Draft for Flexible Exploration Architecture - FLEX

The EDS provides the trans-lunar injection (TLI) velocity, and is discarded. After some days of coasting, at the Moon, the LSAM performs the lunar orbit insertion (LOI). The crew transfers from the CEV to the LSAM and descends to the surface of the Moon, spends some time there and comes back up. (The LSAM performed the LOI and the descent with its descent stage, and the ascent from the lunar surface with its ascent stage.) Then the crew transfers from the LSAM to the CEV and discards the LSAM. The CEV does a trans Earth insertion burn (TEI), and days later re-enters Earth's atmosphere for an ocean landing.

This is quite similar to Apollo, except in Apollo, the CEV equivalent CSM did the LOI burn. The LSAM is the LEM, and the EDS is the S-IVB.

Here in table 1 are the names and masses of the different spacecraft.

Table 1: Stack Mass Breakdown. t means 1000 kg

Name	Empty mass, t	Propellant mass, t	Propellants
ESAS:
CEV	14	9	NTO/MMH per 2006-07-22 (design in flux)
LSAM	14	30	LOX/LH2 for descent stage, ? for ascent stage
EDS	18 ESTIMATE	85* ESTIMATE	LOX/LH2
TOTAL:	47	124
Apollo:
Apollo CSM	12	18	N2O4/UDMH
Apollo LEM	4	11	N2O4/UDMH
Apollo S-IVB	13	~60*	LOX/LH2
TOTAL:	29	99

*: This is propellants in LEO. Part was expended to reach orbit.

In the view of the author, this is all quite sensible and has been proven to work by Apollo. The CEV, without the EDS or LSAM, is also envisioned to be used for missions to the International Space Station. Also parts of the architecture are to be used for Mars missions. The plan enhances the capabilities of Apollo, by putting more mass to the Lunar surface, spending longer there, making it possible to go to any latitude and return any time. Also unmanned cargo missions are in the plan.

Getting the Stack to Orbit, ESAS 1.5 launch architecture

ESAS proposes a "1.5 launch" architecture. The EDS and LSAM are launched fully fueled by a big rocket, named Ares V. Then the crew is launched in a CEV with a medium-sized rocket, called Ares I. They dock in space, and off you go, to the moon!.

The rockets do not exist at the time of writing (august 2006) and would be developed with multiple billion dollars of effort.

Ares I would use a modified space shuttle solid rocket booster as first stage and a new custom second stage, powered by a J-2X engine using LOX/LH2 propellants. Orbited mass around 23-25 tons. (Metric tons, 1000 kg.)

Ares V would use two solids and a main stage powered by existing RS-68 engines from Delta IV (LOX/LH2), that would be lighted on the ground. The upper stage would be the EDS, powered by a single J-2X. The orbited mass could potentially be as big as 160 tons.

One advantage of this is that you can have a big moon stack since it's launched in two pieces. Also doing the ISS missions is easy and can be started quite early, since you only use the smaller Ares I for that. Using Shuttle and Apollo heritage hardware should make it somewhat cheaper than designing things from scratch. Existing Space Shuttle pad infrastructure could also be used to some extent. But still, the projected flight rate of the rockets is low and the costs of developing and maintaining the capabilities loom in the many billions of dollars.

Flexible Architecture

The Proposal

I propose a completely different architecture from ESAS for getting the Lunar stack to orbit. From LEO onwards the mission can be very similar to the ESAS-derived architecture, but how you get to LEO is done completely differently.

We see that over two thirds of the whole mass of the stack is propellants. We also can see that the empty weight of each of the three components is less than 20 tons. Why don't we launch the components separately, and also the propellants separately, on smaller rockets that already exist? Combine them in orbit, and it's just as good as a single launch or 1.5 launch scenario from there on.

That way the development and maintainment of manufacturing and operation capability of new rockets, which is very expensive, can be completely avoided. Also, the architecture would have multiple ways of launching things and when unexpected problems would pop up, they could be maneuvered around. If one launcher failed, the type could be grounded for months and the launches be conducted by other launchers. This is not possible if custom big launchers made specially for the moon program are used. Also, increasing the number of launch of current rockets at the present dry market situation would not increase the price per year much at all. Meaning cheaper prices per rocket. The existing rocket factories already have idle standing armies. The ESAS architecture would create two new rockets which would have a long launch interval and standing factories and manufacturing and launch personnel. That is expensive. With cheaper launch costs, more money could be diverted for actual in-space activities.

