What s Cheaper To Fly: Rocket or TBCC? Why?

What s Cheaper To Fly: Rocket or TBCC? Why? Michael J. Kelly 1, Ronald P. Menich 2, and John R. Olds 3 SpaceWorks Engineering, Inc. (SEI), Atlanta, GA, 30338 Cost estimating for large aerospace projects has long been considered a difficult problem. Often, for reusable launch vehicle (RLV) programs, the ongoing maintenance and other ground operations costs end up dominating the life cycle cost estimates. This presents the conceptual designer with a pair of challenges. First, how does one consistently and reliably estimate these recurring costs? Second, what sorts of vehicle design decisions can one make to minimize those costs? SpaceWorks Engineering, Inc. (SEI) has developed, through research and quantitative modeling, a discrete event simulation to predict the time and resources required to maintain a fleet of RLVs for orbital missions. The model analyzes the various processes that must take place, and the team of technicians needed to perform those tasks. Built using the software Arena and MS Excel, this model, Descartes-Hyperport, is currently being utilized by various government teams to generate life cycle cost estimates. As the model has developed and been exercised more thoroughly, the development team has noticed patterns in the output files, pointing towards key vehicle criteria that can help or hurt the various program metrics. This paper uses the examples of two vehicles, one a rocket-based two-stage-to-orbit RLV, the other a turbine-based combined-cycle (TBCC) concept, both designed for the same mission profile, to guide a discussion of these patterns. Does the relative simplicity of the all-rocket vehicle make maintenance easier? Are there performance advantages to the combined cycle design that allow for other systems to be improved? What are the primary drivers of program cost, and what are the decisions any conceptual designer can make to reduce those costs? With both the theoretical underpinnings and the software capacity to allow for quick, efficient analysis of a wide trade space, vehicle design can now include ground operations metrics as an early-stage figure of merit. This can lead to lower costs, and streamlined performance, for the next generation of reusable access to space. Nomenclature APU = Auxiliary Power Unit DDT&E = Design, Development, Test, and Evaluation DES = Discrete Event Simulation DMSJ = Dual-Mode Scramjet FOM = Figure of Merit JSS = Joint Systems Study MMH = Maintenance Man-Hour NRA = NASA Research Announcement RBCC = Rocket-Based Combined-Cycle RLV = Reusable Launch Vehicle SEI = SpaceWorks Engineering, Inc. SSME = Space Shuttle Main Engine TBCC = Turbine-Based Combined-Cycle TPS = Thermal Protection System TSTO = Two-Stage-To-Orbit VBA = Visual Basic for Applications 1 Operations Engineer, Engineering Economics Group, 1200 Ashwood Parkway Ste. 506, AIAA Member. 2 Space Scenario Analyst, Engineering Economics Group, 1200 Ashwood Parkway Ste. 506, AIAA Member. 3 Principal Engineer, 1200 Ashwood Parkway Ste. 506, AIAA Associate Fellow. 1

S I. Introduction INCE the second half of 2008, groups within NASA s Langley Research Center and the Air Force Research Laboratory, along with several contractors have been working in parallel to develop two-stage reusable vehicles concepts capable of carrying 20,000 lbs. of payload to low Earth orbit. (Specifically, to 100 nautical miles, circular, 28.5 degrees inclination.) This Joint Systems Study (JSS) will ultimately compare vehicle concept designs based solely on rocket propulsion, concepts incorporating rocket-based combined-cycle (RBCC) propulsion, and concepts incorporating turbine-based combined-cycle (TBCC) propulsion. All concepts would call for production of three complete flight units, and as a fleet those three vehicles would perform twelve missions per year. The goal of the JSS is to design various vehicles to meet the same performance standards (margins, reserves, payload, etc.), so that comparisons can then be drawn for a large number of figures of merit (FOMs), including cost-based metrics. These comparisons can fairly be attributed to the differences in propulsion and configureation, since the vehicles will otherwise be similar. SpaceWorks Engineering, Inc. (SEI) has taken on various roles with the multiple teams working on the JSS. Among these roles, SEI has performed analyses of the projected costs of some of the vehicles, including up-front costs such as design, development, test, and evaluation (DDT&E) and theoretical first unit production, as well as the ongoing costs of ground operations and the potential future costs of vehicle failures. In this full life-cycle cost (or campaign cost ) analysis, the ground operations tend to dominate all others over the projected 30-year life of a program. SEI uses a tool called Hyperport to conduct operations cost analysis. While the joint systems study has yet to be completed, this paper discusses the analyses performed so far on the ground operations costs of some of the