11th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference AIAA 2006-7031 6-8 September 2006, Portsmouth, Virginia SACD’s Support of the Hyper-X Program Jeffrey S. Robinson* and John G. Martin† NASA Langley Research Center, Hampton, VA 23681 NASA’s highly successful Hyper-X program demonstrated numerous hypersonic air-breathing vehicle related technologies including scramjet performance, advanced materials and hot structures, GN&C, and integrated vehicle performance resulting in, for the first time ever, acceleration of a vehicle powered by a scramjet engine. The Systems Analysis and Concepts Directorate (SACD) at NASA’s Langley Research Center played a major role in the integrated team providing critical support, analysis, and leadership to the Hyper-X Program throughout the program’s entire life and were key to its ultimate success. Engineers in SACD’s Vehicle Analysis Branch (VAB) were involved in all stages and aspects of the program, from conceptual design prior to contract award, through preliminary design and hardware development, and in to, during, and after each of the three flights. Working closely with other engineers at Langley and Dryden, as well as industry partners, roughly 20 members of SACD were involved throughout the evolution of the Hyper-X program in nearly all disciplines, including lead roles in several areas. Engineers from VAB led the aerodynamic database development, the propulsion database development, and the stage separation analysis and database development effort. Others played major roles in structures, aerothermal, GN&C, trajectory analysis and flight simulation, as well as providing CFD support for aerodynamic, propulsion, and aerothermal analysis. Abbreviations BET Best Estimated Trajectory CFD Computational Fluid Dynamics DFRC Dryden Flight Research Center GN&C Guidance, Navigation & Control HXFE Hyper-X Flight Engine HXLV Hyper-X Launch Vehicle HXRV Hyper-X Research Vehicle IDT Integrated Design Team LaRC Langley Research Center NASP National AeroSpace Plane OML Outer Mold Line POST Program to Optimize Simulated Trajectories RTF Return to Flight SACD Systems Analysis and Concepts Directorate VAB Vehicle Analysis Branch * Aerospace Engineer, Vehicle Analysis Branch, M/S 451, Senior Member, AIAA † Aerospace Engineer, Vehicle Analysis Branch, M/S 451, Senior Member, AIAA 1 American Institute of Aeronautics and Astronautics This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. I. Introduction In 1996 NASA initiated the Hyper-X Program, a jointly conducted effort by the NASA Langley Research Center (LaRC) and the NASA Dryden Flight Research Center (DFRC), as part of an initiative to mature the technologies associated with hypersonic airbreathing propulsion1. Unlike its predecessor, the U.S. National Aero-Space Plane (NASP) program, Hyper-X was a very focused program offering an incremental approach to developing and demonstrating scramjet technologies. During the NASP program, attempts were made to develop and integrate many new, unproven technologies into a full-scale flight test vehicle. In hindsight, this was an overly ambitious goal that was both technically and programmatically unachievable, given the relative immaturity of the various technologies and the budgetary constraints of the time. By contrast, the primary focus of the Hyper-X program was the development and demonstration of critical scramjet engine technologies, using several small, relatively low cost, flight demonstrator vehicles. This philosophy was a direct outcome of NASA's "better, faster, cheaper" approach to flight projects and programs in general. The primary goals of the Hyper-X program were to demonstrate and validate the technologies, the experimental techniques, and the computational methods and tools required to design and develop hypersonic aircraft with airframe- integrated dual-mode scramjet Figure 1. 3-view of Hyper-X Research Vehicle propulsion systems. Hypersonic airbreathing propulsion systems, studied in the laboratory environment for over 40 years, had never been flight tested on a complete airframe integrated vehicle configuration. Three Hyper-X flight test vehicles (Figure 1), the first two of which were to fly at Mach 7 and the third at Mach 10, would provide the first opportunity to obtain data on airframe integrated scramjet propulsion systems at true flight conditions. Prior to contract award in 1997, systems analysis experts from VAB, working with their counterparts from McDonnell Douglas in St. Louis, performed the initial systems analysis studies that led to the Hyper-X vehicle design. Hyper-X gets its shape, or outer mold line (OML) design, from the Mach 10 Dual-Fuel Global-Reach vehicle2,3. VAB engineers were involved in developing the keel line, OML, baseline propulsion and aerodynamic databases, design loads, mechanical design, thermal analysis, vehicle performance, stage separation analysis, and basic system requirements, all of which became part of the government furnished items at the time of contract award in the spring of 1997. After the contract to finalize the design and build the three flight vehicles was awarded, engineers in VAB continued to support the program as the preliminary design evolved through membership in the program’s Integrated Design Teams (IDTs). 