Some of the propellants of the current architecture are hypergolic and some are not easily stored in space. The problematic propellants might be best to send right alongside their craft if the propellant mass is not big. Also a well-insulated big propellant depot could be useful to store propellant continuously from one Moon mission to the next and to provide a safety margin and prevent waste of leftover propellants.

The Rockets

There is at least one launcher in USA that can put 20-ton class payloads into low earth orbit, and multiple that can put 9-ton payloads. The evolved expendable launch vehicles (EELV) Delta IV and Atlas V have demonstrated good operation since 2002. Both rockets are optimized for putting payloads to geostationary orbit and could be changed somewhat to have better performance to LEO, but that is not absolutely necessary. The key is to keep commonality and use volume production to get unit price down.

Delta IV by Boeing has an RS-68-powered first stage using LOX/LH2 as propellants, and a new upper stage using an RL-10 variant. The RS-68 was designed for low cost, and the Delta IV plant at Decatur, Alabama was designed for bigger production rates than what has materialized. Delta IV heavy uses three first stages strapped together. It has only had one launch yet (by august 2006) though, which experienced a few seconds early shut down of the booster stages, but the problem should be corrected by now. The Delta IV heavy is the only current US rocket that can orbit masses of over 20 tons.

Atlas V by Lockeed Martin has a Russian RD-180 engine powering it's first stage that has LOX/RP-1 as propellant. This dense propellant makes it quite compact compared to Delta IV. The second stage is a Centaur, using RL-10. The Atlas V (as well as the intermediate model, Atlas III that had the same main engine and upper stage) has a perfect launch record. The Atlas V Heavy rocket with 3 first stages in parallel exists as a design, but has not been built. It could be built within a reasonable short time, if necessary.

The performance numbers for LEO 28 degree launches of existing high-payload US rockets are as follows:

There are also other US medium-sized rockets that have not yet flown, but are already in development for the commercial and NASA's COTS space station supply market, and should have their first flights in the next two years (in 2008): Falcon 9 by SpaceX and K-1 by Rocketplane Kistler.

There are many rockets outside USA that operate currently and can lift payloads in the over 10 t class. Many countries could be happily participating in the Moon program. Adding many types of launchers creates more possibilities for problems, but it also adds redundancy to the program. In the event of launch failure by one party the mission can be carried through by sharing the weight to others. Some of the designs that are currently in use are optimized for GEO (at least Ariane 5 G/ECA, Zenit 3 SL), they might be changed with some expense to launch more payload to LEO. Inclination proves to be the biggest hurdle. Many countries do not have possibility to launch to a 28 degree inclination.

There's also developments, like Japan's H-2B that is supposed to lift 16 t.

Propellant Choice

Name	Empty mass, t	Propellant mass	Of which LOX, t
CEV	14	9	0
LSAM	14	30	25
EDS	18	85*	73
TOTAL:	46	124	98

LOX is a pretty obvious choice. The stack consists of 46 tons of spacecraft, 124 tons of propellants of which 98 tons is liquid oxygen. Our depot needs to only hold liquid oxygen to be extremely useful for a lunar architecture.

The Payloads

Overview:

In detail:

Depot

The propellant depot needs to hold over 100 t of liquid oxygen. It could be an EDS with a stretched tank and no LH2 tank. There are already plans for stablilizing the EDS & LSAM since the CEV needs to dock with them in the 1.5 launch ESAS scenario. The way to launch the EDS-class payloads is described later.

The propellant depot would also have the multitude of sensors and a communications system to guide the approaching vehicles to a safe distance. It could also have a robotic arm so the approaching vehicles don't have to actively dock. They can be captured by the robot arm and berthed, reducing precise maneuvering requirements.

Tanker

This vehicle is either an upperstage with a stretched LOX tank, if only a few types of launchers are used as tankers, or a commonalized serial manufactured maneuvering vehicle of about 8 to 9 tons of mass holding about 7 tons of liquid oxygen. It can be launched to a precise 28.5 degree orbit by any launch vehicle capable of launching such a mass there. If feasible, multiple stretched variants can be built for different rockets to use all their payload mass and volume. Alternatively, a single variant is used, but only partially fueled when launched with smaller rockets.