concepts being studied, and what conclusions can already be drawn from this work. II. What is Descartes-Hyperport? In early 2008, SpaceWorks began work on a NASA Research Announcement (NRA) contract to develop discrete event simulation (DES) models to aid cost estimation for next-generation reusable launch vehicles (RLVs). The specific DES methodologies employed by the SpaceWorks team were referred to as Descartes. Central to these methodologies is the utilization of Rockwell Automation s simulation software package called Arena, which allows for easy manipulation of DES entities, processes, and resources, as well as including a Visual Basic for Applications (VBA) module. Each Descartes model built includes an Arena model (.doe) file, in which the developers construct a modular representation of the process flows involved. Any numerical parameters of that model (such as process times, resources required for processes, etc.) are abstracted to variable arrays. VBA is used to both import starting values for many of the variables from a MS Excel input file, as well as to export final values for many other variables to a separate output Excel file. The Excel files themselves are as vital to the Descartes model as the.doe file. The complete flow of information starts in a user-input sheet in the input file. A user fills in relevant information about the vehicle being modeled, then runs a Visual Basic macro. In early models, this was a VBA macro, but it is now VB.net since new models are being built using Excel 2007. This macro takes the user information and, using both built-in formulas as well as a custom database derived from historical analogies, generates values for a wide range of variably arrays. As described above, these are grabbed by the VBA code in the.doe file, which then runs for some moderately large number of statistically-independent replications. Each replication s outputs are exported to the output Excel file. After all replications are over, Arena closes and the data is summarized in a single spreadsheet view. A user can then go back to the input sheet and change whatever values they need, and re-run the model, saving to a new output file for comparisons to the original. Using the example of Hyperport, this information flow is shown in Figure 1 below. 2

Figure 1. Sequential flow of information in Hyperport, typical of Descartes models. Descartes-Hyperport was the first full model built under the NRA contract, and was completed in August 2009. The purpose of Hyperport is to model the ground operations required at facilities used to turn around reusable launch vehicles, analogous to the Orbiter Processing Facility, Vehicle Assembly Building, and shuttle launch pads at Kennedy Space Center. There are over 300 individual processes represented in Hyperport, from the vehicle s landing and taxiing, through all aspects of subsystem inspection and repair, on to vehicle assembly, fueling, and, ultimately, the next launch. Special attention was paid to the modeling of propulsion systems and thermal protection systems (TPS), since the team s knowledge of the shuttle program implied that these would be relatively laborintensive portions of the complete turnaround process. Also modeled in the primary vehicle turnaround phase were inspection and repairs of power systems, avionics, environmental controls, hydraulics and actuators, reaction control systems, mechanical and pyrotechnic systems, cockpit and crew cabins, orbital maneuvering systems, thermal controls, and payload. Hyperport can simulate turnaround of vehicles with either one or two reusable stages, as well as expendable stages. A reusable upper stage (or single stage) vehicle can include the full list of subsystems, any combination or rocket, turbine, and ram- or scramjet propulsion, and up to five TPS materials. A booster stage can have any of the above other than orbital maneuvering, thermal controls, and payload. Figures 2 and 3 show screenshots of the top-level of the Hyperport model and the orbiter turnaround submodel, respectively. The small white boxes arranged in a column within Figure 3 are a second layer of submodels, which contain the subsystemspecific process models. One of the primary challenges in building Hyperport was the collection of valid data to fill in the historical analogy database (DescartesData.xls in Figure 1). Since there has only been one true RLV program with real operational history, obviously the space shuttle program was a major source. However, changes in technologies have led to most current operations analysis models (such as Boeing s SOVOCS and NASA Langley s RMAT) predicting much faster, cheaper turnarounds than those experienced by the shuttle. The developers weighed these and other factors when building the Hyperport database. For reference, the space shuttle takes between 200 and 400,000 maintenance man-hours (MMH), depending on how many tasks are included in the estimate, from one flight to the next. If Hyperport is used to simulate the shuttle program, using only shuttle-derived analogy data, the model produces an estimate of approximately 250,000 MMH. Using the regular database, which incorporates other sources, the shuttle estimate is reduced to about 60,000 MMH, so this number would serve as a better comparison point for the other vehicles discussed in this paper. 3

