Advancements in Electric Thrusters

[Impaler]

Per a request in the Mars Colonial Transport thread I'm going to present some of my findings on the current State of the Art in Electric thrusters.  Show how the technology has progressed in the last decade and try to look forward another 10 years to get an idea what we should be planning missions around.

I'd certainly encourage anyone else with knowledge in the field to contribute, but I'd like to keep discussion on thrusters systems only, and not get into power-sources as that really is a separate technology with it's own development history and trajectory.


GRIDDED ION

To most people gridded-ion or just 'ion engine' IS electric-propulsion, to my knowledge so far every deep space mission utilizing EP has used an ion engine and they are the engine type that NASA likes to show off most frequently as they a tech that was largely 'invented here'.  Deep-Space-1 was the first in a whole generation of probes that have revolutionized asteroid visitation missions in the US as well as in Europe and Japan.  Gridded Ion engines work at low power but achieve high ISP but low thrust, this makes them ideal for missions that are power constrained and need very high total DeltaV over a long period, in other words probes to the asteroid belt or outer-solar-system.

The development of Ion engines has seen them scale up rapidly since it was begun in earnest in the Early 1990's starting with the engine on DS-1 the NSTAR was followed by NEXT which has is now 'on the shelf' ready to fly once a mission needs it,  the next engine is the HiPEP which is in late stages of development.  The speculative concepts out their now are for 'Dual 4 grid' designs which offer extraordinary ISP while being a very simple modification of the existing design.

NSTAR  2.3 kw     3,300s
NEXT    6.9 kw     4,100s
HiPEP   37 kw      9,000s
DS4G   250 kw   19,000s

The problem with Gridded Ion is that it's got fundamentally low thrust and low thrust density and the ISP has now far over-run what is ideal for most desired missions because we want to get to the destination and do some science before the spacecraft or the mission operators on Earth are geriatric.  Thus Ion engine development is slowing down as any improvement along the current trend is only opening up potential missions to VERY distant outer-solar-system destinations and their are simply less missions to these kinds of destinations to drive new engine development.  The Europa mission which is looking more likely all the time will likely use NEXT or HiPEP and we will then be 'done' with development of new ion engines and just use the existing ones as workhorse engines for the probes that need them.


HALL THRUSTER

Hall is where all the action and excitement is now, if you've been following SEP closely you know that everyone is gaga over them right now.  The biggest thing HALL's have going for them is that they are the workhorse engine for commercial satellite station keeping and are set to become the engine of choice for going from GTO to GEO.  This means HALL development is a not just for science, it's a commercially supported tech, something that Gridded Ion never had, the Air-force is also naturally interested in anything that can be used in it's satellites to rapidly change orbits for timely reconnaissance.  HALL thrusters have a long heritage and flight history in Russia as well.  A whole range of sizes exist but I'll focus on the largest units available and planned.  The largest thruster used in space so far in the BPT-4000 from ArcJet, the NASA developed 457-Mv2 has been tested and has yet to fly, the X3 a triple nested design is still in early testing and offers a huge improvement in thrust:area density, it would be possible to fit a MW or two of such thrusters onto the butt-end of a conceivable space-craft something not possible with earlier thrusters.

BPT-4000             4.5 kw    1950s
NASA 457Mv2       50 kw     2740s
X3                      200 kw    4,000s

HALL thrusters used to have significant erosion problems that made them only able to handle a modest total propellent throughput, this relegated them to station-keeping.  But these problems have been radically reduced such that newer systems will last years under continual operation and can thus achieve the kind of total impulse necessary for deep space missions.  Also we are about to see a switch from Xenon to Krypton propellents and about 1ks bump in ISP as a result, Kryptons lower atomic mass translates to higher ISP, but it's more costly to ionize.  But as you go to higher ISP ionization is less and less of the total energy expenditure (more is used to accelerate the propellent) so a switch to a lighter propellent has a lower overall efficiency hit then it would at low ISP making it worthwhile.  And Krypton is 1/10 the cost of Xenon to boot.  Newer HALL designs also offer substantial 'throttle' in which ISP is lowered in exchange for higher thrust, 'low gear' is generally around half the ISP and triple the thrust of the efficient 'high gear'.


