How brute-efficiency enables practical electric flight

Anytime the topic of electric aircraft comes up, the immediate response by the middle-brow (and some high-brow) is that it’s obviously impractical due to the mass-sensitivity of flight and the obvious energy density (or they usually intend “specific energy”) advantage of gasoline and jet fuel over batteries. And this is not a totally wrong claim.

This Wikipedia page is a convenient reference:

Aviation Gasoline has a specific energy of 44MegaJoules/kilogram given the high heating value of burning gasoline (high heating value includes the heat energy in water vapor, low heating value does not). A nice safe Lithium Iron Phosphate battery might only pack around 123Watt-hours/kg, or 0.44Wh/kg. Literally a factor of 100 difference in pound for pound energy density and therefore range (or, even worse, they’ll tell you, because the fuel weight is lost during flight). So obviously electric flight is a non-starter. But is it?

A typical gasoline engine might consume about 285grams of fuel per kWh of mechanical power. A turboprop like the PT6 might consume 308grams of jet fuel per kWh. Efficiency in the range of 27-28%.
Electric powertrains can achieve, under cruise conditions, efficiencies on the order of 95% (98% discharge efficiency from the battery and 97-98% from the motor, maybe 0.5% loss from the controller, etc… and note that the longer you take to discharge from the battery, the more efficient it is). (Less for very high power take-offs, etc.)
Electric powertrains are therefore a factor of 3 more efficient. Secondarily, a high end electric car battery can achieve about 275Wh/kg or 1MJ/kg. Next-generation chemistries (solid state lithium, lithium-sulfur, lithium metal anode, etc) that are available in sampling scale are 400Wh/kg, with some labscale cells achieving 500Wh/kg or even 650Wh/kg. Lets use the sampling level of 400Wh/kg, or about 1.44MJ/kg as you could actually probably build a large battery out of these cells today if you had enough pull. Now the difference between burning fuel and battery electric is down to a factor of 9.
Now comes the other huge factor, often ignored: Lift to drag ratio.
A Cessna 172 or a Piper Cub J-3 might get a lift to drag ratio of 8. The best sailplanes (Eta, Concordia, and Nixus) can get about 70, so about a factor of 9 better. And similar speeds as the Cessna and Piper Cub, too!

The full equation for an electric aircraft is:

Range = E* (1/g) (overall efficiency) (L/D) (battery weight as percentage of total weight).

So if the specific energy is 0.5MJ/kg, overall efficiency is .95*.85 (95% for battery and motor, 85% for propeller), L/D of 20 (about the same as a modern airliner), and 25% battery weight percentage, g is 9.8m/s^2, and therefore the total equation is:
0.5MJ/kg*(1/(9.8m/s^2))*(0.95*.85)*(20)*(0.25) = 206 kilometers.
A more aggressive aircraft may use 1MJ/kg batteries, a lift to drag ratio of 50, and a battery weight percentage of 35%:
1MJ/kg*(1/(9.8m/s^2))*(0.95.85)*(50)*(0.35) = 1442 kilometers.
Now we are getting somewhere!
A Piper Cub J-3, of which 5500 are in flying order today (and very popular in the Alaska bush), gets a range of about 354 kilometers.

A really aggressive approach would use a lift to drag ratio of 70, 400Wh/kg (1.44MJ/kg) batteries, and a battery weight ratio of 50% would get: 4153 kilometers of range. That’ll get you transatlantic distances without additional cleverness.

Antares 23E has L/D = 60.

Another version of the Antares, the Antares 20 with a lift to drag ratio of 56, was used as the basis for an electric airplane with enough batteries for a 450km range.

Now, all this being said, there’s a huge advantage to, at the end of the day, choosing much lower battery mass and structure mass, or picking cheaper chemistry (lithium iron phosphate instead of state of the art lithium ion) and structural materials (automotive grade carbon fiber or even fiberglass instead of state of the art aerospace carbon fiber) and getting by with a 1000km range instead, or with a much greater payload fraction. But it’s important to know what the limits are.

Posted in Uncategorized | 7 Comments

The Unreasonable Energy Efficiency of Conventional Orbital Rockets, Part One: Versus the F-150

Falcon 9 launch. SpaceX’s Flickr photostream.

Orbital rockets do not *seem* efficient. They take off in what amounts to a mushroom cloud, using extravagant amounts of propellant to rise even the first 100 meters into the air to clear the tower. Yet, their efficiency rivals that of your modern pickup truck. Let me be clear by what I mean by efficiency, here: Energy efficiency, not necessarily mass efficiency. That is usually a strange assumption, but I hope it will make more sense as we go on., DiamondBack Covers CC BY-SA 2.0 DEED Attribution-ShareAlike 2.0 Generic

\eta = \frac{\textup{Orbital Energy}}{\textup{Chemical Energy}}

An efficiency expressed as a percent has a numerator and a denominator. The numerator for orbit is the minimum possible energy that could possibly be required, such that if you did somehow beat it, you could conceivably *harvest* energy. In our cases, it’s the total potential and kinetic energy that the payload needs to be. For orbit, that’s the energy required to lift up a payload to ~100-200km and accelerate it perfectly to orbital velocity. That total is about 32MegaJoules (MJ) per kilogram. (A Joule is 1 Watt for 1 second. 1 kilowatt-hour is 1000*3600 Joules or 3.6MJ, 1055 Joules is equal to a BTU, and 1 Gigajoule is roughly 1 MMBTU.) And the denominator is the chemical (or any other kind) of input energy (and we will use the High Heating Value of any fuel, i.e. all the heat generated, including that it the water vapor).