Figure 2. Screenshot of Hyperport.doe Figure 3. Screenshot of orbiter turnaround submodel For more details on the development and structure of Hyperport, see reference #1. 4

III. Baseline Vehicle Analyses As mentioned in the introduction, due to involvement in the Joint Systems Study, SEI has access to a set of vehicle concepts designed to perform very similar missions. To date, SEI has performed various analyses on a rocket-based two-stage-to-orbit (TSTO) vehicle and on a TSTO vehicle with a TBCC booster and rocket-powered orbiter. Neither vehicle is manned. Both are fully reusable. Both are intended to complete similar missions, namely that a fleet of three vehicles would make twelve flights per year, carrying 20,000-pound payloads to a 100-nautical mile circular low Earth orbit. A. All-Rocket Vehicle Description The All-Rocket concept, as it is known, consists of two cylindrical-fuselaged winged bodies that are mated piggy-back style and launched vertically, with staging in the Mach 5-6 range. The booster is 120 feet long, the orbiter measures 102 feet. The booster is powered by a set of 4 LOX/RP engines, each producing about a quarter more thrust than a space shuttle main engine (SSME). The orbiter is powered by four scaled-down versions of the same engine design, in this case with about a third of the thrust of an SSME. Some details about the vehicle are restricted at the time of this paper and are thus omitted. When using Hyperport for operations analysis, the first step to studying the ground operations costs of any vehicle is to fill in the necessary vehicle information into the Hyperport.xls user input sheet. For the booster, this included all the values in Table 1. For the orbiter, it included the values in Table 2. The user input sheet also requires a set of program parameters, including things like fleet size, facilities available, and some basic calendar parameters. For both the All-Rocket and TBCC vehicles, all of these parameters have the same values. The fleet sizes were set at 3 orbiters and 3 boosters. The facilities were defined as 3 turnaround spaces for each stage, 1 stage integration facility, and 1 launchpad space, all of which are available 250 days per year with two 8-hour shifts per workday. The learning rate was set at 85%. Airframes are set to spend 6 months at an off-site depot heavy maintenance facility once every 10 flights. Hourly labor rates, intended to represent fully-encumbered costs per employee, were set at $75 for general technicians, $90 for specialists, $135 for mission planning and engineering support personnel, and the ratio of support personnel to technicians was set at 7:1. This 7:1 number ends up driving a significant fraction of the total program cost, and is derived directly from the space shuttle program. We feel that while this ratio represents an area with significant room for improvement, it is somewhat concept-independent and depends rather on the culture of the organization operating the vehicle. Any program that can do better than 7:1 will probably save significant money over a program that cannot. While it is the hope of the authors that the next generation of NASA, Air Force, or private launch vehicles will improve on that ratio, the specific steps toward that improvement are not only beyond the scope of this paper, but beyond the scope of the Hyperport tool. Table 1: Hyperport input information for All-Rocket booster General Information Stage dry weight (omitted) Mission length 1 hour Thermal Protection Advanced C-C Leading Edges 250 sq. ft. Electrical Number of high-voltage batteries 4 Number of fuel cell stacks 0 Number of APUs 0 Different voltage systems 2 Hydraulics Centralized hydraulics? No Number of sets of actuators 10 Reaction Control Surfaces Forward RCS thruster quantity 10 Forward RCS tank quantity 5 Aft RCS thruster quantity 0 Aft RCS tank quantity 0 Rocket Propulsion Number of engines 4 Engine thrust (omitted) Engine complexity Somewhat less than SSME Fuel, oxidizer RP-2, LOX Total fuel burned in flight (omitted) How thorough is the IVHM? Very thorough 5

