There may be several stages of
payload (satellite) deployments, sometimes even with orbital-adjustment burns in
between. If the user desires this, one could close the flower buds back up for
intermediate journeys between different orbits. If nothing else, closing the “bud”
back up would protect the payload from being impacted by orbiting trash, during
these journeys. Opening and closing the petals will require minimal forces in a
vacuum.
Next, the de-orbit burn is performed. I’m assuming that we’d leave the
flower fully opened up for this, or open it back up immediately after the burn.
Now, in the outermost, thinnest whispers of air, we can kill a maximum
magnitude of speed, spreading out the heat and speed-wasting efforts, in the
“calm before the storm”. More slight stress now means less peak stress later.
So fully opening up the petals during this time would be best. But stresses on
the petal-spars and Bifrost hinges will remain minimal, during this time.
As the heat-storm builds up, the forces will amplify on the petals. The
petals will degrade into grid fins. During this time, the petals-becoming-grid-fins
should retract more and more, and not be allowed to be subjected to dangerously
excessive forces. Imbalanced forces from one petal to another (as some
degrade into grid fins faster than others) can be corrected for, by deploying the
most-degraded (less force-gathering) grid fins out further than those who are less
degraded. Measuring the tilt of the vehicle may be “good enough” to inform the
decisions of the avionics package, with regard to this. One might wish to equip
the support structures internal to the Bifrost hinge system with “strain gauges”,
and hydraulic pressures within the hydraulic circuits, to help inform the avionics
control circuits, is an added parenthetical thought, though. During this time, there
will be tremendous forces pushing IN on the arc-pushers, making them hard to
push OUT, but ridiculously easy to pull IN.
At some time towards the end of the process of petals degrading into grid
fins, or shortly thereafter, we should enter a slow “rotisserie roll” mode (to spread
the heat on the rocket body), which is caused by a slow process of rolling around
the rim of the deck, a pattern of the grid fins being deployed in an imbalanced
fashion. The rotisserie roll will be accompanied by a corkscrew descent pattern
for the entire vehicle. Once again, the grid fins may not be deployed very far
outwards, and the Bifrost hinges will be hard to push out, but easy to retract. For
more details about the rotisserie roll, and a drawing, see the parent document.
Figure #3 there is of interest.
Now, during this (rotisserie roll) mode, will the grid fins ever be deployed
to their maximum reach? It’s hard to tell… How much mass are we willing to
spend to make the grid-fins and grid-fins spars, the Bifrost hinges and supports,
and the payload deck strong? How much of the grid fins will be planar (as in the
spars for interfacing to the Bifrost hinge’s pusher arcs), v/s how perforated will
the grid fins be? There are many variables at play here.
Once we’ve reduced our speed (to sub-sonic levels) in the lower (much
thicker) atmosphere, where we might want to deploy parachutes, the forces on
the grid fins will be much reduced. Here, for maximum added “parachute effect”
of the grid fins, they’d best be deployed to their maximum possible reach. But
forces at this stage, on the grid fins, will be minimal. If parachute fabric can
survive here, grid fins (and their supports) will survive with colors flying!
Back to the Bifrost Hinges Designs
So the most critical remaining open question from above is, during
“rotisserie roll” mode, will we, or will we not, want to maximally extend our pusher
arcs outwards, in any maximal-stress mode? (I think this is the only time we’d
maximally extend them in a high-stress mode, but I could be wrong). ***IF*** we
NEVER extend maximally, in a high-stress mode, then a question arises, which
is, “Can we attain any safe mass reductions in our Bifrost hinges design, at the
maximal reaches of the highest degrees of extension”?
We can’t play games with the strength of the deck, or of the post or braces
that fix (in space) the hydraulic motors and cogwheels (“drivetrain”), or the true
hinge, or the drivetrain itself, or the spars on the petals-becoming-grid-fins. All
that is left is pusher-arc itself. There, if our conditions are met, we can weaken
the points on the pusher-arcs, at the points furthest-removed from where these
arcs meet the petals. This assumes that pushing hydraulic motors and
cogwheels are clustered at the ends of the pusher-arcs which are closest to the
petals (grid fins), as they should be. There, at the far-removed tips of the pusher-
arcs, we can add voids (holes), or internal inclusions of lower-density (weaker)
metals, or other materials. It’s not much, but there it is! We can weaken the
remote tips of the pusher-arcs, if the design (and operational) parameters allow.
Mass reductions are precious here, in ANY space-travel context! Enough said.
Now, all of that being said, let’s describe one more variation (or set of
variations) on the Bifrost Hinge. This is another top contender, and perhaps
even the best of them all.
We have already briefly discussed a variation of what is shown above, in
Figure #4, where the multiple pusher-arcs can be smeared together, and tracks
of holes can be placed into the smeared-together plate (in the shape of a
truncated pie-slice). Cogwheels could push on these arc-shaped tracks of holes.
Let’s take this in a different direction. Place no holes in the truncated-pie-
slice-shaped plate, and turn it into a dry friction plate. This means that we don’t
intend to lubricate it. The dry friction plate mates to a strong spar on the petal, of
course. On both sides of the friction plate, there are truncated-funnel shapes,
which strongly squeeze the friction plate together. The truncated funnel shapes
have multiple grooves cut into the ODs (Outer Diameters) where V-belts can
couple to multiple hydraulic motors. The V-belt grooves are cut deep enough, so
that the V-belts don’t touch the friction plate. Parenthetically, V-belts could here
be replaced by chains, with cogs on the funnels. I prefer the idea of using V-
belts, because the funnels are easier to manufacture, and there’s no cogs to be
broken off. V-belts would need to be strong and durable, and inspected (or
replaced) between uses. A torn V-belt caught up between a funnel and the
friction plate could easily (and quickly) spell disaster!
