Read Latex

Tuesday, January 22, 2019

Computing and the Future 5 - Spacecraft Lifetime



First Impressions of an Article on Spacecraft Lifetime 



- image credit Boeing


The Space Review:
Moore's Law, Wright's Law and the countdown to exponential space




In this article, Berleantz et. al. asks the question, "Is there a Moore's law that can be established around the proxy variable of Spacecraft Lifetime?" He goes about devising one using the precedents of Wright's Law for aircraft and Moore's law for integrated circuits.


Definitions

Moore's Law


In 1965, Gordon Moore noticed that the number of transistors per square inch on integrated circuits had doubled every year since their invention. Moore's law predicts that this trend will continue into the foreseeable future. ... The 18-month mark is the current definition of Moore's law.

Wright’s Law

Also called the Rule of Experience, was discovered by Theodore P. Wright and described in the paper, “Factors affecting the cost of airplanes” in the 1936 Journal of Aeronautical Sciences[1]. The simple form of the law is “we learn by doing” and the cost of each unit produced decreases as a function of the cumulative number of units produced.

Spacecraft Lifetime

What does it mean for a spacecraft to be alive? The structure (structs) of a spacecraft could be intact, but its communication (comm) systems could have failed due to radiation damage. Structs and comm could be intact but an instrument (inst) could have failed (Hubble).  Struct, comm, and inst could be intact but attitude control (gyro) could have failed. You get the idea. Within each of these categories are gating factors, such as:

  • radiation budget
  • cosmic ray energy
  • collision energy
  • number of revolutions
  • component MTBF, e.g. capacitors

Key Idea: Moore's and Wright's Law's apply to the cost of making something per unit of performance, and don't speak to how long that manufactured article will last. The cost of something and how long it will last are, to first order, orthogonal attributes.

I say "first order" because of the implication that if something costs more, it may last longer, per the engineering adage, "Good, fast cheap. Pick two."

It might be useful to think in a statistical mechanics sort of way, using the notion of "mean free path" and compute the probability of collision.

Think of the ensemble spacecraft as a particle in a statistical gas and ask how long it will go before it collides with something that will damage it.




Colliding with a paint fleck in orbit can have the same impact as a slug fired from a handgun.


- go there

There's a lot of debris in orbit, so the density of the statistical gas varies significantly with altitude and orbital parameters.



- go there

We need to include those particles in our statistical gas whose kinetic energies are on the order of those required to produce structural or electronic damage. As such our gas consists of several families of "molecules" including cosmic rays, radiation fields and space debris.

The situation that drove the accuracy of Moore's law was that it was based on a
"printed" technology, the principle variable over time being feature size over a planar two-dimensional area. This gave p
rinting technology the particular exponential property that led to Moore's observations of transistor density. The interdisciplinary 'noise' that arose came from changing substrates and logic gate voltages did not impact the predictions. Perhaps that is since printing does not care what 'color' the ink is, the ink-color being analogous to the substrate, such as DTL, TTL, MOS, CMOS, and GAN.

To assess the proxy variable  'spacecraft lifetime' consider the predecessor variable 'aircraft lifetime'.

Consider for example the B-52 Stratofortress making its maiden flight in April of 1952. No aerospace engineer in their right mind would think that a flock of these would still be flying 67 years later, but that is exactly the situation.

Enabling this spectacular lifetime has been continual ground maintenance and propulsion and electronic upgrades. The basic structural chassis of the B-52 has remained the same but comm, nav and warfare electronics have been changed out many times over with advances like those predicted by Moore’s law. In that sense, the B-52, excepting the airframe, bears little resemblance to the April, 1952 version. It is also worth noting that a B-52 can be landed and serviced. An exploratory spacecraft typically cannot. We will revisit this in a moment.

Getting back to our (cost per unit of performance) vs. lifetime idea however, notice that production of B-52's has stopped, so Moore's law transforms from a rapidly growing exponential to an unchanging constant, or even undefined given they aren't made anymore.

Consider the care with which spacecraft are assembled using special materials, tooling and processes. These include accounting for dramatic thermal changes from launch into the hot, cold, hot vacuum of space.

Consider the principal risks besides the degradation due to thermal cycling:

  • risk of collision 
  • risk of exposure to radiation
  • risk of MTBF of finite electronic component lifetime (see Capacitor Lifetime)

None of these can be eliminated and each of them can bring a sudden death to a perfectly well-designed and properly functioning spacecraft. In these cases, everything is fine until it isn't and then there is very little one can do about it, especially considering the long travel times.

All of these are risks difficult to predict accurately and all can be complete showstoppers which leads me back to the mean free path model of spacecraft lifetime, rather that cost per unit of performance.

To see what such a curve looks like in the presence of noise I invite you to the mybinder example described in Computing and the Future 3 - Algorithms and Prediction. In this you can try various amounts of noise and various numbers of data points and degrees of polynomials. You could even plug in your own data since the Python notebook is open-sourced.

The idea to use the proxy variable spacecraft lifetime is interesting, but factors other than time may have more predictive value. I previously mentioned mean free path and radiation, both of which are temporally varying fields.