Table 2: Hyperport input information for All-Rocket orbiter General Information Stage dry weight (omitted) Mission length 24 hours Thermal Protection Advanced C-C Leading Edges 250 sq. ft. AETB tiles w/tufi 1,945 sq. ft. PBI blankets 900 sq. ft. Conformal Reusable Insulation 5,450 sq. ft Electrical Number of high-voltage batteries 4 Number of fuel cell stacks 0 Number of APUs 0 Different voltage systems 2 Hydraulics Centralized hydraulics? No Number of sets of actuators 10 Reaction Control Surfaces Forward RCS thruster quantity 10 Forward RCS tank quantity 5 Aft RCS thruster quantity 10 Aft RCS tank quantity 5 Orbital Maneuvering Systems Number of OMS engines 0 Rocket Propulsion Number of engines 4 Engine thrust (omitted) Engine complexity Somewhat less than SSME Fuel, oxidizer RP-2, LOX Total fuel burned in flight (omitted) How thorough is the IVHM? Very thorough Payload Weight 20,000 lbs. Volume 3393 cu. ft. B. Baseline All-Rocket Analysis The first set of calculations performed by the VBA macros produces estimates of the maintenance man-hours required for each of the processes in the Arena model. The actual estimate produced is for the MMH needed to turn around the first vehicle, and then a learning curve is applied to (most) of the processes for each successive vehicle. The output provided at the end of the simulation gives a calculated average number of hours over a full simulated program. Since these MMH numbers are constant, no matter how many technicians are available to do the work (and, therefore, what flight rate can be achieved), they are the first results found during any Hyperport-driven operations analysis. For the All-Rocket vehicle, estimates of average MMH required for turnaround are: Table 3: Maintenance Man-Hour estimates for All-Rocket vehicle Total Vehicle Turnaround 18,220 Landing/Safing Activity 404 Stage Integration and Preflight 1,621 Booster Orbiter Total Stage Turnaround 6,299 9,897 Airframe Turnaround 3,145 8,527 Thermal Protection 133 4,148 Power Systems 725 725 Avionics 153 288 Environmental Controls 227 227 Hydraulics and Actuators 800 800 Reaction Control Systems 373 746 Mechanical Systems 733 838 Thermal Controls - 423 Payload Loading - 333 Orbital Maneuvering Systems - 0 Rocket Engines 3,154 1,369 6

As stated previously, the mission for this vehicle calls for a flight rate of 12 flights per year for the entire fleet. Using the MMH information available after the first model run, SpaceWorks began estimating the number of technicians of each specialty that would be needed to reach that rate. The labor quantities were adjusted through multiple iterations until a set of technicians was found that could achieve this flight rate for the minimum possible cost. The solution reached after these iterations required a total of 84 technicians per shift (or 168 per day), including 6 TPS specialists, 5 electricians, 9 mechanics, 11 propulsion specialists, 36 general turnaround technicians, 10 stage integration technicians, and 7 other specialists split between various areas. The reason for the large quantity of generalists lies in the structure of the model, where many turnaround processes require at least one specialist to be initiated, and beyond that require a minimum ratio of specialists to generalists to continue working. Since about 26% of the stage turnaround work is on TPS, one could think of an average of 9 of the 36 general technicians working on TPS at any given time. However, their designation as generalists gives a manager more flexibility in scheduling, and the simulation reflects this flexibility. Given the 7:1 ratio of support staff to technicians, combined with the costs associated with each class of labor, the estimated total labor cost per flight of the all-rocket vehicle came out as $28.7 million. Adding in estimated costs for fuel and for spare parts (also calculated in Hyperport), the total cost per flight estimate stands at $31.5 million. C. Turbine-Based Combined-Cycle Vehicle Description The TBCC concept is structured a little bit differently from the All-Rocket. Both stages are still winged bodies, but with less regular shapes. The 224-foot-long booster has a flat top to accommodate the orbiter, which is mounted on the top, and the bottom of the fuselage contains the flow-paths for the 6 turbines and 3 dual-mode scramjets (DMSJs). The booster also has a row of tail rockets, which, at the time of this analysis, included 8 individual engines. The 131-foot-long LOX/Hydrogen rocket-powered orbiter is less different in shape from the All-Rocket orbiter, but it does include a vertical tail, and is somewhat heavier. Like its All-Rocket counterpart, it was a set of 4 tail rockets. The vehicle takes off horizontally, using the booster s rockets and turbojets. It switches to DMSJ propulsion around Mach 3, with the rockets turning back on for staging at Mach 10. The rest of the Hyperportrelevant design details are given in tables 4 and 5. Table 4: Hyperport input information for TBCC Booster Stage General Information Stage dry weight 381,636 Mission length 2 hours Thermal Protection Advanced C-C Leading Edges 90 sq. ft. Structurally-Integrated TPS 11,700 sq. ft. FRSI blankets 25,758 sq. ft. Electrical Number of high-voltage batteries 2 Number of fuel cell stacks 0 Number of APUs 4 Different voltage systems 2 Hydraulics Centralized hydraulics? Yes Number of sets of actuators 12 Reaction Control Surfaces Forward RCS thruster quantity 6 Forward RCS tank quantity 3 Aft RCS thruster quantity 12 Aft RCS tank quantity 6 Rocket Propulsion Number of engines 8 Engine thrust 112,000 Engine complexity Less than SSME Fuel, oxidizer Hydrogen, LOX Total fuel burned in flight 245,000 lbs. How thorough is the IVHM? Moderate Turbine Propulsion Number of engines 6 Total fuel burned in flight 72,000 DMSJ Propulsion Number of engines 3 Total fuel burned in flight 40,000 Inlet variable geometry? Yes Integration into airframe Partially integrated How thorough is IVHM? Very thorough 7

