"Space in 2015 will be very different than it is today. Lower cost access to space will lower the traditionally high barriers to market entry, ushering in an entrepreneurial gold rush. This new commercial environment will favor small companies that can rapidly identify new markets and field innovative products. Andrews Space & Technology is focused on being one of those companies"

- AS&T Vision Statement

Andrews Space & Technology (AS&T) was founded with the long term vision that, eventually, launch costs can be reduced to the point that new "future markets" will be enabled. In addition, these markets have to be a critical part of any business plan.

This paper uses AS&T's Market Analysis efforts as a starting point for a systems engineering analysis to derive 2^nd (Next) * Generation RLV Design Requirements.

Typically, new markets are referred to as Emerging Markets. However, all of our analysis indicates that the next wave of commercial space commerce will be driven by established "brick and mortar" companies figuring out how to improve their bottom line by doing business is space. As a result, there are only a few new (emerging) markets. As a result, we have chosen to adopt the term Future Markets, or Future Commercial Markets.

Future Markets must play a vital role in developing a commercial 2^nd Gen RLV. A commercial RLV Business Case will not close without something more than ISS Logistics and communications satellites. However, an analysis of the existing markets indicates that increased flight rates are not feasible. The bulk of the addressable commercial launches is GEO Satellites, which are projected to number 30 per year over the next decade. The satellite manufacturer's have a preference for multiple launch providers to ensure a backup. And despite public statements about high prices, launch cost is not a significant part of a satellite service provider's business plan.

If the commercial satellite market does not justify a business plan, what other options are there? The obvious answer, is the US Government in the form of ISS logistics. The US portion of ISS logistics is on the order of 80,000 lb per year of cargo (pressurized) and payloads (unpressurized).

Launch Vehicle	Flights
Ariane (4 and 5)	12
Atlas (2A and 2AS)	4
Delta 2	7
Long March (2, 3 and 4)	4
Proton	7
Soyuz	10
Titan (2 and 4)	3
Zenit (3SL - SeaLaunch)	3
TOTAL	50

AS&T's analysis indicates that there are three aspects to fielding a Commercial 2^nd Gen RLV system:

1. Markets: Which "Future Markets" can be enabled to make a business case close?

2. RLV Architecture: What RLV system has the inherent safety, operability, and low recurring costs to address / enable the future markets?

3. Incubating Future Markets: What is required to incubate the Future Markets to enable a 2^nd Gen RLV?

Andrews Space & Technology conducted the Future Space Transportation Study (FSTS) under NRA8-27 (2^nd Gen RLV Risk Reduction), which examined approximately 20% of the Future Markets identified in past future market studies (CSTS, 1994). As part of its NRA8-27 effort, AS&T developed a multiphase analysis process to identify possible future commercial markets and flow down requirements to the space transportation system vehicle level. The complete roadmap of the FSTS effort, with its intermediate and final data products, is illustrated in Figure 1. Although the diagram flows from left to right, in actuality the process derivation starts at the right side with 2^nd Generation RLV Design Requirements and works backwards through the business process.

A 2^nd Gen RLV, which is specified to meet the "shuttle design reference mission" of delivering either 50,000 lb of payload or 20,000 lb of cargo per flight, ISS logistics requires only six flights a year, hardly sufficient to "close" a commercial business case. As a result, additional flights and markets are required. Military missions hold promise, but flight rates are tied to the likelihood of military conflict. In addition, it is unlikely that the military will procure launch services, especially for force projection missions, from a "commercial" RLV provider (commercial bombers?). Therefore, the to close a "commercial" 2^nd Gen RLV business plan the business must turn to Future Commercial Markets.

The driving force for determining the 2^nd Generation RLV Design Requirements is the 2^nd Generation RLV Business Plan. From the business plan key design requirements, or attributes, can be determined. These include flight rate, system reliability, payload performance, launch environments, system cost, recurring cost and operational requirements. These attributes are essentially the system level 2^nd Gen RLV design requirements, from which the stage and subsystem requirements are derived.

Continuing to work backwards, the 2^nd Generation RLV Business Plan is predicated on the business plans of its customers. The information derived from the customer's business plans; such as market elasticity, demand versus launch price, and infrastructure requirements; is used to develop the 2^nd Generation RLV business case. The 2^nd Generation RLV Business Case and the Future Space User Business Case, as described above, can be performed either individually or in an integrated fashion. These two activities are highly inter-dependent. Because of this, Andrews Space & Technology has placed particular emphasis on studying future markets in conjunction with their derived design requirements.