Take the 2021 F-150, for instance. It has a payload capacity of 1,765 lbs and a curb weight of 4,705 lbs. ( It has a combined EPA fuel efficiency of about 20mpg (presumably without a full payload, but for our purposes, we’ll assume a full payload). The “Ecoboost” engine by FORD has a peak efficiency of about 33% (, and there’s about a 20% loss of power through the transmission for an automatic transmission to the rear wheels (F150 Forum). The maximum freeway grade is 6% (which is to say, a rise of 6 meters for every 100 meters traveled). So if you were to carry a full payload of dirt up a height of 6 miles, it’d take about 100 miles horizontal distance. The total energy in this case is not just 100 miles divided by 20mpg, because we’re also traveling up in altitude, so we have to take into account the mass of the vehicle plus payload and the efficiency of the engine and drivetrain. As before, the engine is 33% efficient and the 20% losses (80% efficiency) mean a total of .8*.33 =.264 or 26.4% efficient. However, your payload is only a fraction of the total weight! Curb weight plus max payload gives 1765lbs+4705lbs = 6470lbs. Payload is just 27.3% of the total mass. So the overall efficiency is .273*.264= 7.2%. 1765lbs is 803kg, times gravity (9.8m/s^2) is 7867.6 Newtons. 6 miles is 9,656 meters, so multiplying these two together gives us the potential energy of our payload (and usually we want it stationary up there, so no kinetic energy component). 7867.6N*9,656m =75970125 Newton-meters (which is the same as Joules). 76MJ potential energy. As we established earlier, the efficiency of the engine, drivetrain, and the payload mass fraction (we can ignore mass of fuel for this short distance) is 76MJ/0.072072 = 1054MJ. It took 1054MJ in gasoline to go up 6 miles. Creative Commons Attribution Share alike

But wait! We also didn’t include the energy losses of traveling that 100 miles horizontal distance. To know that, we need to know the energy content of the fuel (as cruise efficiency is given in miles per gallon). That is typically 46.4MJ per kg, with fuel having a density of 0.737kg per liter, or 129.4 MJ / US gallon. 100 miles horizontal distance at 20mpg needs 5 gallons, which at 129.4MJ/gallon, means 647.25MJ in chemical energy in horizontal travel. (This is a restriction of road vehicles… they don’t travel straight up! That reduces achievable efficiency, although perhaps we could improve things with a steeper grade, at the expense of more complication in payload considerations.) The total energy needed is therefore 1055MJ+647.25MJ = 1702MJ in gasoline form. The total efficiency is therefore just: 76MJ/(1702MJ) = 4.5%.

This is for a modern vehicle using its max built-in payload capacity at the typical max freeway grade (other roads can do 7-10% grade or even higher, but there are other issues with such aggressive grades that could violate other assumptions).

How does this compare to a rocket?

To make things simple, we’ll use Falcon 9. I’m using Ed Kyle’s wonderful website (may it rest in peace!) which has some pretty decent estimates for propellant masses for both stages of Falcon 9:
SpaceLaunchReport (

For “Falcon 9 Block 5 Merlin 1D Fuller Thrust,” He estimates 418.7tonnes (t) of propellant for the first stage and 111.5t of prop for the second stage (total of 111.5t+418.7t=530.2t). The quoted expendable payload is 22.8t, so each kg to orbit requires 23.25kg of propellant. Merlin 1D has an oxygen:fuel mass ratio of 2.36 (Merlin Evolution), reducing that 23.25kg down by a factor of (1+2.36)=3.36 to 6.92kgRP-1/kgLEO. RP-1 fuel (kerosene) has an energy density of 43MJ/kg, so Falcon 9 needs just 297.6MJ per kg launched to LEO, expendably. Note that 16.33kg of liquid oxygen is also needed, and it takes about 200kWh per tonne of cryogenic liquid oxygen at modern cryogenic separation plants, or 11.8MJ additional energy per kg to orbit. This is only ~3% of the fuel energy, but we include it here anyway. (Not including cryogenic chilling of the kerosene at this time, but that should be much less than the liquid oxygen, so a rounding error here… although the LOx also takes some energy to subchill as well.) So a total of 309MJ/kgLEO, so given 32MJ theoretical total orbital energy from the ground, the Falcon 9’s energy efficiency in expendable mode is 32MJ/(309MJ) = 10.3%. Over double our maxed out pickup truck!

This isn’t a hard limit, either. Merlin is a gas generator rocket, much less efficient than a staged combustion engine. It also operates fuel-rich, which wastes chemical energy. An ox-rich staged combustion kerolox engine should do even better. In principle, you can improve the efficiency of a rocket by blending in water in the beginning part of the first stage burn and using a hydrolox upper stage (especially if near-stoichiometric or ox-rich). Higher pressure (i.e. from staged combustion) and altitude compensation could further improve efficiency, as could structural mass improvements. Operating with 3 stages instead of 2 could also in principle help, staging off dry mass even earlier.

But how could a rocket be so efficient?

It’s worth looking at the ratio of the upper stage dry mass to the payload mass. In the case of our truck, the payload was just 27.3% of the “burnout” mass, but for Falcon 9, the upper stage burnout mass would be just 4.5t or so, therefore the 22.8t payload would be about 83.5% of the burnout mass, 3 times better than the pickup truck. This is enabled by staging. A single stage rocket would not have this fabulous burnout mass efficiency.