Let me propose the alternate proxy variables, spacecraft material & substrate.

There’s been a drive in civil aviation to use composite materials that consist of an epoxy resin matrix and fibrous materials such as graphite, fiberglass and boron. These materials have certain properties in a terrestrial environment where ambient pressure is available and other properties in the space environment where they outgas. Outgassing is rapid at first then decays according to some half-life characteristic of the material, its thermal, pressure, and photooxidative environment.

Spacecraft material is as important as any other property previously enumerated in determining its functional lifetime. This has, in gross anatomy terms, two components:


  • spacecraft structural material
  • spacecraft electronic materials and operating voltages

These can be quantified according to the MFP lifetime calculation given above. There is the need to harden these against collisions with either macro particles like paint flecks, or nanoparticles like fast protons and gamma rays which also have the ability to destroy them quickly.

So I assert that principal determinant of spacecraft lifetime is not the year that spacecraft was made but rather the materials that it was made of and their characteristic operating voltage. For example CMOS operating at 1.3 V switching voltages is more vulnerable to damage by static electricity than the 5V static tolerant TTL logic it replaced. It replaced TTL logic because of miniaturization and power consumption concerns. As we go to smaller and smaller, circuit feature size and spacecraft containing said circuits become more vulnerable to failure. A lower energy incident can disable the functioning circuitry and we know from muon collision that it’s difficult to shield against all kinds of insults, say with heavy lead or Bremsstrahlung emitting water-based shields.

Now let’s play this out in the context of ancient aircraft. The material from which WW1-era aircraft made of wood and covered with cotton and cellulose nitrate caused them to be vulnerable to both fire and photodegradation. Doped fabric has a much shorter lifetime than a corresponding aluminum-skin aircraft because aluminum is not as vulnerable to these environmental insults. Aluminum is still vulnerable to cracking, fatigue from vibration and thermal cycling however.

So here at an end of our "First Impression" we can connect Moore’s law with spacecraft lifetime. The drivers of Moore’s law has been reduction in feature size.

We can look at Moore’s law as not being a cost law or a speed law but really being a size law - that is driven by the characteristic feature size of the circuit.

As we decrease feature size there is MONSTER over the hill. The MONSTER is this:  We decrease spacecraft lifetime exponentially because it takes less of an radiation insult to disable it. We have in effect decreased the effective mean free path.

We might compensate by fielding spacecraft with technologies not so easily ablated by radiation or paint flecks - where the feature size of the computers that are installed on such spacecraft remain relatively large.

Alternately we may employ redundancy to achieve the same effect with the understanding that a single paint fleck or muon pushing explosively through 4 processors will still destroy the lot of them.

So now we only need two assumptions 

Spacecraft lifetime is proportional to the complexity of its electronic equipment.

Each successive spacecraft generation will last half as long as it’s predecessor because the feature size has gone by down by a factor of Moore's law

This leads us to the bizarre and humorous situation that in order for us to reliably field a spacecraft in successive Moore’s law generations, we must double the shielding surrounding it every 18 months and when we doubled the shielding we double the weight of the spacecraft and when we double the weight of the spacecraft we double the launch cost and in the limit we find ourselves with a spacecraft that can no longer be launched because it isn’t heavy enough to be reliable. This places us right back where we started - on the ground. If nothing else, this argument may provide a law for the most powerful computer that can be launched with a specified lifetime.

Evolutionarily successful satellites such as planets solve this problem by having two things:

  • an atmosphere and 
  • a magnetic field

Planetary evolution teaches us that, without a sustained magnetic field, one cannot continue to have an atmosphere! Ablation by the sun simply removes it, as we see on our moon, and on Mars, where there also happens to be no air, no food and until recently discovered, no water.

We might take a lesson from nature and equip our satellites with a magnetic field and an atmosphere so as to shield them from the harmful effects of radiation and material collision during the course of their journeys. (Which in the limit could be a good argument for staying home and taking care of the satellite we currently occupy, but I digress...)

This leads us to ask, "What combination of atmosphere and magnetic field yield a sustainable spacecraft/satellite?" This will determine the minimum launch mass.

The earth gets its magnetic field from nuclear energy, that fuels the hydrodynamic motion of a molten metal core.

From specific energy and specific power considerations, spacecraft would also have to use nuclear energy to generate a sustaining magnetic field. Note that unlike our earth, the atmosphere of a spacecraft could be charged particles held in place that would serve as a cushion for radiation and particulate impacts. This would enable the mean free path to be on our side.

The maximum lifetime of such spacecraft would then be determined by the half-life of the materials used for the radioisotope thermal generator or RTG, like Voyager 1 and 2 are using.

Because of material and orbital considerations this might lead us to mine comets and asteroids for their ionizable materials, for potential and kinetic energy.

Instead of mining those materials and returning them to earth we might create strapon packages that would 'hijack' comets/asteroids and redirect them to the destination of our choice using them as our material and energetic resources along the way.