2 American Institute of Aeronautics and Astronautics II. Propulsion Support in the propulsion discipline included participation in engine redesigns and analyzing engine ground test data4,5. Effective utilization of scramjet engines requires careful integration with the air vehicle. This integration synergistically combines aerodynamic forces with propulsive cycle functions of the engine. Due to the highly integrated nature of the hypersonic vehicle design problem, the large flight envelope, and the large number of design variables, the use of a statistical design approach in design is effective. Modern Design-of- Experiments (MDOE) was used throughout the Hyper-X program, for both systems analysis and experimental testing. One specific example was the flush wall injector design, which engineers from VAB were directly involved with. Independent parameters selected for this study included fuel injection angle, injector total pressure, fuel equivalence ratio, fuel split, injector to gap spacing, mach number, and combustor length. A test point matrix was defined for a minimum, nominal, and maximum value for each of the independent variables. This MDOE study used three- dimensional CFD to solve the scramjet combustor reacting flow fields. Forebody and inlet CFD solutions provided initial conditions for the 3-D combustor CFD. A limited number of 2-D nozzle solution where performed, using the combustor solutions for inflow conditions, to characterize the nozzle performance. Over a dozen responses were generated from the study including mixing efficiency, combustion efficiency, combustor total pressure recovery, nozzle coefficient, combustor static pressure, combustor Mach number, and other engine performance related parameters. Each of these responses was post processed from individual CFD solution data planes, or combustor cross-sections, or other CFD output files. Finally, regression equations were developed for each of the responses and were used to help guide the injector redesign. Figure 2. SRGULL analysis with flight and HXFE. Figure 3. Mach 7 acceleration predictions. VAB’s propulsion engineer also had the primary responsibility of producing the official predicted engine performance databases for the Mach 7 flight experiment6. This database along with the Mach 7 engine conceptual design, ground test data analysis and post-flight data analysis were performed using an integrated engine performance analysis code called SRGULL. The database evolved with time due to the execution of the ground test program and CFD analysis efforts. The final pre-test version of the database derived its combustion efficiency from the Hyper-X Flight Engine (HXFE) ground tests but was also decremented to account for the 3 American Institute of Aeronautics and Astronautics expected reduced efficiency in flight for “clean” air. SRGULL results for the “on point” test condition are shown in Figure 2. Three flight Mach numbers, three angles of attack, four dynamic pressures, and six fuel equivalence ratios were included in the database and supplied to the X-43A simulation team who used the propulsion database in conjunction with other vehicle data to construct the vehicle control laws and fly simulated trajectories. These simulations were key to insuring a successful flight test. In addition to the nominal engine performance predictions, a range of minimum and maximum expected values of engine axial force, normal force and pitching moment for each of the points in the database was also provided. Uncertainty values were chosen to try and capture the extremes of possible engine performance. These uncertainty values were necessary so that Monte Carlo analysis could be performed to assess potential flight test results and to stress the flight and engine control laws. Figure 3 shows the pre-flight expected acceleration during the engine experiment overlaid with actual flight data. III. Aerodynamic Database VAB’s aerodynamics experts held prime responsibility for the development of the research vehicle’s aerodynamic database. Coordinating and compiling data from thousands of wind tunnel runs in numerous facilities with the results from engineering predictive analyses and CFD calculations to generate the vehicle’s aerodynamic database was a highly complex problem that was critical to flight success. The Hyper-X aerodynamic database7,8 was comprised of data which supported the mission through all phases of flight, as shown in Figure 4, beginning with the HXLV dispense from the B-52, the ascent of the HXLV to the test condition, the separation of the X-43A vehicle from the HXLV, the engine test including the powered and unpowered post test tare measurements, and the descent of the research vehicle to subsonic terminal conditions. Figure 4. Snapshot of various wind tunnel tests, models, and facilities needed showing the regions of the Hyper-X flight profile that each test covered. 