Second, rockets operate at extreme pressures, comparable to a race car’s supercharged engine in the case of Merlin to double that pressure for Raptor. AND rockets expand from that extremely high pressure to vacuum. Those two things combined allow higher thermodynamic efficiency than a car engine’s 33%. Greater than 50% thermodynamic efficiency is possible. Plus, rockets have sufficient thrust to weight ratio to accelerate (initially) straight up and quickly out of the drag-causing atmosphere whereas trucks have to deal with rolling resistance and air resistance throughout the whole trip. All these things together mean that in spite of the fact that rockets have to carry their oxidizer with them and in spite of the rocket equation and the fact that rockets can only push on their own exhaust (and not the whole earth like a wheeled vehicle), a two-stage modern chemical rocket can beat a modern pickup truck in load moving efficiency in terms payload final total (potential+kinetic) energy.

Granted, this is for expendable mode. However, a droneship recovery is now about 80% (18.4t) of that previous 22.8t payload. And even doubling the mass of the upper stage to allow recovery would still leave you at 60% original efficiency, or about 6.3% (~500MJ/kgLEO) for full reuse, still greater than the pickup truck. (Note that my calculations for estimated Starship efficiency give a similar 500MJ/kg, in spite of the higher performance Raptor, and I think that’s due to how non-optimally heavy the Starship upper stage is compared to the first stage… in part because it brings the whole fairing to orbit.)

This all might seem unfair. After all, I’m just counting the altitude, not the distance traveled when making this efficiency calculation. But this might betray a lack of appreciation for orbit: in space, there’s no air resistance to speak of. Kinetic and potential energy are, for most intents and purposes (outside of very low earth orbit), conserved. On Earth’s surface, there is friction everywhere and travel anywhere requires energy expenditure even without a net change in altitude. So rocket travel in space is, if anything, even more efficient than I express here. However, there’s a big entrance fee required: orbital velocity. Rockets don’t require massive energy because they’re efficient. That is largely just the cost admission. Orbit is just fundamentally high-energy, and engineers have had to develop some of the most efficient and remarkable machines ever devised to access this realm.

This energy efficiency comparison applies to aircraft as well, and even to other orbital launch concepts. Single Stage vehicles, regardless of propulsion method, will tend to do worse due to the greater dry mass. And high Isp propulsion used for the first stage is wasted energy. Perhaps I’ll do my next post comparing to Skylon. This topic is a continuation of some of my previous posts on fundamental rocket efficiency considerations.

But let’s leave you with one last comparison: Fuel efficiency of an A380 jumbo jet is about 72 miles per gallon on long haul flights (Wiki: Fuel economy in aircraft). Given the density and specific energy of jet fuel, the energy needed to transport a 160 pound adult halfway across the planet on an A380 (in a seat, stopping to refuel) is the same as launching them to orbit on an expendable Falcon 9 (without a seat, no refueling). So rocket travel is not *inherently* far more energy-intensive than long-haul air travel. It’s the same order of magnitude.

Posted in Uncategorized | 6 Comments

SPS in the Van Allens

There are some of us that don’t see the business case as stated (millions of tons per year) for the SpaceX Starship. The size of the vehicle has led to a number of problem with permitting, ground support equipment, and complexity of the largest spaceflight vehicle to date. Short version is that development is longer, riskier, and more expensive than with a smaller vehicle using similar technology. It is well oversized for the missions claimed for Starlink as simple math gives under 800 total flights for the heavyweight Starlink2 constellation. Mars colonization doesn’t seem to have a visible investment strategy other than Elon impoverishing himself along with everything he can obtain from government and investors. The known Artemis and other missions are well under the capabilities claimed for Starship. The thousands of annual flights at 100-200 tons per flight need another market. Otherwise a much smaller, simpler vehicle with less development investment and risk could do the known projects, and could have been in revenue service years ago.

Assuming that there is an unstated market for the projected capabilities of the Starship/Superheavy combo that is being kept on the down low for the moment, what could it be? Orbital tourism will very likely expand considerably over the next decades. Many of us think it possible that the ramp up will be slow enough and safety/regulatory concerns high enough to extend that market beyond the nearer term profits necessary to get an appropriate ROI.

A recent discussion brought up another possibility that may fit the hardware development scenario currently unfolding. That is Solar Power Satellites in an orbit low enough to vastly reduce the financial barriers to entry compared to SPS in GEO. The orbit also needs to be high enough to avoid most of the orbital debris and working satellites in LEO. That orbit appears to be in the lower Van Allen belts. About 80% sunlight in most of these orbits of interest. Somewhere around 3,000-4,000 kilometers up, the radiation deters most responsible satellite operators in addition to the extra distance from the surface. The orbit being higher also degrades surface imaging to some degree. A bit of extra latency for communications is also unwanted.

If an SPS can operate at those altitudes, it has a serious effect on the assumptions normally associated with the concept. The primary effect is that beaming power from a maximum of 15% of the distance to GEO reduces minimum antenna and rectenna sizes to 2.25% (15%^2=2.25%) of the area of those in GEO. Minimum size of beaming antenna 100 meters diameter and minimum rectenna being under a kilometer in diameter. Also by being in a relatively low orbit suggests that components can be carried up from Earth without first building a Lunar and orbital industry.

There are serious objections to the concept starting with the radiation environment. There were two suggestions concerning this. One was that the units take the mass hit to be radiation hardened since Starship is going to carry hundreds of tons per flight for costs lower that of the Falcon9 or any other rocket in the world. The other suggestion was that Tethers Unlimited had a viable concept for draining the radiation from the Van Allen belts. In either case, the mass lift capabilities of the Starship come into play.