In either case, the funnels are free to spin on bearings at the small-tip area
of the funnel (which has been truncated), and at or near the open end of the
funnel as well. Probably best, a spin axis (axle) would run clear through the
entire funnel, and would be lubricated. All of this (the entire drivetrain and
support structures) would need to be ruggedly built. Possibly, some areas (of the
funnel) could include voids (or inclusions of less-dense, weaker metals or other
materials), AKA be “honeycombed”, for saving mass.
The funnels, motors, and V-belts can all be heat-walled, with only one
significant area of heat exposure, and that’s where the (much-squeezed) friction
plate travels in and out of the slot between the two rolling funnels. Advantages
here, then, are that no lubrication is exposed to high temperatures, and that we
allow “slippage”. If there are sudden jolts (“peak loads”) of high forces imparted
to the petals (grid fins) and their spars and the friction plates, we’ll not get broken
cogs… We’ll instead get slippage between the friction plate and the funnels, and
possibly also between the funnels and the V-Belts.
The above is an introductory summary of where we’re going here, and
why. Keep in mind that the linear distance travelled on the surface of the funnel
(for a given degree of rotation) is much greater at the OD (Outer Diameter) than
the ID. We want to optimally lay out a “grip line” where the friction plate meets
the 2 “squeezer funnels”, and the linear speed rates of all elements agree with
each other, all along the “grip line”. Else our design elements conflict with one
another, and we get “binding” instead of relatively free motions.
Now before we start in on some drawings, it might be best to look at some
basics of the design here. I like to go intuitive and experimental on these things,
with a smattering of theory. The truncated pie-slice friction plate here, if we
ignore the thickness of it, can be modelled by a piece of paper, or the walls of a
cardboard box. Get yourself a funnel, where the funnel walls come together at
90 degrees at the tip of the funnel (if you can find one, go “close enough” and
imagine it being 90 degrees, or play with some paper and scissors, maybe even
some tape too). Imagine the friction plate being “rolled out” by your funnel. Or
place paint all around the outer surfaces of at least half of a funnel, and roll it on
a piece of paper, with the tip of the funnel held in one spot. The paint laid out on
the paper will look like our “friction plate”. A ½ (180 degrees) of a roll of the
funnel is needed to “lay out” the 1/4th of a circle that is “laid down” in our 1/4th
slice of pie, that represents the friction plate. As we imagine the spinning funnel
on the far side of the friction plate, we also see that the “grip line” MUST reside
(no choice here) at 45 degrees, halfway between the petal wall and the deck.
This assumes we put the funnel in “square” with the petal and deck. We can
make SOME adjustments to that, as we’ll see further below.
The whole scheme here can be envisioned as a sphere, with the center of
the sphere being in the middle of the center of the true hinge (where petal meets
deck). The two squeezer funnels are tapered, cylindrical cones, whose tips
“want to” reside at the exact center of our sphere, if we want precise agreement
along the length of the “grip line”. The friction plate (ignoring thickness) can be
envisioned as being “laid out” by the same kind of rolling cylindrical cone (right
angles of the sides, don’t forget) as we’ve described.
It is NOT a good idea to have the “grip line” between the two rolling
funnels meet at any kind of offset (of center of rotation) from the center of rotation
of the friction plate, whose center of rotation MUST reside in the middle of the
hinge! “Geometry testing by drawing software” (Visio in this case, in the below
drawing) should prove this point. “H” is the hinge (the true standard) and H1, H2,
and H3 are sample alternate centers of rotation, for the funnels, whose “grip
lines” shift with them. “Grip lines mismatched” means troubles, of course! This
means cheating is dangerous! The centers of rotations for the funnels need to
be at the center of the hinge, and, if offset, offset VERY little! The point of the
below drawing might be a little bit subtle, but a moment or two of consideration
should help to clarify matters.
More descriptions of this: We are building a “slip clutch” or “torque limiter”.
The two spinning-top-style funnels can squeeze the friction plate very hard, and
still remain fairly free to spin, while gripping (and moving) the plate firmly. This is
true IF the above rules are followed, and IF the funnels have 90-degree tips.
(This last part is true because the friction plate meets the petal at 90 degrees).
Otherwise, due to mismatched “grip lines”, your device will bind badly!
It’s probably high time for more drawings.
H
H1H2 H3
True v/s Shifted
Centers of
Rotation and
Grip
Lines
Friction
Plate
“Cheating” is Penalized!
True
Avoid or Minimize Offsets!
Figure #6
The below drawing (of a funnel) assumes that we use V-belts. We’ll
assign this idea as being less than optimal shortly, but here, let’s cover it anyway.
V-Belt Driven Funnel
H
Structural Posts
Friction Plate
Grip
Line
Axle & Bearings
for funnel
Friction Plate
Extended Below Deck
“Funnel”
V-Belt
Cut-Out for
V-Belt
Motor &
Pulley
(Void)
Figure #7
Several things quickly become apparent. The funnel bangs into our petal,
and the deck! Both petal and deck can be “bulged” or moved out of the way. We
can’t put holes into the petal (which needs to stay airtight on the way up), but we
can put airtight pockets into the deck, for the funnels. For the friction plate, we
can put pockets (“scabbards”), or, perhaps rubbery plastic air-seals (gaskets)
attached to the slot-lips to seal the minimal clearance-spaces between the deck-
slot and the friction plate. These gaskets would be free to burn or abrade away
later, during descent. If we want to push the petals out more than 45 degrees,
we’ll have to grow the friction plate below deck, with a slot cut into the deck. I’m
not sure how far the petals would have to be extended, to balance out their
usefulness v/s the costs of growing the friction plates. This is beyond my
expertise! PS, note, the axle (along with the entire funnel) will tilt towards the
viewer at the tops of the axle and funnel. The drawing is busy enough, without
trying to show that.