The precursor to developing a Future Space User Business Case is the identification of companies or industries that might benefit from doing business onorbit. These companies must have one, if not several, compelling reasons to relocate portions of, or expand, their business to a space based facility. Specific reasons include improvements in product quality, product quantity, the product's uniqueness, the ability to enable a new product or market, reductions in time to market, and reductions in production cycle time. All of these lead to an improvement in the company's bottom line or ability to turn a profit.

The process described above can be depicted graphically. In Figure 2, the Future Markets derive the space infrastructure and platforms, which in turn drive the space transportation logistics requirements. From this process we can make some observations. First, it is clear that if even a small percentage of the potential future markets come to fruition, it will significantly alter how the aerospace community perceives space transportation. The future markets will create a large, diverse space infrastructure predominately located in LEO. To address the logistics requirements of this infrastructure, RLV designers will be driven from the current mode of designing highly complex, inoperable, closed standard systems. The Future Markets will force the industry towards safe, human rated, highly operable space transportation. Second, one of NASA's goals for 2^nd Gen RLV is to maximize the convergence between commercial mission requirements and NASA's. Launching commercial satellites to GTO is very different than delivering seven crew members to ISS. However, if the future markets are incorporated, one of the primary commercial markets becomes deploying and maintaining a manned LEO infrastructure, which is convergent with NASA's.

AS&T analyzed the data collected from the FSTS interviews and utilized a system engineering process to identify a broad requirements set of 50 requirement / attribute pairs (Figure 3). The various attribute/requirement pairs were chosen to reflect the needs of the markets that are to be served, while maintaining the minimum number of limitations imposed on the transportation system designer. All of the collected attributes were sorted into six major categories (Scheduling, Operations Performance, Interfaces, Business, and Provider Specific), including the important distinction between requirements imposed by the customer of a space transportation industry (Customer Specific), and those determined by the "space-line" and imposed on the vehicle manufacturer directly (Provider Specific). Requirements values were derived for each individual market segment and the most limiting values distilled based on the investigated markets (see Figure 4 as an example).

From AS&T's Future Space Transportation Study, we have concluded the following:

FSTS Conclusion #1: This Future Space Transportation Study was a limited scope effort that analyzed approximately 20% of the potential future markets, as outlined by the Commercial Space Transportation Study (CSTS) published in 1994. The results of the limited market analyses supported the general conclusions put forth by the CSTS: that the space launch market is in-elastic above a certain launch price point (approximately $600 per pound) and elastic for prices below. At this time, AS&T has conducted insufficient analysis to make further recommendations on the size, shape and slope of the elasticity curve. Continuing market analysis to define elasticity is critical to the continued growth and evolution of the space launch industry.

FSTS Conclusion #2: Many of the future markets will be enabled once the frequency and cost of space access achieves thresholds that allow established terrestrial industries to earn money from space based operations. New revenues will come from multiple established industries, which will reduce the investment risk of fielding a 2^nd Generation Launch System. As an example, many emerging launch vehicle companies (i.e. Kistler Aerospace Corporation, Kelly Space & Technology, Pioneer Rocketplane, Rotary Rocket, etc.) relied almost solely on the emergence of LEO communication satellite constellations, a new and unproven industry itself, to attract investment and achieve commercial viability. This created a situation where business risk was piled on top of business risk. In contrast, the FSTS market study indicates that future market revenues should come from many different business sectors and consist of capturing very small fractions of large established industries. Figure 6 highlights an example based on the markets studied under FSTS.

The tourism industry has annual revenues of US$1 trillion. Adventure Travel comprises approximately US$200 billion of those. Assuming that safety can be improved and costs significantly reduced, it is not unreasonable that a 2^nd Generation Launch System can capture (or add) 1% or US$2 billion in annual revenues from commercial passenger travel and tourism.