Table 5: Hyperport input information for TBCC Orbiter Stage General Information Stage dry weight 77,322 lbs. Mission length 24 hours Thermal Protection Advanced C-C Leading Edges 47 sq. ft. Structurally-Integrated TPS 3,618 sq. ft. TABI blankets 6,034 sq. ft. PBI blankets 140 sq. ft Electrical Number of high-voltage batteries 0 Number of fuel cell stacks 3 Number of APUs 4 Different voltage systems 2 Hydraulics Centralized hydraulics? Yes Number of sets of actuators 9 Reaction Control Surfaces Forward RCS thruster quantity 6 Forward RCS tank quantity 3 Aft RCS thruster quantity 12 Aft RCS tank quantity 6 Orbital Maneuvering Systems Number of OMS engines 2 Rocket Propulsion Number of engines 4 Engine thrust 112,807 Engine complexity Somewhat less than SSME Fuel, oxidizer Hydrogen, LOX Total fuel burned in flight 37,000 lbs. How thorough is the IVHM? Very thorough Payload Weight 20,000 lbs. Volume 3,393 cu. ft. All of the facilities and workplace calendar assumptions, as well as learning rates and hourly costs, were placed at the same settings as for the All-Rocket. While looking over this table and comparing to the All-Rocket, one can already see several places where there are significant differences beyond the propulsion systems. For example, this concept obviously has a much larger quantity of TPS, as well as multiple power/electrical systems. Trades that explore the significance of these differences are discussed in the next section. D. Baseline TBCC Analysis The average MMH numbers captured from the TBCC vehicle are given in table 6. The minimum-cost solution that achieves the desired flight rate requires a total of 171 technicians per shift. This includes 10 TPS specialists, 15 electricians, 11 mechanics, 10 propulsion specialists, 70 general turnaround technicians, 32 stage integration technicians, and 23 other specialists. The total estimated labor cost is $58.3 million per flight, and the cost including fuel and spares is $68.8 million. 8

Table 6: Maintenance Man-Hour estimates for TBCC vehicle Total Vehicle Turnaround 39,078 Landing/Safing Activity 543 Stage Integration and Preflight 3,509 Booster Orbiter Total Stage Turnaround 21,201 13,825 Airframe Turnaround 11,850 12,652 Thermal Protection 5,369 3,997 Power Systems 2,957 3,704 Avionics 153 288 Environmental Controls 227 227 Hydraulics and Actuators 1,598 1,358 Reaction Control Systems 711 711 Mechanical Systems 835 941 Thermal Controls - 423 Payload Loading - 333 Orbital Maneuvering Systems - 671 Propulsion Turnaround 9,351 1,173 Rocket Engines 2,506 1,173 Turbine Engines 4,071 - DMSJ Engines 2,773 - IV. Comparing the All-Rocket and TBCC Vehicles Given any of the various top-level numerical comparisons, we can now answer the first question asked in the title of this paper. Whether comparing 18,220 MMH to 39,078, 84 technicians to 171, or $31.5 M to $68.8 M, it is clear that, given the assumptions inherent in this study, a purely rocket-based two-stage fully reusable RLV is cheaper to fly than a TBCC vehicle, by roughly a factor of two. While there are other analysis methods that may come to different specific numerical conclusions, the factor-of-two magnitude of the difference shown here represents what is likely to be an insurmountable gap when comparing these two vehicle concepts. But what are the subsystems driving this separation? The majority of subsystems on the TBCC vehicle are projected to take more time than their All-Rocket counterparts. In some cases, work required tends to scale upward with the size of the vehicle. In other cases, decisions made by the respective design teams could have pushed the TBCC vehicle higher. Rather than explore every subsystem individually, consider the average maintenance man-hours required for the top labor-intensive subsystems of each vehicle: Table 7: Top contributors to total MMH for both vehicles All-Rocket TBCC Subsystem MMH Subsystem MMH 1. Orbiter TPS 4,148 1. Booster Propulsion 9,351 2. Booster Propulsion 3,154 2. Booster TPS 5,369 3. Stage Integration/Preflight 1,621 3. Orbiter TPS 3,997 4. Orbiter Propulsion 1,369 4. Orbiter Power Systems 3,704 5. Orb. Mechanical Systems 838 5. Stage Integration/Preflight 3,509 Stage integration and preflight activities appear on both lists, and indeed the TBCC vehicle needs about twice as much work, but to a large extent stage integration can be considered inherently labor-intensive, and it seems intuitive that it would scale up with larger vehicles. The All-Rocket orbiter mechanical systems just edged several other areas on that vehicle for a spot in the top five, so they are not a major concern either. This leaves the TPS and propulsion as areas ripe for further discussion, along with the orbiter power systems due to their ranking on the TBCC table. The methodology taken by the authors to analyze the relative impact of these systems was to complete additional sets of Hyperport runs, based on changes made to the vehicle designs, primarily the TBCC. In each of these cases, 9