More than one set of V-belts could be used (I showed only one, to cut
clutter). The “grip” of the V-belts on the funnels would be good and long, and
gently curved, for low wear and tear on the V-belts. However, one would have to
find room to squeeze a motor and pulley (per each V-belt) in there as well. If
they are kept small, we can find some room (between the funnel and the deck,
away from the friction plate, and away from where the deck will have to be
bulged down to make room for the lowest parts of the funnel). Any other space,
especially if we want a large pulley, eats up valuable room. And a small pulley
sharply bends the V-belt, causing rapid wear. Parenthetically, to envision all this
in 3D, it helps to place 2 funnels in 2 cardboard boxes, with the boxes wall-to-
wall, and the tips of the funnels close together, in the bottom corners of the 2
boxes. The walls between the 2 boxes is the friction plate.
Probably yet worse, to protect the V-belts from the heat, we’ll have to
shield (“heat wall”) most of the funnel (except we can’t do that where the funnel
touches the friction plate, of course). That means we eat up more space, and
have to “invade” the petal and the deck some more. Note that we ARE free to
rotate the funnel clockwise a tad if we want to (in the above views), to avoid
bulging the petal as much, but that, of course, invades the deck yet that much
more. Also note, the cut-outs for the V-belts (grooves in the funnels) will
segment (void parts of) our grip line, reducing the grip of our slip-clutch. So it’s
probably best to totally drop the ideas about using V-belts.
Now why did I like V-belts in the first place? For “defense in depth”
against “peak loading” on the slip-clutch function here. V-belts can slip, without
breaking cogs. But if we think about it, looking at figures 2-3-4-5, the slip-
clutches could be embedded into the couplings between hydraulic motors, and
the driving cogwheels. The driving cogwheels will still have inertia, so EXTREME
(and sudden) peak loads could still break cogs (the slip-clutch function is further
removed from the true hinge, with the use of driving cogwheels). All in all,
though… V-belts have too many disadvantages!
More minor parenthetical notes: The bearings at both ends of the axle
through the funnel will need to be lubricated, so therefore, heat-shielded as well
(regardless of V-belts v/s cogwheels being used). The bearings stay in fixed
locations, so we are fine (I just don’t show the obvious things in the drawings, to
reduce clutter). Also, I make no comments here as to dimensions or to scale, in
a lot of cases. How large will the “Bifrost Hinges” system need to be, with
respect to the sizes of the petals, and the entire deck? I don’t know! I don’t have
the expertise required, here, to hazard a good guess, even. “Not tiny” is clear,
though, because they will need to be STRONG!
Also note, the above drawing over-simplifies structural supports, above
and beyond not showing braces. The shorter post close to the true hinge can be
just one post. The taller post should be 3 or 4 posts, since the support as shown,
if duplicated on the side where the funnel meets the friction plate at the “grip
line”, should be moved to become two posts, straddling the grip line, with an
overhead structure to mount the bearing. Alternately, the axle can be longer, and
the post further away from the petal (and longer as well). I think that 2 shorter
posts is best. For symmetry and overall compactness, we might want to do the
same thing on the near side (close to the viewer). Re-stated, the drawing
overlaps the far-side post (hiding behind the front post) and the grip line a bit,
which is clearly not possible. Almost the entire oversimplified “support structure”
drawing above is a “glib fib”, because the funnel is tilted more complexly, in 3D,
than what is shown. We need firm supports, that fit correctly, and aren’t
excessively massive, is really about “enough said”, I think.
If we combine the funnel with cogwheels, we could do something like the
below:
H
“Funnel”
Friction Plate
Extended Below Deck
Petal
Wall
(Bulged)
(Partially
Hollowed Out) (Firmly Mounted)
Bearings
Structural
Post
Motor-Driven
Cogwheels
Cogwheels-Driven Funnel
Figure #8
The number of cogwheels could be 1 or more… And the bulges for
accommodating the funnels can be placed into the petals or the deck, or both. I
show (above) budging the petals out of the way. Upon thinking about it some
more, bulging the deck is probably far better. Why? Because tilting the funnels
(towards the petals) and where the friction plate becomes squeezed between
them may invite FOD (Foreign Object Damage), since materials from the petal
(as it degrades in high winds) may get caught in there. Tilting the funnels away
from the petals reduces this risk.
Tilting the funnels towards the deck, instead, means that we have to
invade the rocket body some more, to extend the friction plate, while still
retaining our ability to maximally extend our petals. In comments immediately
below Figure #7, we discussed adding rubbery gaskets to air-seal the narrow
gaps between the frictions plates and slots in the deck. This approach, too,
would risk the gaskets (as they degrade) creating FOD between the funnels and
the friction plate. So we might as well sheath (think “scabbards”) the entire
potential travel distances of the friction plates, keeping an air seal. The furthest-
away-from-the-petals points of the scabbards can meet (or be braced against)
the rocket body wall, below deck. This helps build structural strength. Search for
a “can opener” search-string further above in this document, to see comments
about structural strength here. In summary, to prevent FOD and to maintain
structural strength, the friction plates should best be sheathed (not gaskets-in-
slots-sealed), the funnels should be tilted into bulges in the deck, and the
sheaths (“scabbards”) should be buttressed against (or otherwise merged into)
the walls of the rocket body. Rocket-body designers will have to accommodate
this invasion, in the name of recycling most of our uppermost stage, if this
version of the Bifrost hinge is used.