The semiconductor and pharmaceutical industries, which have approximately US$550 billion in combined annual revenues, spend between 10% and 15% on Research & Development. High technology industries are typified by a fiercely competitive landscape, which has every company seeking a competitive advantage and causes companies to take high risks. If a 2^nd Generation Launch System can provide the companies with frequent low-cost access to orbiting research facilities, it is well within the grasp of reality that these companies could spend at least 1% of their annual R&D budgets on space based research, which could easily total another US $500 million. These revenues, US$2.5 billion for R&D and passenger travel, are nearly equal to current commercial GEO satellite launch revenues, and can significantly help the business case of a commercial 2^nd Generation LV.

FSTS Conclusion #3: Based on the FSTS path finding study, which represents the first comprehensive system study to derive transportation design requirements for the future markets, the study team concluded that a 2^nd Generation Launch Vehicle, designed to address future markets, must be designed to support the business cycles demanded by the future user community. As an example, both airline companies interviewed outlined the need to limit the time from when a passenger boards a vehicle to when they arrive at their destination. Specifically, the airlines would prefer to limit the time between when a passenger boards to when they are launched to two hours, and to limit the transit time from launch to arrival at the destination to six hours. For the Space Shuttle and Soyuz systems, this span averages approximately two to three days due to the relaxed launch window and extensive orbit phasing operations. To correct for this, a 2^nd Generation vehicle must routinely meet a very narrow launch window (measured in seconds) in all-weather conditions.

As another example, semiconductor companies developing a new generation of microchips, build multibillion-dollar factories, pay off their capital investments and generate huge profits (80% profit margins) all in the span of 18 to 24 months. For these companies, R&D campaigns are measured in hours, days and weeks. Currently, it takes years to plan, design, and implement orbital tests. Until these disparate business cycles are reconciled by improvements in space transportation and on-orbit infrastructure, many of the future markets will remain unaddressable.

FSTS Conclusion #4: Future markets must be developed in concert with a 2^nd Generation Launch Vehicle. It was clear from the study team's interviews that very few people outside the space industry understand the opportunities of space and how they might benefit their business. Furthermore, the space infrastructure required to address the needs of the future markets is very different than what is operating today. Many of these future markets require new facilities and processes, in addition to the Earth to Orbit transportation infrastructure, which require years to develop and deploy. As a result, any space transportation service provider who expects to address future markets cannot, must not, rely on a "build it and they will come" philosophy. It is incumbent upon industry and NASA to devise a future market incubation plan that serves to: 1) promote space awareness to non-aerospace companies; 2) incubate near term future markets (e.g. space tourism); and 3) act as "stepping stones" that will lead to fully developed, robust commercial space commerce.

Working from the market analysis and conclusions described above, AS&T employed a system engineering process to derive detailed space transportation system requirements. A very top-level summary can be found in Figure 7. An analysis of the requirements reveals the following observations. A 2^nd (Next) Generation RLV should:

• Emphasize Safety: Addressing a manned LEO infrastructure, and transporting common citizens, requires achieving loss of vehicle (LOV) reliabilities in excess of 1 / 10,000.

• Emphasize Operability: Yearly flight rates in 2020 could approach 150 flights per year (every other day), requiring almost constant launch availability. System should be either capable of operating in all weather conditions, or have a HTHL type architecture to operate out of a central operating base and fly to an area with weather suitable for launch. In addition, vehicle turnaround times should be less than one week, which requires subsystems to go tens to hundreds of flights without heavy maintenance.

• Minimize Recurring Costs: $500 / lb implies launch prices of $10M for 20,000 lb of cargo to ISS. Launch costs must be on the order of $5M per flight, which is split roughly evenly between operating overhead, propellant and spares (amortization of subsystem replacement parts - e.g. engines).

• Maximize Flexibility: Future markets are, by nature, unpredictable. The Next Generation RLV must be able to address "real world" business cycles that change every few years. In addition, this may require adopting "open standards" for stage interfaces and all subsystems.