4 American Institute of Aeronautics and Astronautics The Orbital Sciences Corporation was responsible for development of the HXLV and all of its associated databases (though much of the data was generated in LaRC wind tunnels), while VAB engineers took the lead on developing the aerodynamic databases for the HXRV. The most complicated of these databases was by far the stage separation database. The X-43A stage separation from the HXLV, which occurred at the extreme environmental conditions associated with flights at Mach 7 and 10, and dynamic pressure of approximately 1000 psf, was a complicated dynamic event which had to be executed precisely so as not to upset the X-43A in a manner such that it could not obtain the steady, controlled flight conditions required to conduct the scramjet engine test. A series of wind tunnel tests were conducted to characterize the aerodynamic forces and moments associated with this two body, mutual interference separation problem. Preliminary estimates of the aerodynamic interference effects were obtained by modifying a wind tunnel model of an early X-43A configuration to permit a non-symmetric HXLV conical adapter to be clamshell mounted directly on the model sting. Several screening tests conducted in the NASA Langley 20-inch Mach 6 and 31-inch Mach 10 wind tunnels provided a rapid initial assessment, but permitted only axial separation between the X-43A model and adapter (no relative vertical or lateral translation and no relative angular displacement), and measured only the effect of the adapter on the research vehicle aerodynamics (as opposed to the simultaneous mutual interference of the two bodies on each other). Once the proof of concept had been defined, a high fidelity, 8.33% scale, multi- component model, which included the entire HXLV configuration, was built and tested at the Arnold Engineering Development Center – von Karman Gas Dynamics Facility (AEDC-VKF) Tunnel B, at Mach 6 test conditions9,10. Six component force and moment data were obtained for both the X-43A and the HXLV booster + adapter combination in close proximity to each other. Early in the program, initial wind tunnel screening tests were conducted to determine the basic X-43A airframe aerodynamics, including stability, control, and performance characteristics. These "quick look" tests were conducted using small scale, rapid fabrication models in the Langley 20-inch Mach 6 and 31-inch Mach 10 facilities, the Boeing (formerly McDonnell Douglas) - St. Louis Polysonic tunnel, and the Boeing North American subsonic tunnel. As the vehicle design matured, additional testing was conducted using larger, higher fidelity models, with very fine gradations in control surface increments. Additional entries using the refined high fidelity models were made in the NASA Langley 16-ft Transonic facility (0.6 < Mach < 1.2), Unitary Plan Wind Tunnel facility (1.5 < Mach < 4.6), the 20-inch Mach 6, and the 31 -inch Mach 10 tunnels, in order to fully bracket the anticipated flight envelope. Due to the relatively small scale of these aerodynamic force and moment wind tunnel models, inlet-open testing (unpowered or powered using a simulant gas technique) was not possible. Again, many of these models and facilities are seen in Figure 4. A comprehensive CFD study was undertaken to provide estimates of the inlet-open unpowered and powered flight aerodynamic characteristics for the Mach 7 vehicles, including the effects of Mach number, angle-of-attack, and sideslip on the X-43A. A number of different CFD codes and tools were utilized to predict the airframe forces and moments associated with the inlet open flight conditions, including both unpowered and powered engine operation modes. These methods included CFD codes, both structured Euler and Navier-Stokes solvers, for external airframe analysis, and propulsion cycle analysis codes which model the scramjet combustion physics and flowpath processes. A sample of the resulting longitudinal aerodynamics is shown in Figure 5. 5 American Institute of Aeronautics and Astronautics Figure 6. X-43A longitudinal forces and moments – inlet Figure 5. X-43A Mach 6 longitudinal aerodynamic open unpowered and powered modes including validation characteristics. results from the 8-ft HTT. Effects of elevon position on the basic lateral-directional characteristics of the HXRV were investigated. At issue was the question of the effect of vehicle sideslip and the expanding propulsion plume acting over the vehicle aftbody, and whether or not the plume would tend to increase or decrease the configuration's basic lateral-directional stability characteristics. The effect of elevator position on the HXRV aileron control effectiveness was investigated, as were the effects of elevator position on the rudder effectiveness. Engineers from SACD were involved at each stage of testing and were ultimately responsible for scaling, adjusting and combining all of the wind tunnel results, CFD analysis, and other information to develop a unified aerodynamic database for each of the HXRV flight phases. Some validation of the inlet closed + inlet open increments + fuel on increments methodology, as well as the accuracy of the data in each of those data sets, was found with results from the HXFE tests in LaRC’s 8-ft High Temperature Tunnel. As shown in Figure 6, the buildup methodology and data agreed very well with the inlet open fueled and unfueled data from the HXFE tests. IV. Aerothermal Analysis VAB’s aerothermodynamics experts were heavily involved in the design and analysis of the research vehicle’s hot structural components, namely the vehicle nose, all-moving horizontal control surfaces, and vertical tails11. VABs engineers developed a unique approach to quantifying and accounting for some extremely complex aerothermodynamic phenomena such gap heating, corner flows, shock impingement, and shock-shock interaction, all of which can cause significant increases in localized heating. Engineers helped identify critical material shortcomings that eventually led to design changes for the third flight vehicle (Mach 10). As seen in Figure 7, the nose, horizontal tail, and vertical tail on the Hyper-X vehicle were all designed as hot structure components with no active cooling or additional thermal protection such as tiles. The nose and horizontal tails employed carbon-carbon leading edges while the vertical tails were constructed out of Haynes 230 alloy with Haynes leading edges for the Mach 7 flight and carbon-carbon leading edges for the Mach 10 flight. 