Also note that (not clearly shown) where the friction plate meets the petal
wall, there’s plenty of room between the funnels, for putting a “V” shaped
structural element thickening the friction plate (bottom, sharp point of the “V”
shape pointing towards the grip line), for strengthening the joint where the petal
meets the friction plate. This is where a strong structural spar resides in the
middle of the petal-becomes-a-grid-fin. We’d have to look “downwards” onto the
friction plate (edge-on, onto the friction plate), between 2 funnels, to show that.
Ask for more drawings, if needed, at RocketSlinger@SBCGlobal.net. Small
parenthetical note: Placing more than one SET of funnels (plus one friction plate
per set) on one true hinge seems to me to be a crazy idea, unless the funnels
(etc.) are very small, or the hinge is very long. I wouldn’t recommend the idea.
Other than that, I think that the “Bifrost Hinge”, in many variations, has now been
thoroughly described.
Routing Airflow to an Energy-Harvesting Device or Devices
As we’ll see later (further below), I believe that a river-boat-style “paddle
wheel” energy-harvesting device will be most sensible for our uses here. So a
crude drawing of such a wheel is shown in the drawings below.
The parent document mentions the idea of placing the energy-harvesting
devices on twin rails or a grid-fins-style perforated plate, to thrust the harvesting
devices “out into the breezes” after payload deployment. Due to lack of high-
temperature lubricants (thus making transfer of mechanical energy via roller
chains be not such a good idea), and high-temperature hydraulic pumps, hoses,
and fluids likewise off bounds, the entire idea of mobile-location energy-
harvesting devices should be written off. And trying to embed the harvesting
devices into the petal walls (with good air seals) is similarly complex and
therefore off limits.
I have grown wiser now! As was apparently retold by Francis Bacon, “If
the mountain will not come to Muhammad, then Muhammad must go to the
mountain”. If we can’t move our energy-harvesting devices into the airstreams,
then we must bring our airstreams to the (fixed locations) energy-harvesting
devices. Where one petal meets another (where they are latched together by
electromechanical latches during ascent), we place bulges in the petals. These
bulges will turn our entire design into a yet more-so “partial hammerhead” design,
but it is worth the price. I envision only 2 (not 4) of these bulges, to preserve
symmetry, while keeping complexity down to a dull roar. Below, we repeat
elements of Figure #4 from the parent document to show what I mean.
Top-Down View of Deck Bulges
Placements in a Partial-Hammerhead
Design
Petals, 4
Opened
Hinges, 4
Powered
Profile of
Rocket
Fuselage
Bulges in
Deck and
Fairings
Bulges in
Deck and
Fairings
Sliding-
Plates
Airflow
Dampers
Figure #9
The purple-colored sliding airflow-damper plates (above) should be
located in line with, or slightly above or below the deck. When closed (shutting
off all airflow), they might mate directly to slots in the bottoms of the petals,
where the petals bulge out, there. I can see no reason why we can’t do it that
way. If, however, we have to leave an arc (half-circle) of structural material out
there, for some reason, after the petals open, that might work as well. If these
structural arcs get blown away during descent, that might be acceptable as well.
The sliding plates can be driven by hydraulic cylinders, with the cylinder
bodies themselves heat-shielded, and the long skinny metal rods left exposed to
the heat. Hydraulic cylinders are fairly simple, which is why I favor their use
here. Another alternate choice is a “screw actuator” type of linear actuator,
driven by a hydraulic motor. Close the damper until after the payload is ready to
deploy. Afterwards, withdraw the damper to allow airflow (unobstructed by the
rocket body) to power one edge of an energy-harvester (I favor a paddle-wheel
design). The sliding damper can double up to adjust how much or how little
airflow is fed to the energy-harvester, which should be located right above the
damper.
Rocket
Body
Airflow in
Descent
Airflow Dampers
Paddle
Wheels
Placements of Paddle Wheels and
Airflow Dampers
Figure #10
The wind-collecting sides of the paddle wheels are now exposed to high
winds (and high pressure), while the sheltered sides are not. The wheel axles
(and the hydraulic pumps that go with them) are now also fairly well sheltered
from the wind and heat, while staying in a fixed location. That’s what this whole
scheme is about!
I don’t know how much or how little wind we want to collect. The above
shows no “air ram-scoops” tacked onto the flanks of the rocket body. We could
put amplifying scoops or reducing scoops there as may be needed. The below
drawing is easy enough to provide, so here it is:
Rocket
Body
Airflow
Damper
Paddle
Wheels
Amplifying V/S Reducing Scoops
Airflow in
Descent Rocket
Body
Amplifying
Scoop
Reducing
Scoop
Figure #11
The reducing scoop may not make sense if one regards it as being
redundant with the airflow-reducing function of choking off the damper a bit.
However, if the adjustable damper is delicate compared to the strength of the
wind, a strongly-built reducing scoop may make more sense than beefing up the
adjustable damper.
It is entirely possible, too, that we could put air-scoops down on the flanks
of the rocket body, and invade the body of the rocket with angled air-pipes, and
bring the airflow fountains (geysers?) up into the deck, further away from the
edges of the deck. We’d still need our sliding dampers to choke off airflow during
ascent, to keep our clean air (around the payload) protected. This method would
eliminate our need to put special bulges in the petals (fairings), at the twin
expenses of making the rocket body more complex, and of impeding straight-line
airflow. This idea seems simple enough to me, that I provide no drawing.
Another idea is to put one single straight-line air-pipe right up the
centerline of the rocket body (in the middle of the rocket engines and everything).