If the somewhat inflexible approach is taken, and a stretched upperstage is used together with a heavy-lift vehicle, the LOX amount can perhaps increase by about 3 tons.

LSAM and CEV launcher

This is a heavy launcher capable of launching the 25-ton CEV as well as the ~20-ton LSAM (with hydrogen but without LOX) to the propellant depot. This could be done by the NASA designed CLV/Ares I, but can as well be done by the Delta IV Heavy or Atlas V Heavy.

EDS launcher

EDS has to function as the second stage of a rocket and has to launch itself to orbit, just like the CLV/Ares I second stage, which is the "sister stage of EDS" in the ESAS architecture. But it must store excess hydrogen for the TLI.

I.e. the payload is the upper stage itself. Such an upper stage can probably be fitted to a Delta IV Heavy launcher too. It is big and filled with hydrogen, while the LOX tank is about third full, just enough oxygen there to make it able to lift it's own mass to orbit.

The propellant depot is just a variant of this EDS, with a bigger LOX tank proportion.

As comparison, the existing Delta IV Heavy second stage weighs 31 t when fueled and 3.5 t when empty. Since the EDS drymasses 18 t, it's 14.5 t additional to the current upper stage. It has to contain 12 t of hydrogen for TLI, bringing the total payload mass to 26.5 t. As the Delta IVH can orbit a payload of 24-25 t, it might be a just a bit too big. The second stage would thus have:

and all the LOX and 4 t of hydrogen would be consumed on the way to orbit.

In reality, the EDS might not need to be so heavy in structure: it doesn't need to carry a full LSAM weight on top of itself. It also doesn't need as big propellant tanks since it doesn't have to insert the LSAM to orbit. (The ESAS EDS is not full when it starts TLI.) The rocket's LEO capability might also be slightly different because of the high-thrust J-2X engine reducing gravity losses but having probably lower ISP than the RL-10. It is also conceivable that multiple RL-10 engines might be used. The current Delta IV heavy can probably not take off with such a heavy payload, and lowering tankage in the boosters or some method of uprating would become necessary. If needed, the propellant depot could also store hydrogen, but that would complicate the design somewhat. Alternatively, the LSAM descent stage could be overtanked (since there's room for payload increase in its launch) and give some of its hydrogen to the EDS stage before TLI. The margins here are so small that better study is required.

Operations

Propellant boiloff is a problem for LOX in LEO although it's much easier than for hydrogen. Multilayer insulation (MLI) would help here with the depot, where the LOX is stored for the longest duration. Active refrigeration is not out of question either. After all, the depot stays in one place and additional mass doesn't hurt any performance. Numbers for this are hard to come by. Orbit height does affect it.

If the depot is launched to a 29 degree orbit, the payload penalty is minimal, but the subsequent launch windows for docking with it extend nicely so that there are two good opportunities within minutes of each other, every day. (According to ESAS.) This makes launch quite easy. Of course launching from Kourou has smaller launch windows. Planetary probe missions already require very precise launch windows and have been succesfully executed by many rockets, including EELV:s.

One factor determining the boiloff is the length of time during which the tankers and the Lunar stack can be launched. But if multiple types of launchers are used, they can be prepared in parallel. Also, if it's cheap to construct pads (for the Atlas V it seems so), it might be good to have multiple launch vehicles of one type on adjacent pads readied for launch.

All the three areas above, propellant boiloff, launch windows and launch intervals are interlinked and require closer study than done here.

An example with 7 ton medium tankers:

Launcher	Payload	LOX in depot, t
Heavy	Propellant depot	0
Medium	tanker 1	7
Medium	tanker 2	14
Medium	tanker 3	21
Medium	tanker 4	28
Medium	tanker 5	35
Medium	tanker 6	42
Medium	tanker 7	49
Medium	tanker 8	56
Medium	tanker 9	63
Medium	tanker 10	70
Medium	tanker 11	77
Medium	tanker 12	84
Medium	tanker 13	91
Medium	tanker 14	98
Heavy	LSAM	73
Heavy	EDS	0
Heavy	CEV	0

An example with 23 ton heavy tankers:

Launcher	Payload	LOX in depot, t
Heavy	Propellant depot	0
Heavy	tanker 1	23
Heavy	tanker 2	46
Heavy	tanker 3	69
Heavy	tanker 4	82
Heavy	tanker 5	105
Heavy	LSAM	80
Heavy	EDS	8
Heavy	CEV	8

Mission Success Probability Analysis

Let us calculate with 98% and 99% launch success probability numbers.