AS&T has analyzed a number of HTHL TSTO architectures, using a system engineering process, to investigate compliance with the safety, reliability, operability and maintainability design goals. For all system architectures, a LOX / LH2 propulsion system was used to maximize performance of an air-launched type system. The concepts and ground rules investigated were:

1. Aerial Refueling: Aerial refueling is proposed by Pioneer Rocketplane Corporation and requires flying off a runway under turbojet engines with the complement of rocket fuel. Once at altitude, the rocketplane rendezvous with an aerial tanker (assumed to be a B747) and transfers up to 350,000 lb of Liquid Oxygen. Once fully fueled, the engines are ignited, the first stage accelerates to Mach 8 and 200,000 ft, at which point the vehicles stage and the second stage continues on to orbit to deliver its cargo. The benefits of aerial refueling are that the vehicle takes off without oxidizer, which makes up 60% of the GTOW, so that the wing, landing gear, and overall vehicle size are significantly smaller, resulting in a lower development cost. In addition, because the vehicle is not fully fueled until it reaches altitude, the abort modes are greater and the odds of crashing into the ground fully fueled are greatly reduced. However, aerial refueling is limited by the propellant storage weight of the tanker aircraft and, as a result, has limited orbital payload capability. For our analysis, it was assumed that a fleet of B747's was not a feasible operational approach, so we limited the analysis to one B747 tanker.

2. Tow Launch: Tow launch is proposed by Kelly Space & Technology, Inc. HTHL TSTO's are limited by the prohibitive wing, landing gear and propulsion system masses required to get them airborne. One option is to use a tow plane to augment the TSTO's thrust at takeoff and low speeds, in effect "lifting" the heavy system off of the runway. The drawbacks of this approach are the reliance at low speeds and low altitude on a tow line, which, if it broke, could leave the TSTO RLV to crash into the ground. In addition, the maximum takeoff weight of the TSTO is limited by the size of the tow plane (most likely a B747). As a result, the maximum ignition mass of a tow-launched system is on the order of 850,000 lb, resulting in pressurized cargoes of around 8,500 lb.

3. Rocket Assisted T.O.: Rocket assisted takeoff uses a combination of rocket engines augmented by a rocket powered sled to get the required thrust and velocity to achieve flight, stage and accelerate into orbit. Rocket assisted takeoff has safety issues since it is relying on lower reliability, lower performing rocket engines to remain airborne. In addition, rocket assisted T.O. has higher \Delta V requirements than traditional air launched systems, and is limited by the 1M lb mass limits of runways, so it has more aggressive weight requirements, which reduce overall margins, safety, and operability.

4. TBCC: Turbine-Based Combined Cycle (TBCC) launch systems utilize turbine engines for takeoff and acceleration to Mach four where a separate ram/scramjet system takes over. Use of the oxygen in air up to Mach eight or ten reduces the gross takeoff-weight since the booster carries no liquid oxygen. However, there are significant structural weight penalties associated with hypersonic flight at high dynamic pressures (~ 2000 psf), so a TBCC sized for shuttle-like payload capability weighs more than a million pounds at takeoff, even with advanced structures and propulsion technologies.

5. Subsonic In-Flight LOX Collection (ACES): Subsonic in-flight LOX collection has the same operational advantages of aerial refueling: a smaller system with more abort modes; but eliminates the issue of using a fleet of tankers by producing the required oxidizer on board while in flight. The advantages of this approach are significant, since a vehicle can fly off the runway with traditional landing gear and jet engines, but still have rocket ignition masses in excess of 1.25Mlb. As a result, ACES is both the safest (most abort modes) and highest performing (theoretical cargo values with 750klb GTOW limit approaching 25klb to ISS) option identified.

Figure 8 provides a performance summary of the options investigated.

High Loss of Vehicle reliability is achieved by having one failure tolerance on the main propulsion system. For a TSTO RLV, turbofan engines can be used as backups to the rocket engines. However, to do this, the system must be designed to 1) withstand a loss of the rocket engine cluster and 2) have sufficient performance margin and wing area to facilitate turbofan powered flight. To address the issue of rocket engine catastrophic failure (i.e loss of the cluster), on-board IVHM could, to the extent possible, detect and contain the failure and shut down the engine. In these cases, rocket propulsion will be terminated and the stage(s) will re-enter, transition to turbofan powered flight and fly to the nearest airport using on-board JP8 and / or residual hydrogen. For additional safety, the AST Team favors the split-expander engine cycle because of its graceful failure modes. Coupling a reliable real-time IVHM system with inherently safe engines (ones with a high ratio between uncontained and contained failures - e.g. 1/10), and an air breathing HTHL architecture, results in a very safe space transportation system.

Having one-failure tolerance on the main propulsion system has tremendous implications with respect to vehicle safety and the cost of certifying the main engines. It means the design can meet 1/10,000 loss of vehicle with an engine catastrophic reliability of only 0.999. (This assumes seven engines and 1/20,000 loss of vehicle from non-propulsive causes).