Fuel and oxidizer tanks might turn into toroids or “donut shapes” to accommodate
this pipe. Now we could have just one “air fountain” in the middle of the deck. I
doubt that rocket-body designers would be too happy to make all of the changes
required (I could be wrong). Once again, this idea is simple enough, to skip the
drawing.
Energy-Harvesting Devices
I favor the paddle-wheel, for its simplicity. I don’t think I need to elaborate
on that much at all. If, as described above, one side of the wheel is in the high
winds, and the other side (as well as the axle and hydraulic pump) is sheltered
(and heat-walled), it should work just fine.
The other is a wind chainsaw, or chain windmill. See the abstract way up
above for a quick summary. See https://en.wikipedia.org/wiki/Bicycle_chain to
understand the terms used to describe the parts of a “roller chain”. Here, for a
chain windmill, we could just add (to the “outer plates” in the chain) rigid wind-
catching wind-vanes, similar to the vanes or paddles on a paddle wheel. We’d
need them, not on each and every outer plate of the chain, but, say, on every 5th
or 6th or 7th one, or some such.
If one side of the chain is in the stiff breeze (two cogged wheels carry the
chain, with the wheels being of a sufficiently large diameter), and the other side is
sheltered, we are doing fine… Except now, we have TWO wheels that need
lubricated, and their axles protected from heat, AND we have NO reliable method
of keeping the chain itself lubricated. This is extra trouble and complexity, for no
good reason, compared to the paddle wheel, in my opinion.
However, I did have 2 ideas that might be useful for a chain windmill” in
some other context, so I’ll briefly describe them “just for grins”. ***IF*** we did
NOT have a clear difference between a stiff breeze on one side, and shelter on
the other (and we could ideally also lubricate the chain, of course), then “folding
the vanes out of the way” on the chain’s return journey would make sense. Here
is what that would look like, in incarnation #1:
Outer Plate of Roller Chain With
Folding Vane
Solid Purple Vane
Wind Direction
Hard
Stop
Dotted Vane
Wind Direction
Figure #12
The hardstop angle could be adjusted if the angle of the wind chainsaw (to
the wind) is desired to be oblique rather than 90 degrees. The hardstop and the
vane should span from one outer plate to the other, on both sides of the roller
chain. Other than that, the drawing should make the idea fairly obvious.
The “butterfly vanes” approach is to have two pins poking upwards out of
the chain (from both outer plates on both sides of the chain), with the pins
serving as pivot axles for the “butterfly wings” (vanes), and each vane containing
the built-in hardstop for the other.
Wind Direction
Butterfly Vanes
Pins (Axles)
Wind
Direction
(Vanes Shown Shorter Than Best, Probably)
Figure #13
The parent document mentions a “screw turbine” as yet another energy-
harvesting device. Now that I know that the windward side of a re-entry vehicle
has pressures roughly 100 times that of the lee side, and have researched the
state of the art for heat-tolerant materials, I believe that the “screw turbines” idea
(whether ducted or not) is a vastly inferior design, for this application, compared
to the champion, which is the paddle wheel. A screw turbine would require
transferring power away from the center-line of the device, which presents all
sorts of troubles, with obstructing airflow and with heat-sheltering components. I
can speculate and provide drawings (contact me as usual at
RocketSlinger@SBCGlobal.net if desired), but I think that the screw turbine
deserves no more discussions or drawings here.
Hydraulics and Hydraulic Power Management
I’m an amateur here, as in so much else. Internet research (and the
Mighty Google, Which Knows All) empowers me here. Bidirectional hydraulic
motors are clearly a “thing”. At a casual glance, see
https://www.globalspec.com/industrial-directory/bidirectional_hydraulic_motors.
So, in the Bifrost hinge, we can do what we want, there. One motor can be used
to open and shut the hinges.
For bidirectional hydraulic oil pumps, see https://www.rg-
group.com/resources/blog/bi-directional-pumps, which says… “NOTE: Don’t
confuse bi-directional pumps with bi-rotational pumps. Bi-rotational pumps can
flow out of either port, but only when rotation reverses. A bi-rotational pump has
one port hooked to tank and the other port piped to the circuit.” Also it says
Normally, bi-directional pumps do not have a port piped to tank. So, in the one
and only place where I recommend the use of a hydraulic cylinder (in the
adjustable airflow dampener feeding the air to the paddle-wheel), we might wish
to go “closed hydraulic circuit” and use a bi-directional pump, to power the
cylinder, here.
As far as the “tank” is concerned, to my knowledge, this is a general-
purpose reservoir of hydraulic fluid, which, unlike a “hydraulic battery”, doesn’t
place the fluid under high pressure. In our application, though, we’ll have to deal
with zero gravity, as in, for example, when we first activate the Bifrost hinges (to
open the petals), before deploying the payload. We can’t deal with “sucking
vacuum” from the tank, because our oil is vacuum-cavitating and floating around
in blobs. We’ll have to create a “tank” (which can have weaker walls, unlike a
“hydraulic battery”) which at least weakly pressurizes the hydraulic fluid, behind a
sliding wall, just like in a hydraulic cylinder, where the wall is called a “seal” (or
piston), usually, it seems. This will keep us from spilling the contents of our tank,
in zero gravity! On the empty side of our seal, we can place (sealed) weak
nitrogen pressure, or mechanical springs. Use nitrogen here, not air; we don’t
want to risk burning our hydraulic fluid.