First, the depot:

The depot is launched empty, and if its launch fails, a new one must be launched. It is to stay in orbit for years supporting future lunar flights too. It must be sent before any other mission and it can be done so that it is not a time critical operation.

Should we therefore, when calculating each mission's success probability, include the launch success probability of the depot? It is not a clear question, but it seems to me that no. It is a program risk that has to be dealt with, not a time-critical per-mission risk.

The LOX tanker flights:

There is a risk of failure, and in case the LOX is launched with medium vehicles, calculating about 7 tons of LOX per launch, about 14 launches might be needed. With 98% reliability, there's a 25% chance of one or more failures, and with 99% reliability, 13% chance. With the sad state of current launchers, it just has to be taken that "failure is an option". You manufacture spares of tankers and rockets too. If you have one spare, the probability of two or more failures with 98% is 3% and with 99% it is 0.8%. When you sustain the program, spares are left over from previous missions and can be used for launches. This should not increase the cost of the program much at all, one just has to have a manufacturing buffer. The LOX boiloff rate is a significant question to the time criticality of the tanker flights. If two failures can be tolerated, the probability of more than two failures is 0.246%, counting with 98% launch probability .

If LOX is launched 23 tons at a time, about 5 launches are required. With 98% success, a 10% chance of failure and with 99%, a 5% chance of failure. With one spare, a 0.4% or 0.1% chance of failure.

So, the tanker launches can be approximated to be quite safe with this kind of planning. With small launchers and 14 tanker launches you have a chance of failure 0.8 - 3% with one spare and less than a quarter of a percent with two spares. With larger launchers and one spare the mission failure probability is 0.1 - 0.4%.

The LSAM, EDS and CEV flights:

Here, mission risk is hard to calculate and we get a decision tree. Perhaps the EDS, as the cheapest component, is launched first. 98% or 99% success. But if it fails, perhaps the mission can be called off. We did not lose the LSAM or CEV, the much costlier parts of the architecture, they can be used at a later date. But the LOX in orbit might boil off. This we do not know and the time of standdown is unknown too. Or perhaps a spare for the EDS had been manufactured, as well as a spare launcher, and the mission can go on with no fundamental flaw in the launcher being found. What is the preptime for another launch etc? These are questions that need a closer look.

Say that half of the LOX boils off, we send new tankers and a new EDS (98 or 99%) and then the LSAM. This again has a probability of failure. It is an uncrewed launch though, and if it fails a new one can be launched or the whole mission delayed. The CEV is still spared. What if the last part, the CEV fails? How long can the EDS and LSAM be stored in orbit? Will they have to be ditched?

Maybe we can calculate it like this, if we assume that the loss of any major component (EDS,LSAM,CEV) ditches the whole mission:

If we assume that components can stay in orbit for long times and/or that LOX boils off slowly and that some spares are manufactured or CEV missions are done in close succession, the numbers in the above paragraph are not true.

If NASA is headstrong and keeps launching spares with close intervals until it succeeds, then the loss probability of each CEV, LSAM and EDS is the same, 2% or 1%.

Critique of ESAS Multi-Launch Analysis

Launch delays

The ESAS report, in section 6.4, calculates that launch delays are a serious reliability-lowering factor in multi-launch scenarios, but the analysis is based on all launching done from a single pad by a single type of launcher, which is clear to give problems. Thus the ESAS analysis must be discarded, when using multiple launch pads or even multiple launcher types, which is proposed in this document.

Spares

The report also doesn't take into account the possibility of spare launchers, which clearly show a huge improvement in mission success probability at least for the tanker portion where the space hardware is inexpensive.

Docking vs Berthing

ESAS considers docking a necessity, while the other alternative, berthing seems unheard of.