the only part of the vehicle changed was the subsystem in question, and that change s effects on other systems were mostly discounted. For example, a change in TPS materials would normally require aerothermodynamics analysis to ensure the new materials could handle the thermal loads of the planned trajectory. For purposes of this study, such considerations were ignored. 1. Reducing TBCC Thermal Protection The first scenario tested was a major reduction in thermal protection for the TBCC vehicle. While this would obviously be an impossible idea, all of the TPS areas other than the carbon-carbon leading edge covers were reduced to zero. This reduced the MMH required for the booster to 15,881, while the orbiter decreased to 9,854 hours. Including some propagating decreases in a few other areas, the total change was 10,267. The 12-flight-per-year solution required 136 technicians, a reduction of 35 (including 9 of the 10 TPS technicians and an additional 26 generalists). The new estimate of labor cost was $46.0 M, with a total cost of $56.4 M, both of which are reductions of about $12.3 M. These results are in line with what would be expected from back-of-the-envelope calculations based on the MMH numbers from the original vehicle table. TPS, in total, requires about 23% of the work during each turnaround, and eliminating TPS from the vehicle saves about 18% of the total cost. Some economics of scale, primarily driven by higher utilization of the workforce, account for the lower percentage savings in cost when compared to the savings in work. 2. Removing Air-Breathing Propulsion From TBCC While the rocket engines alone would not be powerful enough to complete the intended trajectory, the TBCC vehicle was re-configured as a rocket-only design by simply saying that the turbojets and DMSJs did not exist. While generally we ignored second-order effects while running these trades, one additional consideration was made here. The primary purpose of the auxiliary power units (APUs) on the TBCC booster is to power the variable geometry components of the air-breathing propulsion flowpaths, as well as the centralized hydraulics. For this reason, in addition to removing the engines from the simulation, we also reduced the APU count from 4 to 2 on the booster. Booster propulsion MMH reduced to just the 2,506 for the rocket, and booster power systems decreased from 2,957 hours to 1,769. The new technician count was 135, including 6 fewer propulsion specialists, 3 fewer electricians, and 27 fewer generalists. Labor cost per flight was down to $45.5 M, bringing the total down $12.7 M to $56.1 M. Again, a 23% reduction in MMH resulted in an 18% reduction in cost. 3. Removing Centralized Hydraulics From TBCC The large decrease from the last trade encouraged the team to explore another trade with easily-modeled beneficial side effects: removing centralized hydraulics from both stages of the TBCC, which would allow for reduction in the need for APUs on both stages. In practical terms this would involve replacing the centralized hydraulics with either electro-hydraulic actuators (EHAs), the technology used in the All-Rocket vehicle, or the next-generation technology of electro-mechanical actuators (EMAs). For the simulation, the change consisted of eliminating the centralized hydraulics and then decreasing the APU count from 4 to 2 on the booster (since some are still required for the variable geometry components) and from 4 to 0 on the orbiter (which assumes the fuel cells are sufficient for powering the other systems) cuts out a total of 4,840 MMH from the power and hydraulics subsystems, bringing the total of that group to 4,777. Technician count fell to 150 from the base of 171, costing $51.5 M per flight. The new total cost estimate of $61.9 M represents a 10% reduction driven by the 12% reduction in MMH. This gives some indication that the superior cost projections for the All-Rocket vehicle are not 100% due to the design decisions that are pre-determined as soon as the propulsion systems are determined. 4. Increasing All-Rocket Thermal Protection The SEI team then wanted to make sure the effects of these changes were symmetric. That is, would adding things to the All-Rocket have a similar effect as taking things off of the TBCC? Given how much savings could be generated on the TBCC if its TPS requirements were not so high, we checked what the effect would be on the All- Rocket vehicle if it had the same TPS requirements. While the All-Rocket orbiter s smaller size would make it impossible for it to even utilize all the square footage, we added it anyway resulting in new MMH estimates of 11,534 for booster turnaround and 9,746 for orbiter turnaround, leading to a total increase of 5,618 hours. This is smaller than the change to the TBCC above since the All-Rocket base case starts with more TPS than the leadingedge-only scenario gives to the TBCC vehicle. Additionally, due to different choices of materials, the TBCC orbiter TPS actually requires several hundred fewer hours of work than the baseline All-Rocket orbiter TPS. 10