Second objection was that the solar collectors and beams would be moving all over the sky with variable distances and angles to the rectennas on the ground. Electronic steering of radar and communications antennas has apparently taken care of that issue. Starlink being quoted as case in point. Various satellites would shift beams during the orbit such that they always had a receiver available. For the limited time they didn’t, and the limited times that the rectennas weren’t receiving, there is the battery tech from Tesla.

Third was the millions of tons of CO2 produced by tens of thousands of annual launches by this huge system. The counter to that is that the clean electricity delivered would more than offset the launch CO2 in the near term and almost eliminate fossil fuel pollution in the long term.

Fourth was that this was a high risk vehicle for the foreseeable future. As in Starlink, mass produced components should be inexpensive enough to risk the occasional bad day. The SPS could create the demand for launch after Starlink2 is up and running in profitable service.

My reservations about Starship haven’t gone away. This is just trying to see what reasonable justification there could be for a launcher to be built by the thousands and each vehicle launching daily with a hundred or more tons of payload. This post is speculation. It is based on skepticism about the purported markets and an expectation that there is a rational reason for going this big this fast.

Posted in Uncategorized | 41 Comments

NASA’s Selection of the Blue Moon Lander for Artemis V

Last week, when NASA announced that they were picking Blue Origin’s National Team to develop a sustainable human lander for the Artemis V mission, what surprised me wasn’t the selection, but the fact that I’ve come around to really liking the decision.

NASA’s Associate Admin for Exploration Systems Development, Jim Free, at the Artemis V lander selection press conference (Credit: NASA/Aubrey Gemignani)

While it’s still recent enough to be relevant, I wanted to share some thoughts on Blue Origin’s updated human lander architecture, why I think this was the right selection in spite of my feelings about their original concept, thoughts on the execution challenges they’ll face, and some of the interesting future possibilities having two fully-reusable lander architectures may open up for NASA. But first, you may be wondering why I was so surprised that I would end up liking Blue Origin’s lander architecture.

Why I Wasn’t A Fan of Blue Moon 1.01

I hope I don’t offend any of my friends who work or worked at Blue Origin by saying this, but if you had to summarize my initial reaction to the National Team’s original HLS lander concept, I would’ve used the word cynical.

It felt cynical, because rather than trying to come up with the best solution for affordably and reliably bring people to/from the lunar surface, they seemed instead to be regurgitating exactly what they thought NASA wanted to see. NASA had a reference mission concept that was a complex, fully expendable system with three stages — an in-space tug, a descent element, and a separate ascent element, so that’s what they had. Blue Origin was a relatively unproven space contractor, so they added not one, but two aerospace primes to their team. Which also happened to maximize the number of congressional districts their project would have work performed in2. It almost felt like they weren’t even trying to win, so much as guarantee that they’d be one of the two solutions picked3. While in theory at least some of the elements in Blue Moon 1.0 could be refueled and reused, some components like the massive lander descent stage had no easy path to future reuse.

The National Team’s original Blue Moon HLS concept, as proposed in 2020 (Credit: Blue Origin)

In contrast, the Dynetics ALPACA design was a creative approach that seemed to be genuinely trying to provide NASA with a good way of getting people and cargo to and from the lunar surface. By the end of the base period, they had shifted to a single-stage lunar lander concept that leveraged in-space refueling, and had a clear pathway to full reuse. The low-slung central crew/cargo attachment point allowed easily delivering crew to/from the lunar surface as well as delivering large cargo modules, without needing multi-story ladders or elevators. The low CG meant that it could likely land on rough terrain with a lower odds of tipping than a design like Starship HLS. It did have the teensy problem that at the time proposals had to be submitted, the design’s mass budget didn’t close yet4, but they did eventually close the design, just not in time for consideration in the Option A evaluation.

Autonomous Logistics Platform for All-Moon Cargo Access (ALPACA) lander concept from 2020 (Credit: Dynetics)

Anyhow, suffice it to say, that going into last week’s HLS Sustainable Lander Development announcement, I was really rooting for Dynetics to win, and didn’t have a very high opinion of Blue Origin’s lander concept. So, when Administrator Nelson announced the National Team had been selected, my first reaction was pretty strong disappointment.

I’m glad I was too busy taking notes from work to tweet my immediate hot take, because by the end of the call, my opinion had shifted pretty dramatically.

Why Blue Moon 2.05 is a Dramatic Improvement Over v1.0

It took the press conference a while to show any details about the new design, but when they unveiled the Blue Moon 2.0 lander, I almost did a double take. The design was superficially similar to the original design, but you pretty quickly noticed some pretty significant differences. Was I seriously seeing a bottom-loader single stage design?

Meet the Blue Moon 2.0 lander concept (Credit Blue Origin)

I never got around to blogging about what I call bottom-loader SSTO landers, but it’s an idea I first learned about almost 20 years ago with t/Space’s CXV Stage 2 concept from their Concept Exploration & Refinement study final report6. It’s one of my three favorite Unorthodox Reusable Lunar Lander Concepts that I’ll hopefully get more chance to blog about in the future. Needless to say, when I saw that, I perked up and started paying real attention.