VASIMIR

I'd like to briefly point out VASIMR and note that while a lot of people talk about VASIMIR as either lovers or haters *cough Zubrin cough* the technology behind VASIMIR was always just ONE concept in a whole pantheon of SEP designs.  And while VASIMIR numbers were really impressive 20 years ago when development began and they were being compared with early <1000s Halls and arc-jets they are now mediocre.  The current proto-type unit is 200 kw and are speculated to be scalable to 1 MW.  It's also aiming for 5,000s which as you can see is basically already done with HALL.  Lastly VASIMIR has an expected thrust:weight ratio at the full MW level that has already been matched by the X3.  The final selling point of variable specific impulse (the VAS in VASIMIR) is likewise nothing special now.  So their are no real selling points to the tech any more and NASA has rightly lost interest, because the tech is both new and won't scale below about 200 kw it has no appeal to satellite designers either commercial or military and as with just about any space-tech if you can't sell it to anyone but NASA it will fail.

ELF

Now if you REALLY want something cutting edge then look at the 'Electrode-less Lorentz Force thrusters'.  This new thruster was only first built in 2009 still has a lot of work needed.  It works by sending out pulsed plasmoids, coherent self rotating doughnuts of plasma, like smoke rings.  Because they are self contained the plasmoids don't have the wide divergence problems or plasma-nozzle detachment losses inherent in HALL or VASIMIR in which some of the thrust is wasted as it spreads at an angle off the desired axis.  Combined with lower collision and ionization losses inside (which simultaneously lead to extremely long life and low waste heat) give the thruster extremely high overall efficiency of >85%.  And ISP is also expected to be in the 5k range using Xenon with much better ability to move into lighter propellents due to the high efficiency.   The thrusters mass:power ratio is expected to be half that of HALL primarily because of a lighters and simpler Power processing unit inherent is pulsed firing, and the thrust density is excellent as well because the device is a narrow tube half the diameter of an the best HALL.  Their are even future plans to mix neutral unionized gas into the plasmoid which would be entrained with it and accelerated without any ionization energy cost, this would basically eliminate the last barrier to utilization of low atomic mass propellents as we no longer lose lose efficiency to ionization.  Scaling up the MW level also looks to be possible.

http://msnwllc.com/Papers/ELF_IEPC-2009-265.pdf

[nadreck]

Thanks for that great summary!

[Burninate]

Much thanks, excellent post.  A few more linked variables that would be of interest: Thrust to weight ratio, thermal efficiency, and thrust to power ratio.  You also mention 'Thrust density'... would this be thrust per cross-sectional area (nozzle area, effectively)?

[Damon Hill]

Separate discussion on the power sources, then?  Persuading nuclear reactors to convert their neutrons to electrons without massive thermal radiators (waste!) has always been a challenge.  As has getting efficiency up and weight down on photovoltaic devices.  Plenty of stuff to discuss here, too.  Just give us a power level to engineer to.

And thanks for the perspective on existing and near-term technology, too.  I knew comsats were going electric in a big way, and was hoping this rising tide would float a lot of deep space probes, too.

[Impaler]

Much thanks, excellent post.  A few more linked variables that would be of interest: Thrust to weight ratio, thermal efficiency, and thrust to power ratio.  You also mention 'Thrust density'... would this be thrust per cross-sectional area (nozzle area, effectively)?

Yes, same terminology as conventional rocket engines, effectively how much thrust per unit area your getting which is really important when it comes to designing a real vehicle rather then just adding up numbers on paper.   Were assuming that the physical size of them is the only real barrier to packing them in clusters and their are no gimbaling needs nor adverse interactions requiring wider spacing.

I'll see if I can find some better thrust/weight ratios, though these are often convoluted as their is a mass for the thruster and a mass for the power-processing equipment, often the later dominates and as we go to higher and higher power system active cooling can come in as another parasitic mass.

Efficiency is generally around 60% for Ion, HALL and VASIMIR systems, only the ELF stood out as particularly high in efficiency.  This efficiency number is electrical:kinetic and all sources of inefficiency are included such as ionization energy (generally the biggest), collision of ions with the side walls of the thruster, waste thermal heat anywhere in the system and off axis cosine lose etc etc.

Thrust to power ratio can be calculated empirically from ISP and efficiency, even for normal rocket engines.  I'll get back to ya with the formula but it is fairly simple if I recall.

Separate discussion on power systems YES please, I don't think I'll be starting that thread though, but I'd certainly be interested in learning more about them and would cross-link too it as a 'sister thread'.