6 American Institute of Aeronautics and Astronautics Figure 7. Hyper-X research vehicle showing hot structure Figure 8. Finite element mesh model for components. vertical tail and predicted temperature distribution for Mach 7 flight. The control surfaces on the HXRV, along with the vehicle nose, were subjected to significant aeroheating without the benefit of a thermal protection system and were therefore referred to as hot structures. The purpose of the aeroheating and thermal analysis process was to predict the probable heating loads on the hot structure components and the resultant temperatures. From these temperatures and gradients, a structural analysis could be performed which would demonstrate the deflections and stresses in the material. The temperature predictions would then be used to verify that all materials remained within their allowable temperature ranges. To produce heat fluxes, the trajectory was discretized for local maxima and minima in the variables that impact heating. Doing so produces a piecewise linear representation of the trajectory. Variables of interest include Mach number, vehicle angle of attack, dynamic pressure, and control surface deflection. Using a variety of engineering level codes, validated by CFD, heat loads were computed at 20 discrete time steps along the 130-second preflight trajectory. 3-D maps of heat fluxes (q) were generated for each of the hot structure components at each given time point in the trajectory. These heat fluxes were obviously dependent on the skin temperature of the component. Thus, after a solution was run for a given set of q maps, the temperature maps were provided back to the aerothermal codes, and new q maps based on the latest temperature prediction were run. Several (3-4) iteration loops were usually required for this cycle to come to closure. Closure was defined as when temperatures between solution sets were varying less than 10°F. Figure 8 shows the finite element mesh model for the HXRV vertical tail as well as the predicted temperature distribution for one of the points in the Mach 7 flight profile. A similar process was followed to analyze the flight data from the Mach 7 flight (and was performed for the Mach 10 flight). Heat loads and the associated thermal responses were produced at 27 points along the Best Estimated Trajectory (BET). The Mach 7 nose had a single thermocouple located on a butt line 0.50 inch off the vehicle centerline and at a fuselage station 0.5-0.7 inches aft of the leading edge. A comparison of the flight data and the postflight analysis results is shown in Figure 9 with two heating estimates believed to bracket the actual thermo- couple location. Two thermocouples were installed in the Mach 10 carbon-carbon nose at 1.0 7 American Institute of Aeronautics and Astronautics inch and 1.5 inches aft of the physical nose and near the vehicle centerline. Figure 10 shows good agreement between flight data and preflight predictions made without applying uncertainty factors. The quality of the Mach 10 comparison is similar to that shown for the Mach 7 leading edge. Figure 9. BET predicted temperature profiles for the Figure 10. Preflight Mach 10 predicted nose temperatures vehicle nose for the Mach 7 flight with flight data with flight data. overlaid. V. Stage Separation Engineers from VAB also led the entire stage separation effort12. As part of the baseline mission, the research vehicle was required to separate from Pegasus-derived rocket booster near the scramjet test condition. Such a separation of two non-axisymmetric vehicles, as shown in Figure 11, in a high Mach number, high dynamic pressure environment had never been done before. This was a huge task and a problem of such complexity, that it actually made making the scramjet work seem simple. Thousands upon thousands of wind tunnel runs measured the interference effects that the two bodies in close proximity had on each other. Tests were conducted to characterize separation mechanisms from which detailed models were built and incorporated into the Figure 11. Depiction of X-43A stage separation. simulation. Additional tests helped to evaluate event sequence timing. A 15 degree-of-freedom (6-HXLV, 6-HXRV, 2-ejector pistons, 1-rotating drop jaw) simulation of the separation event was developed by VAB engineers and contractors, and hundreds of thousands of simulations were run to examine potential outcomes and to help determine the best flight controls approach. This tool, called SepSim, models all of the vehicle dynamics, separation mechanics, and aerodynamics for both vehicles and uses an industry standard simulation code called ADAMS to supply general multi-body equations of motion, simulation integration, and input/output capabilities. The SepSim team coded user subroutines for aerodynamic forces and moments, control system characteristics, atmosphere modeling, bolt 8 American Institute of Aeronautics and Astronautics and ejector piston characteristics, and HXLV and HXRV actuator characteristics. By