If I had to summarize the highlights we could glean of the proposed Blue Moon 2.0 lander architecture, I’d point out three key features:

  • Bottom-Loader SSTO Lander: Crew or cargo pod on the bottom, propellant tanks on top. Enables easy surface access without cranes or ladders. Keeps the CG low for reduced tipping risks. Keeps the load paths for the rocket higher efficiency. Provides the best thermal isolation between the warm parts7 and the parts that want to be kept really, really cold8. This is the part that Blue Origin would be developing
  • Reusable Cis-Lunar Refueling Tug: While they never showed any pictures of the tug, this element would bring LOX and LH2 propellant from LEO to NRHO to refuel the lander, and return to be refueled again for reuse. This is the part that Lockheed Martin would be developing.
  • Reusable From the Start and ISRU Compatible: By going to a single-stage architecture, there’s a clear and easy path forward to refueling — initially in NRHO using tugs coming from LEO, but eventually also on the lunar surface9. Also, while LH2 is harder to handle than Methane, LOX/LH2 can be derived readily from lunar water ice sources10, enabling a switch over from terrestrial to lunar sourcing once ISRU is proven out/debugged/scaled up.

In short, Blue Origin responded to their Option A loss in 2021 by significantly improving their offering to NASA, offering a solution that was innovative and actually worth funding.

Why I think Blue Moon 2.0 Was the Right Call

I’m still a fan of Dynetics’ ALPACA and LLAMA concepts, and I hope they find some way to see the light of day. But given what we know now, I think the Blue Moon 2.0 concept was the right call for NASA, and not a politically-motivated decision, or one that only won because a space billionaire bought his way to success.

First, and most importantly, I think Blue Moon 2.0 helped close the innovation gap between the National Team and Dynetics. Blue Moon 2.0 captures many of the benefits that ALPACA brought to the table, including: lower CG for better landing on uneven terrain, crew/cargo located close to the ground for easy ingress/egress and loading/unloading, ability to deliver significant cargo mass to the surface, and a single-stage design with a clear path to full-reusability. In some ways it was better than ALPACA, by providing a cleaner load path and easier thermal isolation of cryogenic tanks from heat sources, a propellant combo that had an easier path to 100% sourcing from lunar ISRU, and a more developed fully-reusable cislunar tanker concept11. There were some relative drawbacks like the challenges of LH2 storage, and the potentially smaller available cargo volume12, but overall they did a good job of narrowing or closing the gap with Dynetics’ solution.

Second, there seemed to be far less zip-code engineering this time around13. There are multiple team members still, but each of them makes logical sense, not just as a way to get more congressional support.

Third, while there’s very real execution risk for Blue, since they haven’t flown anything anywhere near this complex, Dynetics carries similar risk, so it’s not really a discriminator.

Fourth, while Bezos’s willingness to subsidize the price to NASA probably made a difference, it was far from the only consideration, and in my opinion was probably more of icing on the cake. In addition to closing the innovation gap with Dynetics, it sounds like Blue did a better job of convincing NASA that they had a design that unambiguously closed technically. If Dynetics had still had the clearly superior concept, and if they had done a better job of making it unambiguous that they had a design that closed for all of NASA’s needs, I think they would’ve had a decent chance of winning, even with being more expensive to NASA.

In the end, for all of these reasons, I think NASA made the right call. That said, while I doubt it will happen, I hope Dynetics finds some way to get their concept fielded14.

But Can Blue Deliver?

This is where I have the strongest reservations. While Blue has now laid out an innovative and exciting architecture that’s worthy of being funded, a concept is only as good as the organization tasked with executing it. And frankly, people have reasons to have reservations about Blue Origin’s ability to execute on a project this complex. Whether you look at how late the BE-4 engines were in development, how long it took New Shepard to transition from flight test into operations, or how long New Glenn has been taking to make visible progress, there’s definitely room to worry that Blue sometimes take the Graditim part of its slogan more seriously than the Ferociter part. One thing Blue Origin has made me realize is that while I’ve had too much experience with having too little money, that there are real risks in having too much money, that has too few requirements for demonstrated traction tied to it.

It’s an open question if Blue can change its company culture and processes quickly enough to be able to deliver on an ambitious project like this on a tight schedule. I hope they can succeed at that evolution though, because if both the National Team and SpaceX are successful, it could lead to a very exciting new world.

What If They Are Successful?

If both SpaceX and the Blue Origin National Team are successful, we enter a really interesting world. As Eric Berger pointed out in this Ars Technica article today that I was quoted in, both architectures are now based solidly on the use of reusable launch, in-space cryogenic storage and transfer, and in-space reuse. As I pointed out in the intro to my unfinished series on Unorthodox Reusable Lunar Lander Concepts, a fully-reusable lander architecture brings a lot of advantages, beyond the obvious ones of cost savings:

  • Lower Marginal Costs: While you’ll still have some fixed costs associated with the lander infrastructure, the marginal cost of such an architecture drops dramatically, since you’re not having to build new lander or in-space tug hardware for every mission.
  • Throttleability: Once you have a stable of multiple reusable landers, where there aren’t any major expendable components, it becomes a lot easier to throttle up or down mission tempos based on budget availability. If you have a year or two that you need more money to fund say Mars system development, you can throttle down to a lower ops tempo without risking losing the capability, unlike what happened during Apollo.
  • Easier International Involvement: While Starship and New Glenn should theoretically be cheaper than any other launch source, if NASA is paying for those launches, it’s still a cost. But with a distributed lift/tanker architecture, it becomes more feasible to allow international partners to contribute propellant or crew or cargo launches to LEO as their part of the mission. Even if their rockets are more expensive, if NASA isn’t having to pay for those launches, it lowers the cost to NASA.