The high-tps All-Rocket vehicle would require 103 technicians for turnaround, resulting in a labor cost of $35.1M per flight and a total cost of $37.9 M, which are increases of about $6.5 M. Here a 31% increase in MMH resulted in a 20% cost increase. These sorts of effects are similar to those observed when removing things from the TBCC vehicle. That is, the percent change in cost scales roughly with the percent change in work, but lags behind it 20-35%. 5. Removing Reusable Orbiters The last trade was a check to see how much could be saved if the orbiters of each vehicle were expendable. This would leave booster turnaround and stage integration activities intact. The results would therefore highlight the differences between only the boosters, which are the stages driving the All-Rocket and TBCC designations. They also can hint at the break-even point for expendable stage production cost in this kind of configuration. The TBCC booster with expendable orbiter would require 23,973 MMH for turnaround activities. This work could be done by 144 technicians, for a labor cost of $48.3 M per flight and a total cost of $57.6 M. The All-Rocket booster with expendable orbiter would require 7,432 MMH for turnaround activities. This work could be done by 34 technicians, for a labor cost of $11.1 M per flight and a total cost of $13.0 M. The large discrepancy between the two, and between the relative advantages of the two, further highlights that the complexities of the multiple propulsion systems, booster TPS, centralized hydraulics and APUs of the TBCC system make it much more expensive than the All-Rocket alternative. Table 8 summarizes the results from all of these trades. Not all subsystems are listed, only those that varied during at least one of the experiments. Within each column, the greyed numbers are those that match the baseline number for that vehicle. Table 8: Summary of results of trade studies TBCC All-Rocket Baseline Less TPS Rockets Only No Cent. Hydraulics Expendable Orbiter Baseline More TPS Expendable Orbiter Total MMH 39,078 28,810 31,314 33,730 23,973 18,220 23,838 7,432 TPS 5,369 49 5,369 5,369 5,369 133 5,369 133 Propulsion 9,351 9,351 2,506 9,351 9,351 3,154 3,154 3,154 Power 2,957 2,957 1,769 1,769 2,957 725 725 725 Hydraulics 1,598 1,598 1,598 960 1,598 800 800 800 TPS 3,997 25 3,997 3,997 0 4,148 3,997 0 Propulsion 1,173 1,173 1,173 1,173 0 1,369 1,369 0 Power 3,704 3,704 3,704 1,328 0 725 725 0 Hydraulics 1,358 1,358 1,598 720 0 800 800 0 Booster Orbiter Technicians 171 136 142 150 144 84 103 34 Labor Cost ($M) 58.3 46.0 47.4 51.5 48.3 28.7 35.1 11.1 Total Cost ($M) 68.8 56.4 58.2 61.9 57.6 31.5 37.9 13.0 V. Conclusions and Designing For Operations One goal of the Joint Systems Study is to draw meaningful comparisons between vehicles with different propulsion systems. To accomplish this goal, vehicles are being designed to perform the same mission utilizing different combinations of rockets, turbines, ramjets, and scramjets. While other design decisions made by independent JSS teams can influence the ultimate outcome, this paper is one of the first of (presumably) many that will attempt to compare the vehicles according to some set of FOMs to determine a winner and then explain the reason for the victory. There are several somewhat-independent conclusions that can be drawn from the results discussed in this paper. Primarily, as said in the last section, the All-Rocket vehicle as designed in this study will have lower ongoing costs than its turbine-based combined-cycle counterpart. A large fraction of this difference in cost can be attributed directly to the cost of maintaining the additional engines. The baseline cost of $68.8 million per flight for the TBCC vehicle would be directly reduced to $58.2 million if it were capable of flying under rocket power alone. Additionally, the thermal protection necessary for a vehicle as large as this, and with a staging point at Mach 10 that requires booster stage thermal protection, is adding about $14.2 million to the cost. Alternatively, adding that same 11