definition, the separation event started when the command was given to blow the explosive bolts attaching the two vehicles and ended 2.5 seconds later when the HXRV had cleared the influence of the HXLV and had recovered to the desired state to begin the engine test. Engineers used SepSim to help develop the control law strategy and feedback loop closure timing for the separation event. At the command to separate, the HXRV actually flew open loop with the horizontal control surfaces moving from their fixed zero degree deflection position to a set position (based on vehicle angle of attack) near what would be the trim position for the HXRV in free flight at that flight condition (cowl closed). Moving at the actuator rate limit, the horizontal control surfaces reached this deflection near the end of the piston push (0.1 seconds into separation). Feedback loops for the sideslip and roll rates were then faded in, followed closely by pitch rate. At the same time, the HXLV was commanded to a hard pitch down to minimize the risk of re-contact. Attitude feedback loops on the HXRV were then faded in, commanding roll and sideslip angles to zero degrees and angle of attack to the test point target of 2.5 degrees. All loops were closed and the HXRV was under full autonomous control, flying free of the HXLV’s influence within 500 ms of the initiation of the separation event. All of this complex sequencing and timing was designed using the SepSim model that VAB engineers helped build. In order to obtain a more comprehensive validation of SepSim, and thus satisfy one objective of the Return-To-Flight activity, VAB engineers developed an independent simulation of the Hyper-X stage separation using the Program to Optimize Simulated Trajectories (POST)13. Previous independent simulation validation efforts relied on SepSim to generate a “delta” force and moment model to account for aerodynamic interference effects during separation. This model was incorporated into the 6-DoF simulation DFRC had built to model the HXRV in free flight, and was then used to “validate” SepSim results. Recent enhancements to POST II allowed for the construction of a completely independent simulation that could validate the SepSim results and verify the numerous simulation models used to predict the flight behavior of both vehicles. Also, by utilizing software specialized for trajectory calculations, it was possible to develop a simulation that provided similar results as SepSim, but was simpler and more concise. As such, the POST II simulation took much less CPU time to execute than the higher fidelity Figure 12. Comparison of angle of attack profile statistics from Monte Carlo analysis between SepSim which made it well suited for conducting SepSim and the POST II simulation. trade studies and sensitivity analyses that required the running of numerous Monte Carlo cases, thus providing the project with an additional analysis tool. A comparison of the two simulation tools is shown in Figure 12. VAB engineers also performed post-flight analysis of both the Mach 7 and Mach 10 flights, comparing flight performance to both simulation models and pre-flight Monte Carlo predictions14. Figure 13 shows a comparison of the axial acceleration expected during the piston push for the Mach 7 flight overlaid with the actual flight data. Good agreement indicated here and in other comparisons indicate that many of the components of the separation event were 9 American Institute of Aeronautics and Astronautics being modeled accurately. However, performance of the control system during the Mach 7 flight showed some lag in the system response to attain the desired angle of attack. This characteristic was traced to an under-prediction of Cm . This effect was corrected for in the Mach 10 flight, 0 and as shown in Figure 14, performance was well with the expected range. Figure 13. Axial acceleration profile during piston Figure 14. Comparison of Mach 10 HXRV angle of push for the Mach 7 flight. Flight data vs. pre-flight attack time history during separation. Flight data vs. Monte Carlo. pre-flight Monte Carlo. VI. Structures Structural engineers from VAB performed numerous critical tasks over the life of the program including dynamic analysis of the launch stack15. Their analysis identified critical problems with longitudinal and lateral bending frequencies. VAB’s engineers, through exhaustive analysis, suggested 0.1 design changes which, when Flight 3 Data Reduction implemented by the program, 8.2 Hz proved crucial to raising the 0.08 launch stack frequencies above those required by the flight 8.3 Hz analysis 1st pitch bending control system. This required 0.06 structural modifications and material changes. To confirm Amplitude their predictions, the project 0.04 initiated several vibration tests which VAB engineers helped 18.0 Hz analysis 2nd pitch bending 17.3 Hz to design. They also analyzed 0.02 the results which convinced them that they were correct and that their suggested design 0 changes would work. Data 0 10 20 30 40 50 gathered from the first failed Frequency [Hz] flight as well as from the Figure 15. Frequency content for Flight 3 Z-acceleration. second and third successful flights ultimately proved that their analysis was accurate. Figure 15 shows the pitch frequency content for the Mach 10 flight. As shown, a first mode bending frequency of 8.2 Hz was seen in 10 American Institute of Aeronautics and Astronautics