In addition to those benefits, a fully-reusable Cislunar tug, like what LM is proposing as their part in the Blue Moon 2.0 architecture, opens up some very interesting possibilities. Once you have a reliable way of getting from LEO to NRHO and back reusably with propellant, it’s a relatively straightforward upgrade to add the ability to ferry crew and/or cargo instead of or in addition to propellant. And since Blue will have already developed a crew cabin that’s safe for up to 30 days on the lunar surface, using a derivative of that as a crew pod on the reusable Cislunar Tug isn’t a crazy option. We don’t have hardly any details on LM’s concept, so there’s a chance they might have something in mind that wouldn’t be able to do the LEO-NRHO-LEO loop with crew or cargo, but most of the most likely options should be fairly straightforward to do that.

Once you have the ability to move crew, cargo, and propellants around from LEO to NRHO with a fully reusable system, do you really need SLS and Orion anymore? The vast majority of the budget being attributed to Artemis was the development and operation of SLS and Orion, but they’re only really capable of one mission per year. If you replaced them with distributed lift and reusable Cislunar tugs for crew/cargo out to NRHO, you could probably enable upping the lunar mission tempo dramatically, while freeing up money for developing lunar surface habitation and ISRU payloads. It’s still a longshot politically, but if SpaceX and the National Team are successful, we could be living in very interesting times.

Posted in Blue Origin, Commercial Space, Lunar Exploration and Development, Propellant Depots, Reusable Lunar Landers, Space Transportation, SpaceX | Tagged , , , , , | 10 Comments

Regenerative cooled turbopump

This started as a short thought on nasaspaceflight and grew into something that resembled a blog post. So I copied and pasted it here. I screwed up the link but this is an ongoing thought from the turbine in chamber stuff from a decade and a half back. November 10 2008 Performance Monoprop being one of the series. I recently realized that the concept could serve as a pressure booster in a gas generator cycle to get performance almost matching that of staged combustion with a higher thrust to weight ratio and much faster and cheaper development.

I seem to be drifting back to an ability to do some simple prototyping and am interested in finding an engine company that might do a bit of business with an inventor.

The provenance of the regenerative cooled turbopump includes Rotary Rocket, ATREX, and a patented “single rotor turbine” from LANL.

Rotary Rocket introduced a concept to avoid turbopumps (kinda) by putting the thrust chambers on the ends of rotating arms. The canted thrust chambers spun the assembly which pressurized the propellants in the feed pipes with the centrifugal force. There was no central drive shaft as the assembly freewheeled. It eliminated the separate gas generator, turbine, drive shaft, impeller, and all of their housings and accessories. The engine was never completed beyond some subscale testing. Which makes sense when one thinks of the problems of getting 72 thrust chambers to operate properly in the high gee field at the ends of those arms.  Just the weight of all those engines rotating at high speeds should make one nervous. Still, the idea of simplifying the engine process is appealing.

The ATREX was a Japanese hydrogen fueled air-turborocket with tip turbine blades mounted to the outside of the compressor. It was claimed to be lightweight and compact. Whether all the claimes made sense or not I don’t know. But putting tip turbine blades on the Roton structure seemed to be an easier path than the mass of  whirling engines. It could be a much smaller structure delivering propellants into a thrust chamber bolted onto the vehicle in the normal way. Then the gas  in the thrust chamber could drive the blades on the way through the throat. Seemed simpler, but exposes the turbine blades to more heat than even regeneratively cooled blades would be able to stand. Also, it leaves no path for cooling the nozzle and thrust chamber.

The Single Rotor Turbine patented by LANL has the hollow turbine blades as the last stage of the compressor. The air flows through the inside of the blades as coolant while being accelerated and compressed. Then the air flows out of the blades into a volute for pressure recovery before entering the combustion chamber. In this way, all of the air is used as turbine coolant before all of it is burned in the combustion chamber and used to drive the turbine. The goal being to raise the allowable turbine inlet temperature for a more efficient engine. Primary direction seemed  to be power generation. There seems to quite a bit of prior work with ideas of this nature as several previous patents are referenced.

Liquid propellants by nature have about three orders of magnitude more mass for a given volume than gas. As such, there is an enormous amount of coolant available inside the turbine blades limited by the requirement that the propellants exit the turbine blades as liquid. It also takes far less energy and velocity to pressurize a given amount of liquid than it does to pressurize gas. It seems apparent that using the LANL concept applied to liquid propellant would result in a relatively simple high pressure turbopump.

In the last post on the regenertive cooled turbine, I skipped over the pump characteristics suggesting using it in a fairly normal layout that happened to allow much higher turbine inlet temperatures. This actually is better as a stand alone turbopump as the driveshaft, housings, and torques involved would be just as heavy as the standard units for a fairly modest gain in capabilities at the expense of R and D on a new system, not to mention the uncertainty. Also, it would almost certainly be tasked to a staged combustion engine which is one of the most expensive and difficult engines to develop.

What I am going to suggest here is more modest in some ways and more radical in others. Develop a small turbopump and gas generator of this nature that free wheels on its’ shaft in the same manner as the Roton engine. Without having the stresses of torque through the turbine blades and disk, torsion in the shaft, and torque limitations in the impellers, the single rotor turbopump can operate at much higher speeds than any normal layout. The turbine inlet temperature can be much higher than any normal turbine even while the disk and blades are much cooler. The torque stresses are minimal with the drive gas on one side of the turbine blade driving against the liquid propellants on the other side of one thickness of metal. The centrifugal stresses will be the same for a given rpm and radius.