TPS to the All-Rocket vehicle would add $6.2 million, so the actual direct penalty to a hypothetical generic design is probably somewhere in between these two numbers. It is reasonable to assume that while some of this extra TPS cost is a direct result of the propulsion systems, the evidence presented by the larger TBCC orbiter requiring less work than the smaller All-Rocket orbiter suggests that there are other decisions including choices of materials that are likely to have significant impact on TPS operations costs. This leads into the second set of conclusions, concerning decisions that can be made during the development of any vehicle to reduce the total life cycle costs. In addition to considering operational impacts when selecting engines and TPS materials, vehicle designers should strive to eliminate the need for centralized hydraulics, and for auxiliary power units. While the specifics of these systems are beyond the scope of this paper, developments in independently-powered electro-hydraulic and electro-mechanical actuators can make them a very practical choice, and the operational benefit of avoiding lengthy inspections of hydraulic lines throughout a vehicle is substantial. Eliminating centralized hydraulics and any other systems that lead to a requirement for APUs also avoids having to work on these traditionally costly subsystems. Vehicles drawing their power from batteries, fuel cells, or, if they will be in orbit for extended periods of time, potentially solar cells, are much easier (and therefore cheaper) to maintain. The last set of trades, concerning the savings from the use of expendable upper stages, gave results that are a bit more difficult to generalize. In both cases, elimination of orbiter processing from the standard ground turnaround flow resulted in a significant financial savings, but the magnitudes varied greatly. To gain real insight into the value of that option, further work would need to be conducted to estimate the costs of producing expendable vehicles suited to this mission and these vehicle configurations, as well as any difference in DDT&E costs between reusable and expendable stages. These conclusions, along with the others still to come as the JSS teams continue their work, can be very useful to the aerospace conceptual design community. By designing complete vehicles and having multiple independent parties discuss and evaluate the designs, the participants in the JSS are providing baseline concepts to be used as comparison points in the future. We are also gaining insights by studying the differences between the vehicles, leading to specific conclusions that can point towards more successful designs in the future. As the United States, international partners, and private companies makes their plans for the next generation of space access, these vehicles and the work done on them may help point towards the best, cheapest solutions in the years and decades to come. 12

Glossary Discrete Event Simulation Discrete Event Simulation (DES), is a computationally efficient technique often used when modeling complex systems. Every interaction between the various entities and resources of a system is modeled as occurring at a discrete event in time. A schedule of these events is maintained. A computer program can then proceed through the schedule, determining the results of each event in turn (which often includes adding additional future events to the schedule) until the schedule is empty or the simulation is otherwise terminated. DES models are usually used probabilistically, and the simulation may be run numerous times, with statistical analysis performed on the outputs. Life-Cycle Cost / Campaign Cost Both of these terms, which are generally regarded as interchangeable, refer to the total, all-inclusive cost of, in this case, a space vehicle program. A complete life-cycle cost (LCC) estimate has to account for DDT&E, production, ongoing operations, and, if applicable, disposal cost at the end of the program. Good LCC estimates may also take into account metrics like vehicle reliability, and estimate total costs due to mission failures over the life of a program. Turbine-Based Combined-Cycle Turbine-Based Combined-Cycle (TBCC) is used as a descriptor of launch vehicles that contain both turbine engines and either ramjet or scramjet engines. In the TBCC vehicle described in this paper, there are turbines, dual-mode scramjets (thermal choke instead of mechanical), as well as rocket engines. Acknowledgments Sponsorship and financial support for major portions of this project are provided by NASA's Aeronautics Research Mission Directorate under 2007 NASA Research Announcement contract NNL08AA30C. The authors would specifically like to thank the Contract Officer s Technical Representative Mrs. Lakisha Crosby, the hypersonic technical point of contact Mr. Jeffrey Robinson, Mr. John Martin, and all of NASA Langley Research Center. The authors gratefully acknowledge the contributions of colleagues at SpaceWorks Engineering, Inc. who contributed to this paper. Mr. A.C. Charania (President of SpaceWorks Commercial at SEI) assisted with the development of Hyperport, along with intern Mr. Chris Cho. Mr. William J.D. Escher (SEI Affiliate) and Mr. Carl Ehrlich (SEI Affiliate) provided sage advice as well as references relevant to operational aspects of past aerospace programs. References 1 Kelly, M., Charania, A., Olds, J., and DePasquale, D., Discrete Event Simulation of Reusable Launch Vehicle Development, Acquisition, and Ground Operations, AIAA-2009-6484 Space 2009, Pasadena, California, September 14-17, 2009. 13