[Robotbeat]

Gridded ion thrusters like NEXT can achieve electrical efficiencies higher than 70%.

[Robotbeat]

Also, while PPSes generally get around 95% efficiency, some can get over 97% efficiency. There are also ways to directly power the thrusters with your solar array so your PPS is basically a bunch of relays and could operate at nearly arbitrarily close to 100% efficiency, so active cooling of the PPS need not be a significant concern. Also, the thrusters' losses are often in ionizing the gases which is nice since that means you don't need to reject that heat through active cooling or somesuch.

[Robotbeat]

In general, "specific power" and Isp range and electrical efficiency are more useful figures of merit than thrust to weight ratio (and it's trivial to find thrust to weight ratio if you know those things, anyway).

[Robotbeat]

Some ion thrusters are capable of 80% efficiency. Usually this is at very high Isp (operating at higher Isp usually makes it easier to get higher efficiency since your ionization loss stays constant even as the kinetic energy per ion increases).

For instance, JIMO's NEXIS thruster was tested at over 75% efficiency when operating at over 7000s Isp.

Similarly, a modification of the NEXT thruster, NEXT STEP, can achieve around 75-78% efficiency at ~4600s

[momerathe]

This is more a general question on mission planning: given that for a fixed power, thrust is inversely proportional to  ISP - how much ISP is too much?

For probes it's not such a big deal, as they'll be thrusting for years regardless. But for squishy humans, you don't want to be spending months spiraling out of Earth's gravity well.

[Burninate]

This is more a general question on mission planning: given that for a fixed power, thrust is inversely proportional to  ISP - how much ISP is too much?

For probes it's not such a big deal, as they'll be thrusting for years regardless. But for squishy humans, you don't want to be spending months spiraling out of Earth's gravity well.

In general, "specific power" and Isp range and electrical efficiency are more useful figures of merit than thrust to weight ratio (and it's trivial to find thrust to weight ratio if you know those things, anyway).

When I was looking at NEXT, I found comparable figures for solar panel mass, thruster mass, and propellant mass for a 1 year burn.  Thrust to weight ratio is absolutely an important metric to compare;  In a short SEP mission (such as anything involving a human or a fickle Congressional funding scheme), solar panels, batteries (surprisingly light for half an orbit's worth), and thrusters need to be a very sizable portion of IMLEO; adding dry mass reduces dV and increases mission duration, partially neutralizing the other criteria.  How much Isp is too much?  That depends on thrust to weight of the propulsion & power system, and mission duration (influenced substantially, actually, by thruster lifetime).  At the extreme, the SEP tugboat concept, particularly for deorbiting space junk of diverse orbital parameters, spends some fraction of its lifetime thrusting at full power while attached to nothing but its propellant tank, and has substantial primary mission-planning decisions made by how many thrusters & solar panels can be packed into a single launch fairing.

[rklaehn]

Per a request in the Mars Colonial Transport thread I'm going to present some of my findings on the current State of the Art in Electric thrusters.

Great summary. Posts like this really add to the quality of the discussion on NSF.

Which of the three concepts would be best suited for use with low molecular mass propellant such as water?

I guess ELF should be very well suited because it potentially can work without ionizing the entire propellant. But the whole concept is very new. Hall effect would also be possible, albeit with reduced efficiency. And gridded ion would be completely out of the question, right?

Do you have any numbers on Isp and efficiency of the various concepts using water or maybe ammonia?

[Impaler]

For mission duration in the 6-8 month range and withing the orbit of Mars (which would be both trips to Mars and to the Near Earth Asteroids) and greater then ~10 tons payload, the preferred EP solution is definitely the latest generation of HALLs like the X3.  Tug concepts from LEO to lunar orbit or Lagrange points could use Ion if your willing to go for slow cargo transfers on the range of a year, but I think HALLs will be used simply because that's the technology that we want to get more experience and validation of right now in large vehicles.

The general vehicle ratio I've seen for HALL based vehicles is around 1/3 hardware (thrusters and power), 1/3 propellent and 1/3 payload.  But improving power density and ISP may push this to around 50% payload and 25% each hardware and propellent.  Beyond that point the additional payload fraction is just not significant enough to be worth the slower transfer when your talking about crew, but it may be desirable for cargo.