With the “brakes off” as compared to normal systems, the tip velocities of the turbine/impeller combo can reach speeds normally reserved for turbo compressors. Tip velocities creating velocity head, and velocity head going as the square of velocity, 40,000 feet of head pressure is theoretically achievable with tip velocities of 1,600 fps.  LOX to over 19,000 psi and RP to over 14,000 psi is theoretically achievable. Cutting those in half with decent pressure recovery might give something close to reality. Using the RP at 7,000 psi as pressure in the gas generator and running as hot as the turbine blades allow should bring the turbine pressure drop to a very modest value allowing 5,000+ psi at turbine exit.

The interesting thing about the single rotor turbopump is that it doesn’t gain weight at the same rate as a conventional system. One that would run all the RP and much of the LOX, or vice versa, in a Merlin would be a unit you could pick up in one hand and not strain. But that would be several bridges too far. Especially as on of the weaknesses of this system is that it gives no reasonable path for regenerative cooling of the thrust chamber and nozzle.

I suggest it might be worthwhile to use such a unit in place of the gas generator on a gas generator cycle engine. Gas generators seem out of style at the moment with so many going for the staged combustion and full flow staged combustion. The reasons are the improved performance compared to the gas generator due to the gas generator exhaust being both lower temperature and lower velocity than the main flow. That percentage of propellant “not pulling its’ weight” is a potent argument. I suggest that replacing the passive gas generator (illustration 2-20 Huzel and Huang page 45) with a very small active gas generator with the single rotor turbopump could boost gas pressure to the standard turbine by 5,000 psi. This could allow either much less propellant to run the standard turbine which would boost system Isp. Or it could have and exhaust of 1,000 psi into a secondary combustion chamber with additional oxidizer creating a respectable Isp in itself. If the gas generator is using 10% of the total and getting 320 seconds in vacuum and the main chamber is getting 360, then the system would only be about 3-4 seconds below that obtained with staged combustion at similar pressures. Considering that the gas generator cycle has inherently better thrust to weight and is easier to develop, a new conservatively designed gas generator engine might be a contender against the state of the art stuff being developed now.

And then there is the option of using them as boost injectors. Initially one. Then later possibly 19 in the manner that the V2/A4 engine recycled the burner cups from a previous.

Pages: [1]   Go Upd

Posted in Uncategorized | 9 Comments

Second Guessing Starship

There seem to be two extremes when it comes to Elon Musk and SpaceX. There are those that believe everything Elon tries will work because–ELON. There are those that believe SpaceX has reached the end of its’ lucky tether because–ELON. The extreme factions make it a bit more difficult for those of us that try to think, or second guess , what will unfold in the SpaceX universe.

Starship typifies the extremes. Some believe that Starship will be in full service within a year delivering 100+ ton payloads multiple times per day at a cost per flight well under even Falcon9. There are others that don’t see Starship becoming operational in this decade. Again the extreme factions throw a bit more murk out there to peer through.

There are still quite a few of us somewhere in the middle that doubt both extremes. For my part, I suspect it will become operational mid decade at a price point above Falcon9 per launch, though well under it per unit of mass. For heavy lift, SLS isn’t even on the radar of those that wish to open up space as a place for humanity to live, work, and grow.

Where I second guess Starship and Superheavy is on the sequence used in the development path. I felt there was a good bit of hubris in going for the largest launch vehicle in history with several concurrent new technologies in one jump. I did not strongly hold these opinions until seeing the difficulties with ground support equipment, regulatory approvals, and ersatz-environmentalist tantrums. Hence Second Guessing. This is not a recommendation going forward, rather a hindsight thought.

Considering that many of the issues are caused by the sheer size of the vehicle, it is worth considering what could have been done with a more modest precursor vehicle. For this post, I am suggesting a Falcon9 type layout, though that may not be the best possible size or configuration. 9 Raptor engines first stage and one Raptor vac on the expended second stage.

(simplifying) Assuming that Raptor has twice the thrust and a higher Isp than the Merlin, the Falcon/Raptor (FR) should have a bit over twice the payload of the Falcon9. Assuming that Elon and company have accurately forecast that Starship will cost less per launch than Falcon, then the configuration here should have a launch cost about 25% that of Falcon9. Assuming that the 3 launches per day per Superheavy are not a total fantasy, then the FR booster should be able to hit a cadence of once per day with RTLS.

Assuming that had been done, what would it have gained? For starters, the Starship test flights last year would have been first stage FR test flights. Those test flights could have been followed by full up test flights by the end of last year with operational missions starting early this year. By now it seems reasonable that launch cadence could be approaching that of the Falcon9.

Construction of (hand waving) 5 meter diameter first stages would take about half the time and materials per unit of height as either Starship or the Superheavy boosters. Remember the problems getting the welds right early on? Learning curve (time) could possibly been halved. Along with smaller faster ground equipment construction with about a quarter of the propellant volume requirements per test. Problems could have been found and addressed even sooner than the current gargantuan effort.