Table 2 compares the casualty rate, the cost of certifying a new engine, and the amortized cost per flight of a vehicle with those new engines, as a function of the demonstrated engine reliability. Assuming certification testing is to a 50% confidence level, a catastrophic reliability of 0.999 requires 500 successful engine tests that cost about $50M (ignoring the cost of engine development and test hardware for the moment). This is certainly affordable, but trying to meet demonstrated reliability for 1/10,000 loss of vehicle with zero failure tolerance is going to require test costs close to $10B.

Engine Catastrophic Reliability	0.995	0.999	0.9995	0.9999	0.99995	0.99999
Engine Certification Cost, $M	10	50	100	500	1000	5000
Casualty Rate w/ Zero Failure Tolerance (7 engines)	1/25	1/122	1/243	1/1157	1/218	1/7607
Amortization Cost /flt w/ Zero Failure Tolerance, $M	51.9	0.5	5.26	1.05	0.53	0.11
Casualty Rate w/ One Failure Tolerance	1/919	1/10619	1/16492	1/19831	1/19957	1/19998
Amortization Cost /flt w/ One Failure Tolerance, $M	1.419	0.058	0.0145	0.0006	0.0001	~0

When addressing the Future Markets, which requires achieving launch prices of $500 per pound or $10M per flight for 20,000 lbs of cargo, every effort must be made to reduce recurring costs. For high flight rate systems (greater than 50 a year), operations costs are driven by general overhead and per flight expendables, either actual (e.g. propellant) or amortized (e.g. rocket engines). Business models, such as those for the airlines, indicate that, for high flight rate systems, per flight costs break out into 1/3 overhead, 1/3 propellant, and 1/3 amortization of subsystem replacement costs. As a result, per flight engine costs are critical. If we assume that the overhead and propellant costs are relatively equivalent for comparable performing TSTO RLV systems, then the major ops cost discriminator is the number of main engines.

2^nd Gen RLV propulsion requirements are currently focused on developing main propulsion engines that have a major overhaul at 100 missions and replacement at 200 missions. This breaks out to "power by the hour" costs of between $150,000 and $250,000 per engine per flight. Figure 9 examines total propulsion system per flight costs based on an average number of $200,000 per engine per flight.

A main contributor to AS&T's preference for HTHL architectures is the lower total thrust requirements versus VTHL type systems. Specifically, air launched systems have lower total \Delta V requirements (28,500 fps versus 30,500 fps), which results in smaller systems requiring less total thrust. In addition, because HTHL systems "fly" out of the atmosphere versus rising on a pillar of flame, they can get by with initial thrust to weights of 0.8 versus 1.25 for vertical takeoff systems, reducing total thrust requirements further. As a result, comparable performing air launched HTHL TSTO's have approximately half the total number of engines compared to their VTHL brethren. This point is characterized in Figure 9, where the total propulsion system ops cost is plotted for a four engined HTHL TSTO versus a ten engined VTHL Bimese VTHL. Note that both options deliver approximately 20,000 lb of cargo to ISS.

Orbital mechanics and the earth's rotation moves the ground track of a station in LEO approximately 1500 miles westward each orbit (about every 90 minutes). As a result, a ground-launched system operating from a single site has approximately two launch opportunities a day, but rarely will the ground track cross directly over the launch platform to allow direct insertion. However, to address both launch opportunities, ground launched systems must be capable of launching in allweather conditions, whether it be a strong windstorm or a snowstorm, and still requires some phasing before rendezvousing with the orbiting destination.

An air-launched system has two operational advantages over ground-launched systems. First, the same system has the ability to operate from multiple bases. As an example, if a HTHL system could operate from MSFC (in Alabama), JSC (in Texas), or Edwards AFB (in California), the odds of having all three operating hubs socked in at once is significantly less than the odds of having thunderstorms at KSC. Second, many of the HTHL air launched concepts have cruise ranges in excess of 1500 miles, which means they can takeoff and fly east or west to intersect the ascending node ground track at least twice a day and the descending node at least twice a day (occasionally three times a day if the ground track passes directly over the operating base). Having multiple operating bases to enable satisfactory weather conditions and four launch opportunities a day should initially be sufficient to service future commercial markets. Most importantly, an air launched system can set up its launch location to allow for immediate rendezvous with the orbiting destination, eliminating the lengthy phasing operations required by most ground launched systems.