One of the things that I am wondering about is, during the “rotisserie roll”,
there will be tremendous amount of air pressure pressing in against one petal-
becoming-a-grid-fin, as we retract it, and extend its neighbor. Instead of bleeding
high pressure fluid back to the (low-pressure) “tank”, and having to re-spend
much energy fighting air pressure on the neighboring petal as we extend it, can
we switch high-pressure fluid from the being-retracted petal’s motor(s), to the
being-extended petal’s motor(s)? Solenoid-switched fluid-flow valves (including
“check valves”) and switches are a given. In electrical power management, we
can combine a motor and a generator and create an “alternator”. Can we do the
same thing in hydraulics, and create a combined hydraulic motor-and-pump?
This would allow us to efficiently do the “rotisserie roll”, by having one combined
pump-motor supply the next, with high-pressure fluid, I think.
The answer isn’t immediately clear.
https://en.wikipedia.org/wiki/Hydraulic_motor seems to imply that it’s not
easy to do that. https://www.plantautomation-technology.com/products/kazel-
hydraulic/hydraulic-motor-pump-combination isn’t very clear (probably involves
electrical power).
https://www.hydraulicspneumatics.com/technologies/accumulators/article/218838
29/accumulators is good background information on accumulators.
https://www.graco.com/us/en/in-plant-
manufacturing/products/lubrication/automatic-lubrication/hydraulic-accumulator-
supply-pumps.html describes air-powered pumps to feed power to accumulators
(that one’s not very directly related to our application). So in any case, in our
“Bifrost Hinges” design, whether we need to combine a (birotational or
bidirectional) hydraulic motor with a pump (as 2 discrete units or as one unified
unit), we CAN convert mechanical motion (from relieving high fluid pressures as
the petal retracts), to pump-power the extension of the next petal. It CAN be
done! If “Graco” can do it (pump hydraulic fluid) with air pressure, we can do it
with hydraulic fluid pressure.
Partially repeated from a few paragraphs up, see https://www.rg-
group.com/resources/blog/bi-directional-pumps, which says… “NOTE: Don’t
confuse bi-directional pumps with bi-rotational pumps. (Deletions here). A bi-
rotational pump has one port hooked to tank and the other port piped to the
circuit.” So then, my question would be, when the petal is pushing in strongly on
us (and we want to allow it to retract), can we use a bi-rotational pump, and pipe
high-pressure fluid, not to the tank, but to an accumulator? Or over to the next
petal that needs to be extended? If the answer is “yes”, then we are “home for
free”! We get to extract power as we retract!
If all else fails, we can STILL do this job of recovering some power, I will
bet! See some of the above drawings of the Bifrost hinges (especially figures #2,
3, 5, and 8) and we can see that there are multiple locations available, in the
drivetrain, for pumps and motors. If need be, we could place a weak
unidirectional motor for powered retraction (as in a vacuum, for post-payload-
deployment petal-retraction, if desired, when the wind isn’t pushing us), plus a
strong unidirectional motor for powered deployment and place-holding (braking),
PLUS a pump, to extract power to be fed back to an accumulator, or to a
neighboring petal that needs extended, as we are retracting the strongly-being-
pushed-on petal on our given instance.
As I understand it (I could be wrong), our strong unidirectional motor (in
this scenario) will be fed high pressure fluid-feed to push outward on the petal. It
will maintain that high fluid pressure (from a pump and-or accumulator) in
“braking” mode. Now we could suddenly relax that motor-brake and feed excess
fluid back to “tank” (don’t let the being-fed high-pressure fluids short-circuit to
tank), at roughly the same time as we engage the pump, to turn strong
mechanical energy (from the petal being pushed inwards by winds) back into
fluid pressure (fed to the accumulator and-or the neighboring petal and Bifrost
hinge, of course). That should do it! (Parenthetically, in electrical switching,
especially with electromechanical relays, we talk of make-before-break, v/s
break-before-make connections. I don’t know what we’d do here with hydraulic
switching).
Certainty concerning the above (or more details) are beyond me! No
hydraulic circuits drawings will be provided, at this time, from Yours Truly!
A “hydraulic battery” will need to be thicker-walled (compared to a “tank”),
to contain higher pressures. See
https://www.hydraulicspneumatics.com/technologies/accumulators/article/218829
59/hydraulic-batteries-save-fuel “Hydraulic ‘Batteries’ Save Fuel”. I’m no
hydraulics expert, but I’m willing to bet that we need one (or more) in the systems
envisioned here. What I can offer up, is an idea: The air (or mechanical springs,
etc.) pressure in an accumulator could be indirectly replaced, in our application,
with the airflow pressure exerted by our vehicle’s re-entry. After our high-speed
re-entry is done, of course, our high-pressure source will be fading away, and our
Bifrost-hinges-stored hydraulic-fluid pressure (more importantly, also our stored
potential mechanical energy) will largely fade away with it also. But during the
maximum-stresses parts of re-entry, this scheme may work just fine.
That is, our special accumulator is an “airflow spoiler” down on the flank of
the rocket body. That does means that we have to move hydraulic fluids down
there and back, in a heat-walled manner. I think (and hope) that this can be
done without invading the body of the rocket all that terribly much. We also
assume that we can find a way to feed hydraulic fluid pressures back and forth
through a Bifrost hinge (see slightly above) to substitute for air pressure in a
hydro-pneumatic accumulator. The “accumulation” function is stored potential
mechanical energy in the form of the extended friction plates (or pusher arcs) in
the Bifrost hinges, which can pump fluid as they are pushed back in by the strong
prevailing winds, is another way to put it. Such a combined airflow spoiler and
accumulator could be located wherever best makes sense.
Combined Accumulator and
Airflow Spoiler
Rocket
Body
Hinge
Bifrost
Airflow in
Descent
H
Airflow Spoiler
Heat-Walled Hydraulic
Fluid Lines
Figure #14
The hinged plate should roughly conform to the rocket body, and be heat-
shielded in the same manner as the surface of the rocket body is heat-shielded,
or better. It is assumed to be solid planar here, not a “grid fin”, although I
suppose it could possibly be grid-fin-style, as another choice. If it is solid planar,
it will gather a lot of force, and so, should be built thick and rugged.