Returning to the atmosphere discussion, our atmosphere is held in place by gravity at a scale height corresponding to the free gas available.

Articulating the obvious - it’s not practical for us to launch from earth objects that are large enough to have a significant gravitational field -  the energy costs of laws are prohibitive. Yet we know we must generate and retain both a magnetic field and an atmosphere of some significant depth so as to insulate ourselves both from radiation and from collisions.

This means that instead of relying on the gravitational force to retain an atmosphere we would utilize an electrostatic or charged envelope around the spacecraft. This makes sense since the electromagnetic forces are orders of magnitude more powerful than gravitational ones.

This charged atmosphere could be dense enough to serve as an ablative and protective shield to deflect incident cosmic radiation and ballistic insult to the spacecraft.

We could then make lifetime estimates based on the size of this envelope and cost estimates based on the mass of the generating equipment necessary to create and maintain it.  This makes a comet hijacking package interesting - a Genesis-style strap-on - as a feasible method of long duration space exploration.

Waxing more fanciful: If we face these constraints based on first principles other civilizations would experience similar constraints. Trying to perform space exploration this way leads us to ask the question if alien civilizations have in fact used a strapon packages to hijack comets and asteroids to navigate to various celestial outposts.




My feeling is that it is better to explore with photons, which travel rapidly and are massless, than with more massive, slower materials.

Since time is limited, let me finish with this conclusion: Aliens, including ourselves are best advised to hijack comets and asteroids with strap-on Genesis packages to explore other worlds in a material way. My justification for this bizarre statement is provided by the arguments above.



References:

Wright TP, (1936). “Factors affecting the costs of airplanes.” Journal of Aeronautical Sciences 10: 302-328

Researcher finds Moore's Law and Wright's Law best predict how tech improves: https://phys.org/news/2013-03-law-wright-tech.html

Wright's Law Edges Out Moore's Law in Predicting Technology Development:
https://goo.gl/HhNpjJ


Do your projects follow Wright’s Law?:
https://www.controleng.com/articles/do-your-projects-follow-wrights-law/


HTML Borders:
https://www.quackit.com/html/codes/html_borders.cfm


Moore's Law:
https://en.wikipedia.org/wiki/Moore%27s_law

Good, Fast, Cheap: You Can Only Pick Two!:
https://goo.gl/tbuwuX

Mean Free Path:
http://hyperphysics.phy-astr.gsu.edu/hbase/Kinetic/menfre.html#c2


Orbital Debris:https://goo.gl/xoDuMF

Scientists Design A Way To Clean Up Space Trash:
https://goo.gl/m8JTgS

The MOS Transistor:
http://www.cs.mun.ca/~paul/transistors/node1.html

The Silicon Engine Timeline:
https://www.computerhistory.org/siliconengine/timeline/

Boeing B-52 Stratofortress:
https://en.wikipedia.org/wiki/Boeing_B-52_Stratofortress

Capacitor Lifetime:
https://goo.gl/Ayned9

Aircraft Fabric Covering:
https://en.wikipedia.org/wiki/Aircraft_fabric_covering

Genesis Project: Star Trek - The Wrath of Khanhttps://goo.gl/GrRVdp

An Interstellar Tourist Barrels Through the Solar System
https://goo.gl/NoFPwQ

1 comment:

Anonymous said...

 In 1965, Gordon Moore noticed that the number of transistors per square inch on integrated circuits had doubled every year since their invention.

Not per square inch, but total (so bigger chips can contribute because they hold more stuff). But if you’ve read the 1965 paper I’m willing to negotiate on the details.

 Moore's and Wright's Law's apply to the cost of making something per unit of performance,

Cost per performance is a modern take on Moore’s law, but not what Moore suggested in 1965. Kurzweil focuses on the cost per performance metric.

 "Good, fast cheap. Pick two."

Fast and cheap are legitimate dimensions. Good is vague and a grab bag of multiple unspecified dimensions.

 Let me propose the alternate proxy variables, spacecraft material & substrate.

The problem with choosing proxies is they must be available on a per-spacecraft basis. Just see the Wikipedia page (esp. the info table) for any craft, or the databased pages at https://nssdc.gsfc.nasa.gov/nmc/SpacecraftQuery.jsp or at the EU site. If the info isn’t in one of those locations, it’s going to be a problem to use.

 Each successive spacecraft generation will last half as long as it’s predecessor because the feature size has gone by down by a factor of Moore's law

Maybe something like that. Cubesat craft have shorter lifespans and are becoming more frequent, so will have an impact on analyses like http://thespacereview.com/article/3273/1. I propose looking at lifespan/mass as an alternative metric. That analysis *is feasible* (the data is available at standard locations) but *has not been done*. It needs to be done! That may well be the best way to see a continuing Moore or Wright trend that holds even though cubesat type craft become common.

 My feeling is that it is better to explore with photons, which travel rapidly and are massless, than with more massive, slower materials.

How about speeding up communication with entanglement (“spooky action at a distance”) - https://www.sciencealert.com/einstein-spooky-action-demonstrated-on-massive-scale-for-first-time