Regulators seem to be having problems with the size of the Superheavy Starship combo. Kilotons of equivalent explosive seems to come up on a regular basis. Decibels created by the largest launch vehicle in history gets some of them going. It seems possible that a FR could be passed off as the next logical Falcon upgrade. Somewhat larger to accommodate the more environmentally friendly fuels. A diameter and height that could (theoretically) use the same launch pads as Falcon9. The higher payload is just a byproduct of more efficient engines, nothing to worry about. Pads at the cape and Vandenberg could be modified relatively quickly to handle both types of Falcon

Environmentalists, real and ersatz, would have a much harder time fighting a vehicle that was just a bit larger than what was in the original EIS, really just a minor upgrade in the scheme of things. With no immediate need for massive expansion due to the smaller vehicle, there would have been far less they could try to block. Everything necessary being permitted and underway before drawing their serious attention.

If the Superheavy/Starship combo hits full operational status shortly, then most everything in this post can be used to ridicule me. If the costs per launch are in the single digit millions within a couple of years, the ridicule can be redoubled. I would accept that ridicule with a smile, and possibly a few belly laughs. I do not expect it to just coast to a smooth operational status in a short time frame. If I am right, then a smaller ship as the FR could be finding the problems with operating an extreme performance methane engine launch vehicle at a quarter of the engine/airframe cost per boom/lesson. And those lessons could have been accumulating starting about a year ago. This is not casting shade on SpaceX, this is about biting off a huge chunk that may be too much to swallow easily for anybody.

It seems that the strongest argument has to do with Starlink2. If FR were operational by now, it could be launching 100 or so Starlink1 per flight or ~50 of the Starlink2 as they come on line. Assuming that Superheavy/Starship could operate for the same cost per launch as Falcon9, then this purported FR should cost about 25% of that. So Starlink1 would be hitting orbit at roughly 12-13% of the launch cost per sat as those launched on Falcon9. Starlink2 at a little over twice the mass would cost perhaps 30% of what it costs to do a Starlink1 with Falcon9. If Superheavy/Starship has a suggested turn around of up to 3 times per day, then it seems that an FR could hit once per day even while ironing out operational bugs. One ship flying daily with 40 Starlink2 on board would be placing 200 of them per 5 day week. That would be 10,000 Starlink2 per airframe per year. With the current build rate at Boca, there would be no need to rely on one airframe. Perhaps three each at Boca, the Cape, and Vandenberg, with Wallops and Kodiak if needed.

Posted in Uncategorized | 21 Comments

Fill ‘er Up: New AIAA Aerospace America Article on Propellant Depots

I was interviewed by John Kelvey for an article on propellant depots, which AIAA just published in this month’s Aerospace America magazine. Here’s an online version of the article for those of you who don’t get the print edition. Also interviewed were my friends Laura Forczyk from Astralytical, Jeremy Schiel from Orbitfab, and Bill Notardonato from Eta Space.

Posted in Propellant Depots, Space Policy, Space Transportation | Leave a comment

Independent Perspectives on Cislunar Depotization

I was invited to participate in a workshop this week in Chantilly, VA focused on cislunar propellant depots. As part of the preparation for the workshop, they asked for stakeholder to provide their perspectives on depot architectures, the benefits of depots, and the technology, market, and policy changes needed to enable them. Based on the inputs I provided, they asked me to share my thoughts with the group.

Over on my Starbright Engineering blog, I wrote a summary of key ideas I think I brought to the table, and key lessons learned:

Posted in Lunar Exploration and Development, Propellant Depots, Space Development, Space Transportation | Leave a comment

Starbright Response to ISAM National Strategy RFC

As I mentioned in my previous post, I’m going to be providing links here to relevant blog posts on my Starbright Engineering LLC blog. This first link is to a comment I submitted to the Office of Science and Technology Policy regarding their In-space Servicing, Assembly, and Manufacturing National Strategy that they released back in April.

Posted in Satellite Servicing, Space Policy, Starbright | Leave a comment

Personal Update: Starbright Engineering LLC

So, last month I did a thing, and left Voyager to strike out on my own. If you follow me on Twitter, or LinkedIn, or Facebook, you probably already saw. For the near-term, I’ve founded Starbright Engineering LLC as a single-person part-time aerospace/startup consulting business. I’m using Starbright to help pay the bills while I explore other options for my next startup.

I have one idea that I’m exploring with a friend, which is a swing-for-the-fences concept that would take me away from aerospace for a while. It’s highly dependent on negotiating a licensing deal with one or two companies though, so I’d prefer not to go into details publicly yet. My goal is to get to a go/no-go decision on launching this new startup by the end of the summer. If we can negotiate a workable license, and the technology is as near-term as I think it may be, and if we can line up adequate funding to get started, this startup will definitely be a worthy place to invest the next decade of my career. And if we’re successful enough, it may loop back around to aerospace toward the end of that time period.

In the meantime, as I mentioned, I’ll be doing part-time consulting for aerospace and deeptech clients. If you’re interested in talking with me, drop me a line via my Starbright contacts page. I have a non-compete agreement with Voyager, and frankly wouldn’t want to directly compete with my friends companies anyway, so we’ll need to make sure that whatever I’d be doing doesn’t directly compete with anything the Voyager family is trying to do, but it’s often worth discussing to see if there’s a way to angle things to be complementary instead of competitive.

Also, I’m going to try an experiment during this exploratory phase — I’m hoping to be a little more active in blogging in the coming days, but I’m planning on posting blog posts that are professionally oriented on my Starbright Blog, and just posting a link over here on Selenian Boondocks. In the long-run if I decide to shut down Starbright, or if I decide that having yet another blog is a horrible idea, I’ll migrate everything back here. But for now we’ll leave Selenian Boondocks for more speculative posts, and leave more professional/policy-oriented posts for the Starbright blog.

Posted in Administrivia, Starbright | Tagged , , , | 5 Comments