Water is certainly an attractive propellent as it's clearly the easiest to extract volatile in asteroids.  The downside is the high ionization cost so it would need to a high ISP system to not be terribly inefficient, also the Oxygen will likely cause serious grid corrosion problems.  My guess is that if we are talking about a small Asteroid visitation probe with then Ion engines are a viable choice but when you go to anything bigger your into the HALL range because you need higher thrust-density.  Halls can be run on just about anything too, even pure oxygen so long as they have some small supply of easily ionized gas to start the ionic-collision chain reaction at the cathode with (it's a bit like starting a cold diesel engine, you need something a higher octane then your normal fuel).  As little as 5% of the total propellent being this low ionization level would be adequate from what I have read.

ELF is certainly very promising when it comes to ISPP, we are very unlikely to be able to efficiently source any propellent elements heavier the argon in-situ.  And while Argon is a good compromise between collect-ability and commonality with existing heavy Nobel-gass propellents their are advantages to going even lower in atomic mass.  Availability of elements increases the lower the mass is and the ISP rises, and if we have the speculated ELF neutral-gas entrainment then these light elements don't cost ionization and don't lower efficiency.  This means we can keep ISP at the ideal range, rather then being forced to higher then desired ISP and slower then desired transit times.

Once we have EP which is compatible with ISPP the combination is incredibly potent.  People have always thought about ISPP in the context of chemical combustion rockets which are so terribly inefficient that they can't generated sufficient DeltaV to even go between the sites of potential resource-extraction without throwing away parts of themselves.  They would then be forced to expend massive amounts of energy to separate oxygen from the collected resource as our solar-system is Oxygen rich and virtually all elements exist in an oxidized form, be they hydrogen or carbon the two elements we would be most likely to use as fuels.  The huge quantity of material needed would demand large processing equipment and high peak power rates if we desired any kind of timely refueling and this power supply would then be idle when actually in transit. 

An Electric propulsion systems on the other hand naturally has a power-source which can be employed for 'free' once the vehicle is stopped at a resource site.  The propellent material extracted dose not need to be chemically altered at the time of collection and the quantities needed are tiny compared to the conventional rocket, all making for a timely process.  The elements collected don't even need to contain any specific oxygen/fuel ratio, and may not even need to be purified, the only thing that will be needed is for them to be liquid between the temperature range in the propellent feed lines and tank, and mixtures tend to have a wider liquid temperature range.  Lastly no Electric propulsion vehicle remotely needs to shed parts of itself to get the DeltaV needed to go between likely resource-sites which actually makes the whole thing viable at all.

It is interesting that while advocates of 'abundant chemical' propulsion freely admit that In-Situ Propellent production is a MUST HAVE for a human mission to Mars.  Their is virtually never discussion of applying the same propellent to an electric system.  If one feels that producing propellents on the Mars surface and using it for assent AND Earth return (as in the original Mars Direct) are viable then it would be more efficient to use that Earth return propellent portion in an electric system once in LMO, thus allowing for more returned mass.  Or in the Semi-direct architecture the ERV which stays in LMO arrives empty and receives return propellent from the assent vehicle.  In both cases the propellent for the electric system could be the conventional Oxygen/Methane mix or if were stingy some of the Nitrogen/Argon waste byproducts from processing the Martian atmosphere could be saved.

[Hanelyp]

... given that for a fixed power, thrust is inversely proportional to  ISP - how much ISP is too much?

To rephrase:  You have a limited mass budget for propulsion.  How much do you allocate for power plant vs. thruster vs. propellant?

But as a rough estimate, exhaust velocity comparable to stage delta-V is generally not a bad balance for overall rocket design.  Manageable mass ratio (but far more than 1/3 of mass as propellant) and not excessive total energy expenditure.

[Nilof]

But as a rough estimate, exhaust velocity comparable to stage delta-V is generally not a bad balance for overall rocket design.  Manageable mass ratio (but far more than 1/3 of mass as propellant) and not excessive total energy expenditure.

Sure, if you are energy-limited rather than propellant-limited. On the other hand, if there is one thing that the inner solar system has an abundance of, that would be solar energy. That tends to shift the optimum towards a higher Isp.


One thing that is useful to mention in the context of the original question is that if you have an engine that allows you to vary your Isp in flight, some maneuvers can also be performed over a longer time without affecting the total travel time and should thus always be done at high Isp low-thrust when possible.

Inclination changes in solar orbit are a good example, and are probably the main reason for developing variable-Isp propulsion systems, since they constitute a big chunk of the delta-v cost of reaching many targets in our solar system such as Mercury, Ceres and Pallas.