The demise of the LEO constellations, and the resulting financial challenges experienced by the emerging launch service providers, demonstrates that it is very difficult to build and system and field a service for markets that do not exist. This is primarily because, in those situations, the business case has too much development or execution risk (will not close without a "then a miracle happens") to attract the hundreds of millions or billions required to develop a Next Generation RLV. As a result, the third element of AS&T's RLV development approach is to identify methods for incubating the Future Markets in parallel with development of the RLV. It turns out that NASA already has a program underway that could do this.

It has already been stated that the Future Markets will primarily involve terrestrial companies learning how to do business in space, particularly in LEO on platforms that can support their R&D, manufacturing, and entertainment enterprises. The global space community has a LEO platform to facilitate these activities: the International Space Station. The primary deficiency, from AS&T's analyses, is the lack of an adequate transportation system that has both the routine flight availability and cargo bandwidth to support space commerce, the two-way flow of goods and services. An analysis of the existing visiting vehicle fleet (those capable of docking with ISS) reveals that there is sufficient up-mass capability, but a severe lack of recoverable down mass capability (Figure 2 depicts the current set of visiting vehicles). Specifically, the Soyuz can recover up to 100 lb of cargo while the Space Shuttle can recover in excess of 10,000 lb. However, both systems are dedicated to crew transfer and, in the case of the Space Shuttle, lead times to get on the cargo manifest is measured in years, if you can get on at all.

As part of the Space Launch Initiative, Congress and OMB allocated approximately US$315M to initiate the Alternate Access to Station program, which is slated to develop a domestic back-up to the current fleet of visiting vehicles. AS&T participated in the first round of program awards and, through our analyses, concluded that the AAS program was prudent to mitigate ISS logistics risk, but was also critical as a 2^nd Gen RLV risk reduction effort. Specifically, a properly sized AAS system capable of recoverable down mass , could provide both the cargo bandwidth and routine access required to help in the incubation of future markets and facilitate space commerce. Furthermore, the AAS program is slated to come on line the in 2005 to 2007 time frame, a good five to seven years ahead of a 2^nd Gen RLV (Figure 10). This will allow the future markets to take root and grow to the extent that they will be demanding more space transportation capability right around the time a 2^nd Gen RLV becomes available.

AS&T's analysis and past experience indicates that Future Space Markets must play a substantial role in the development of a 2^nd (Next) Generation RLV. Regardless of the system design solution, AS&T's analysis indicates that there are three aspects to fielding a Commercial 2^nd Gen RLV system:

1. Markets: Which "Future Markets" can be enabled to make a business case close?

2. RLV Architecture: What RLV system has the inherent safety, operability, and low recurring costs to address / enable the future markets?

3. Incubating Future Markets: What is required to incubate the Future Markets to enable a 2^nd Gen RLV?

This paper explored each topic in more depth. Specifically, servicing the future markets requires a system that is safe and reliable, flexible enough to accommodate market changes, and capable of high flight rates and low cost operations. AS&T's requirement derivation activities have shown that such a system requires airplane-like operating characteristics and airplane-type safety and design margins. This led AS&T to a HTHL air-launched TSTO RLV solution, which employs subsonic in-flight LOX production. This system allows a two-stage vehicle, the size and weight of a B777 at takeoff, to deliver the same pressurized cargo to the ISS as the Space Shuttle.

For more information on Andrews Space & Technology and subsonic in-flight LOX collection, please visit our web site at: www.andrews-space.com.

1. J L Leingang, L Q Maurice and L R Carreiro, 1996, " InFlight Oxidizer Collection Systems for Airbreathing Boosters", from Developments in High-Speed-Vehicle Propulsion Systems, pp 333-384, Vol 165, AIAA Progress in Aeronautics & Astronautics
2. V V Balepin, 1996, " Air Collection Systems", from Developments in High-Speed-Vehicle Propulsion Systems, pp 385-419, Vol 165, AIAA Progress in Aeronautics & Astronautics
3. Andrews Space & Technology, 2001, "Future Space Transportation Study Final Report", pp ii-iv, Andrews Space & Technology
4. NASA, 1994, "Commercial Space Transportation Study Final Report", pp 393-402, NASA