The airflow spoiler could possibly be located near the bulge in the deck
and the petals, and the optional reducing air scoop; see figures #9, 10, and 11.
The spoiler would perhaps protect the sliding plate (airflow damper) and paddle
wheel from too-strong airflows. If this is going to be a purpose of the combined
airflow spoiler and accumulator, the two sometimes-conflicting purposes must
somehow be balanced, of course.
The spoilers might best be added in balanced pairs, to maintain overall
symmetry. Note also that the “power storage” function here of a spoiler plate’s
extension on a Bifrost hinge is really the same as what we have in a petal-
becomes-a-grid-fin and a Bifrost hinge. The difference is that the former isn’t
envisioned as being part of the “rotisserie roll” function, and the latter is. So the
former is more purely an “auxiliary power accumulator”. Once again, all of this
assumes that we can pull fluid-pumping power out of extended-against-the-wind
grid fins and spoilers, which I do firmly believe is a safe assumption.
Avionics and Flight Control
I’ve previously commented (both here and in the parent document) about
the “rotisserie roll” and an associated corkscrew descent pattern for the entire
vehicle. The avionics circuits (controlling the hydraulics of course) may, from
time to time, shift from overall flight-path corrections as being the primary
concern, to maintaining the rotisserie roll. If all is working well, most of the time,
both objectives can be met at the same time. And as previously remarked,
during the time that the petals degrade and become grid fins, imbalances here
can be corrected by adjusting which petal is further deployed than another. I
have nothing more to add to that, other than the below…
Immediately below Figure #8, I added comments about FOD (Foreign
Object Damage), as materials from degrading petals may get caught between
elements of the Bifrost hinge (of any design variation, really). We should design
to prevent this, both in the Bifrost hinge, and in the materials that will ablate,
burn, or fall away, as petals degrade.
However, we should probably design into the avionics, methods of getting
ingested FOD materials out of the Bifrost hinges. If sensors (motion sensors,
hydraulic fluid pressure sensors, strain gauges, etc.) detect a “stuck hinge”, the
hinge should be extended a bit, to see if that fixes it (disgorge the FOD
materials). Make several attempts before giving up. So along with maintaining
the flight path and the rotisserie roll, this might best be a third duty of the avionics
circuits, with the duties interleaved as need be.
Conclusions
“Bifrost Hinges” (in any of several design variations) would make good
powered hinges for a shuttlecock-style atmospheric re-entry. Paddlewheels
could be used to collect power. Airflow spoilers could serve combined purposes
of slowing the descent (spoiling airflow of course) and as “hydraulic
accumulators”, which store energy in the form of potential mechanical energy, as
Bifrost hinges are extended against strong winds.
Open questions remain. Can a port from a bi-rotational pump be routed to
an accumulator instead of a tank, thereby conserving power? Can a hydraulic
pump be combined with a hydraulic motor, without electrical power being
involved? Are the above design notes optimal, especially concerning hydraulics?
What would a detailed “hydraulic circuit” diagram look like, for a Bifrost hinge? I
plan to see if I can get a hydraulics expert to comment. Maybe I’ll write or co-
author another paper to flesh this out some more. Please contact me if you are
(or know such) an expert who might be willing to comment, contribute, or co-
author. See my email address immediately below.
I have no special expertise or any more plausible ideas concerning any
associated matters here, so I will sign off at this time. This concludes my ideas
as of this time. Comments or questions (or idea contributions) are welcomed at
RocketSlinger@SBCGlobal.net
Stay tuned… Talk to me! RocketSlinger@SBCGlobal.net
References
Stauffer, Titus. (2021). Methods of Decelerating a Spacecraft Through
Atmospheric Re-entry Using a Shuttlecock-like Design.
Stauffer, Titus. (2019). Designs for Passively, Thermally Gated Fluid Flow
Switches.
Back to main site at www.rocketslinger.com
Article
Full-text available
Abstract This sub-page to www.rocketslinger.com is meant to describe methods for converting a vacuum rocket engine nozzle (with a large bell shape) into one that can operate efficiently and safely at sea level (with a much smaller bell shape, with today's typical design). Three primary alternatives are described… '1) One could put an array of hinged gates (NOT permitting one-way flow of hot rocket exhaust gasses into the lower-pressure ambient environment, for vacuum or low-ambient-pressure applications) into the extended areas of the bell shape (those exceeding what is needed for sea-level operation). However, the one-way-gas-gates WOULD allow ambient air to flow (at sea level or in other higher-pressure environments) inwards into the rocket nozzle, thereby reducing "back flow" of air (along the insides of the outermost rim of the nozzle, when the nozzle is "over-expanded"), which is associated with inefficiency, chaotic oscillations, and danger. '2) One could design the outermost (lower) extensions of the nozzle to be used in a vacuum environment (as in, for example, the Space-X Falcon 9 interim upper stage, formerly being design-tweaked to begin emulating the upper stage of the "BFR" system, AKA, the BF Spaceship), but to fall away (be discarded) immediately after the final de-orbit burn. This converts the bell shape from a vacuum nozzle into a sea-level nozzle, by drastically reducing the nozzle in size. '3) One could formulate the materials and construction of the outermost extensions of the nozzle to be burned away by the heat of atmospheric re-entry. This might be an especially attractive option if one uses a shuttlecock-style re-entry method with the rocket engine(s) at the "fore" end during re-entry. Options for precisely this method (rocket engines at the fore end in shuttlecock-style re-entry) have already been described (citations provided here). Designing these outermost nozzle portions to be sacrificially burned or ablated away will help protect the engines, and facilitate their re-use.
Article
Full-text available
Pre-Summary This sub-page to www.rocketslinger.com is a companion or supplementary document to other documents found at Research Gate. As previously documented, conventional (existing) satellite-deployment fairings could be modified and subdivided into (an ideal number being postulated to be 4) powered-hinge-attached fairings segments renamed to be "petals", as they open (like a flower) for satellite deployment, and thereafter serve as the "feathers" in a badminton-style shuttlecock, for descent (re-entry). As these "feathers" degrade in re-entry-generated heat and pressure, they evolve into embedded grid fins. The powered hinges and hydraulic energy management used to accomplish this have already (albeit incompletely) been described. Here, we add additional design elements, notes, and drawings to emphasize that the highest-velocity phases of re-entry could be substantially prolonged (therefore prolonging the exchange of kinetic and momentum energy in exchange for heat energy, AKA reducing "heat flux", and reducing overall thermal stresses to vehicle structures) by using lifting-body principles for the highest-thermal-stresses phases of the re-entry process. The shuttlecock elements (and flight modes) earlier described aren't entirely negated here; they are more-so (much!) improved upon. Also important is this: What is described here is a "punctuated equilibrium" of flight of a lifting body, where the vehicle periodically rolls by 180 degrees. "Belly" and "back" of the vehicle are periodically reversed, to absorb and shed heat more equally, across all surfaces.
Article
Full-text available
Pre-Summary This sub-page to www.rocketslinger.com describes how payload fairings for an upper stage could be modified to remain attached to the satellite(s) deployment mechanism(s), and to one another, and also (optionally but highly desirable) to the entire interim upper stage (fuselage, fuel and oxidizer tanks, and rocket engines, etc.) as well. Whereas a commonly used design today involves the use of 2 (two) clamshell fairings, which are jettisoned before satellite deployment, the base design here envisions "N" fairing-segments, or "flower petals", or simply "petals". The petals can also be thought of as "feathers" in a feathered badminton shuttlecock. Each of 2 clamshell fairings are now subdivided (sliced) into smaller segments, with each shell-slice being vaguely reminiscent, not only of a flower petal, but also of the skin-slices of a sliced orange. For this document, "N" will be set to "4", with one (marked) exception. Each petal remains attached to the base of the assembly (ideally but not always, this assembly is the entire upper stage), at the base of the petal, by hydraulic actuators. During re-entry, the petals are sometimes more-retracted (less deployed) and sometimes more-deployed (less retracted), as one travels radially around the periphery of the bulk (center of mass) of the "shuttlecock". "N" = 4 petals is enough to control the descent path (vector), as controlled by the hydraulic (or otherwise-powered) actuators. Unlike the grid fins used by Space X on Falcon 9, these petals will not rotate on an axis that skewers the centerline of the fuselage, but rather, will be retracted to be almost vertically parallel with the fuselage when fully retracted, and nearly perpendicular to the fuselage when fully deployed. This is the "degree of freedom" which is needed already anyway, to protect the payload on ascent ("flower" is in the "bud" stage; payload = pollen not yet deployed, so to speak). The "bud" must turn into a "flower" to deliver the payload. "Petal control" is a bit complex, but highly important! Not only are the petals adjusted to control the descent path, but also, from one side of the center of mass (of the fuselage or capsule) to the other, the deployed petals are more-deployed on one side than on the other. This lop-sided deployment slowly spins around the center of mass. Thus, the (usually cylindrical) center of mass tilts, and does a slow "rotisserie roll" to spread the heat of re-entry across all of its surfaces. Each petal also gets to spend some time partially protected from the highest re-entry heat as well. A given fuselage-surface (or petal) element will slowly alternate time in the heat v/s time in the relatively sheltered slipstream. The probably-ideal flight path might best be described as a corkscrew. This will prolong the travel distance and time, providing more time to shed heat. Solid petals would gather too much force from the hot air or plasma during re-entry, requiring excessive mass (for both petals and actuators) for strength, to avoid being sheared off. This is why Space X uses "grid fins" rather than planar solids. But we need solids to fully protect the payload during ascent. "Having your cake and eating it too" is obtained here as follows: The petals are manufactured as grids, but are filled in (grid-holes plugged) by optimized plastics or ceramics (or other lower-melting-or-burning-point materials) formulated to melt, burn, and-or fall away during the heat of re-entry. This means more work for refurbing the petals before re-using them (if they are re-used at all), but this is a small price to pay for recycling an entire upper stage.
Article
Full-text available
From RocketSlinger@SBCGlobal.net (email me there please)… This is a sub-site to main site at www.rocketslinger.com … Abstract / Pre-Summary This sub-page to www.rocketslinger.com is meant to describe methods for designing a passively powered (unpowered by electronics) flow-switch for the control of the flow of fluids. The designs are kept as simple as possible. The temperature-based discriminatory function is based on the differential expansion coefficient of different metals (or other materials), AKA "coefficient of thermal expansion". See https://en.wikipedia.org/wiki/Thermal_expansion. The variations of such a design (as described here) could be applied to many different applications of a thermally controlled fluid-flow switch. However, the primary envisioned application is for a "transpiration cooled" atmospheric-re-entry skin of a spacecraft. A common example of "transpiration cooling" would be the evaporation of sweat from the skin of a human or other mammal. As with other sub-pages of www.rocketslinger.com , the intent here is to "defensively publish" propulsion-related (and "misc.") ideas, to make them available to everyone "for free", and to prevent "patent trolling" of (mostly) simple, basic ideas.