[Robotbeat]

Optimum exhaust velocity (i.e. 9.8m/s^2*Isp) if minimizing propulsion system mass (both propellant and dry mass) is usually the characteristic velocity, defined as:

Characteristic velocity = sqrt(2*(system efficiency)*(dry mass total specific power)*(thrusting time))

So if you want to know the optimum Isp, there's a pretty straightforward equation to give you a decent optimum. It's from page 27 of this pdf (which is labeled as slide 21), and the full form of the equation is on page 26 (slide 20):
http://bluebox.ippt.pan.pl/~sbarral/misc/EP_course.pdf

If you want to play around with the full form of the equation which gives you a more refined optimum exhaust velocity, here it is:
http://www.wolframalpha.com/input/?i=maximize{e^(-7/v)-v^2/20^2*(1-e^(-7/v))}

Note that in that example, I used a mission delta-v of 7 and a characteristic velocity of 20. (Units are whatever you want them to be, but I tend to use km/s).

[tea monster]

How much "Ummph" can you get out of one of these designs? How are they getting higher thrust out of these engines and what sort of figures are there?

The big deal with the VASIMIR was the fact that you could kick it into overdrive to get out of orbit before coasting. Every mission I've heard of spends MONTHS spiraling out of orbit before they even start the journey proper to where they are going. For any kind of manned mission, this is - quite literally - 'loopy'.

[Robotbeat]

Read what I just wrote. VASIMR also needs to spiral out, it depends on specific power.

But for a manned mission, the crew would join the spacecraft only after it spiraled out of LEO. Also, in such a situation a very small chemical kick stage helps (the malletEDIT:smaller the kick stage, the larger the Oberth "multiplier" you get).

EDIT:dang auto-incorrect.

[Impaler]

Read what I just wrote. VASIMR also needs to spiral out, it depends on specific power.

But for a manned mission, the crew would join the spacecraft only after it spiraled out of LEO. Also, in such a situation a very small chemical kick stage helps (the mallet the kick stage, the larger the Oberth "multiplier" you get.

Yes, EVERY SEP mission has base-lined this high-earth-orbit rendezvous (which something like an Orion craft would be ideal for) for decades.  I find it very annoying when people grip about the duration of an Earth-escape-spiral and try to say the crew will die in the Van-Allen radiation belt so their-for my-pet-high-thrust-propulsion (chemical or nuclear) is the ONLY solution.

In fact we should be going as SLOW as possible at the highest possible ISP during spiral, because it is a pure Delta-V accumulation, their is no Synod to worry about, no crew on board.  All we need to worry about in space-craft wear-and-tear and if were going on a multi-year mission the vehicle will be expected to last a good long time (particularly if it is reusable) so spending one year or even two in the spiral is very reasonable.  You can conserve your propellent for a faster higher thrust maneuver to go heliocentric and keep crew radiation dosage low when it actually counts.

Interestingly enough the Ad-Astra papers which seek to tout the advantages of the Variable Impulse of their engine show that Variability is actually quite over-blown.  At least the heliocentric portion of a journey the huge range that VASIMIR claims to be able to offers dose very little.  The total mission time reduction for a variable vs constant ISP system was only 10% over a fixed ISP system.  And with most other SEP technologies having a modest range of ISP it is likely that the VASIMIR systems high range would contribute even less in a real world comparison.

[rklaehn]

Read what I just wrote. VASIMR also needs to spiral out, it depends on specific power.

But for a manned mission, the crew would join the spacecraft only after it spiraled out of LEO. Also, in such a situation a very small chemical kick stage helps (the mallet the kick stage, the larger the Oberth "multiplier" you get.

Yes, EVERY SEP mission has base-lined this high-earth-orbit rendezvous (which something like an Orion craft would be ideal for) for decades.  I find it very annoying when people grip about the duration of an Earth-escape-spiral and try to say the crew will die in the Van-Allen radiation belt so their-for my-pet-high-thrust-propulsion (chemical or nuclear) is the ONLY solution.

But what about the spacecraft itself? Wouldn't this also be damaged by repeated the passage through the van allen belts? For example, if you had a spacecraft using thin-film solar cells which offer very high power/weight? (Maybe somewhat similar in design to IKAROS)

Wouldn't the charged particles from the van allen belts drastically reduce the efficiency of the thin film cells by the time you are in EML2?