In berthing the approaching vehicle needs to just arrive in the vicinity of the depot, which can grapple the vehicle with a robotic arm, and berth it against a docking or refueling port. In FLEX, no big pressurized human passageways need to be achieved in autonomous dockings, only the CEV-LSAM tunnel is such, just like in the ESAS baseline configuration.

It is worth making the propellant depot active and thus somewhat more expensive to increase the easiness of docking, since it will stay on orbit for quite long anyway, while the approaching vehicles are discarded. NASA doesn't currently have unmanned docking or even rendezvous capability, but it could be developed. The American military XSS-11 testbed and the Russian Progress/Kurs system used to supply MIR and ISS show good promise toward this.

Conclusion

It must be concluded that ESAS didn't cover much of the multi-launch territory for some reason, and it definitely didn't analyze anything much close to the configuration proposed in FLEX.

Further Concepts

In the ESAS scenario, the EDS has to expend part of its propellants to reach orbit. But in the FLEX scenario, it only has to expend hydrogen, LOX is retanked, and thus it can reach a better mass fraction. Could it also perform the TLI burn? Could the CEV be unneeded (all the crew descends to the Moon anyway) and the two-stage LSAM with a heat shield perform a direct lunar landing mission without lunar orbit rendezvous (LOR)?

This of course still leaves the ISS problem for transferring the crew. But if the COTS commercial supply pulls through, this could be a non-issue. We shall see.

Alternatively, one can keep the architecture but use more propellants to increase margins and make some of the crafts reusable. Like the LSAM, make it single-stage and use methane and LOX. It can stay at Earth-Moon L1 or L2 Lagrange point between visits to the Lunar surface. There probably has to be a methane depot at LEO then too, since the methane mass is substantial. There is also a possibility of producing LOX on the lunar surface. That remains to be demonstrated.

If the Lunar missions are done for a long time and often enough, the program can function as a market for tens of medium-size launches per year to low Earth orbit, essentially doubling the size of the current world's launch market. That could be the perfect market for a completely reusable vehicle. Some studies have suggested that the breakeven point of reusable launch vehicles is 50 launches per year. Potentially it could be lower for partially reusable vehicles. If such a cheaper to fly - vehicle would be created, there is the potential for entering a virtuous cycle where space launches increase and new uses and markets are found for space applications and price goes lower and launches increase ever more.

Summary

FLEX proposes launching all hardware in less than 26 metric ton chunks. First an empty fuel depot is launched, then many tankers one by one to fill it with liquid oxygen. Then follow the LSAM (without LOX in the descent stage) and close after, the EDS (without LOX). These are sufficiently light when empty that they can be launched with a Delta IV Heavy or similar launcher. Finally a CEV is launched with a crew. The LSAM and EDS take their LOX from the depot, and the stack travels to the moon for a normal mission. The same depot can be used for the next mission.

In short, the FLEX architecture is superficially both more flexible and cheaper than the NASA-proposed ESAS architecture. No new rockets need to be developed. No rarely-used expensive launcher capabilities need to be maintained. (They are very people-expensive.) The FLEX launchers are interchangeable, any rocket can be used to send propellant to the depot, it is no different from a satellite launch. Any sufficiently large and safe rocket to launch the depot, EDS, LSAM or CEV. This can include reusable launch vehicles or anything.

The flexibility allows NASA to stimulate the launch market, if it chooses to pursue that route. With a sustained pace of Moon missions, existing prices per launch will drop. Also new launch service providers can emerge to give NASA more "bang for the buck". Both are good for Lunar space exploration and all other space exploration too. Because of the refueling aspect, many of the in-space vehicles can also be changed/designed to be reusable, which can lower cost significantly in the long run.

More in-depth study is needed to find out the details and more closer price, availability, schedule and operations issues.

Appendix 1, probability mathematics

For the sake of portability, exponent is marked with sign '.

With 14 launches, the probability of all succeeding, P(0 failures), is naturally 0.98'14 = 0.753.

Probability of any one of those failing, P(1 failures), is 0.98'13 * 14*0.02 = 0.215.

Probability of any two failing, P(2 failures) is 0.98'12 * 14*13/2*0.02'2 = 0.029

Thus the probability of three or more failing is 1-( P(0 failures)+P(1 failures)+P(2 failures) ) = 0.00246.

List of sources, figures and tables

Sources: