Analysis of Aeroheating Augmentation and Control Interference Due to Reaction Control System Jets on Blunt Capsules. by Artem Alexander Dyakonov

Transcription

1 ABSTRACT DYAKONOV, ARTEM ALEXANDER. Analysis of Aeroheating Augmentation and Control Interference Due to Reaction Control System Jets on Blunt Capsules. (Under the direction of Dr. Fred R. DeJarnette.) Atmospheric entry capsules are frequently fitted with rocket nozzles as part of a Reaction Control System (RCS), which is used for attitude control and guidance maneuvers during entry. These rocket nozzles are installed on the rear wall of the capsule and the interaction of their exhaust with capsule s wake causes changes in aeroheating and aerodynamics. Changes in aeroheating may influence design of Thermal Protection System (TPS) and material secection, while changes in the surface pressure can cause unbalanced moments on the capsule and interfere with the native RCS control authority. Aerodynamic initerference of RCS must be understood and bounded for a sound controller design. The method to analyze aerothermal and aerodynamic effects of RCS on blunt capsules, presented in this dissertation, has been used in analyses of Mars Phoenix, Mars Science Laboratory (MSL) and Crewed Exploration Vehicle (CEV) RCS. As a result a working RCS design paradigm has been developed. This paradigm has been successfully applied to the design of RCS of Mars Science Laboratory (MSL) entry capsule.

2 Analysis of Aeroheating Augmentation and Control Interference Due to Reaction Control System Jets on Blunt Capsules by Artem Alexander Dyakonov A dissertation submitted to the Graduate Faculty of North Carolina State University in partial fulfillment of the requirements for the Degree of Doctor of Philosophy Aerospace Engineering Raleigh, North Carolina 2010 APPROVED BY: Dr. Hassan A. Hassan Dr. D. Scott McRae Dr. Jack R. Edwards Dr. Pierre A. Gremaud Dr. Fred R. DeJarnette Chair of Advisory Committee

3 DEDICATION To my family. ii

4 BIOGRAPHY The author was born on February 26th, 1976, in Moscow, modern Russia. His middle school years were largely spent at the school number 179, and he finished high school number 1182 (current number 1550). After high school he got accepted to Moscow Institute of Automobiles and Roads, in the curriculum of wheel and caterpillar - based vehicles due to interest in mechanical engineering. There he finished first year of study before coming to USA with his family at age 17 in the Summer of He lived in Wichita, KS, in Greensboro, Durham and Raleigh in NC, and presently living in Newport News, VA. Since coming to the US he attended Wichita Area Vocational Technical School, where learned auto-repair and English language and various aspects of American culture, attended Wichita State University, North Carolina A&T State University in Greensboro (received BS in mechanical engineering), North Carolina State University in Raleigh NC (received MS in mechanical engineering). While at NC State he began working on doctoral degree in aerospace engineering, but began his job at NASA Langley. He continued to work toward his degree while working at NASA, and completed this Dissertation on the effects reaction control thrusters on aerodynamics and areroheating of blunt entry capsules. iii

5 ACKNOWLEDGEMENTS I would like to thank my grandfather, Arkadiy E. Mikirov, for having been a defining influence for me, encouraging my early curiosity, interest in science and engineering. I think I owe him my intuition and desire to seek out and solve challenging problems. He left us early, I wish I got to know him more. I would like to thank my advisor for his mentorship and help, and for his patience with a working student. I would like to thank my committee for having shaped my understanding of the field of hypersonics, and for helping me explore my own ability. I d like to thank my family for their encouragement, understanding and support. I would like to thank my colleagues at NASA Langley Research Center for an endless supply of challenging problems, for their help and for useful discussions that were instrumental in the present work. iv

6 TABLE OF CONTENTS List of Tables vi List of Figures vii Chapter 1 Introduction Introduction Literature Data from past missions Apollo Program Viking Program Winged Vehicles, Space Shuttle Orbiter Chapter 2 Description of the Problem Summary Shocklayer Shoulder Expansion Mixing layer and the recirculating zone Thruster nozzle internal flow Thruster plume Plume and Nozzle Scaling Chapter 3 Analysis CFD Modeling Analysis of Recent Flight Vehicles Mars Phoenix Capsule Aero-RCS Analysis Mars Science Laboratory RCS-Aerodynamics and RCS-Aeroheating Analysis Analysis of Orion CEV RCS aeroheating Flight CFD Model Summary and Conclusions References v

7 LIST OF TABLES Table 2.1 MSL hypersonic conditions Table 2.2 Product moles as function of NH3 dissociation Table 2.3 Combustion product as function of NH3 dissociation, assuming 0.25% water by mass Table 2.4 Values of γ at different stations within the nozzle for a frozen mixture of 87.11% N2, 12.54% H2 and.3% NH3 by mass vi

8 LIST OF FIGURES Figure 1.1 Computed(LAURA) MSL capsule flowfield at Mach 18 hypersonic condition, streamlines, contours of pressure Figure 1.2 Layout of Apollo CM RCS, borrowed from [2] Figure 1.3 View of Apollo 8 capsule showing effects of aeroheating due to RCS, source - JSC photoarchives Figure 2.1 Variance in shock stand-off and shape at Mars, U=2km/sec, density=3.483e- 3 kg/m3 (LAURA) Figure 2.2 MSL trajectory profile Figure 2.3 Properties above the wake shear layer for hypersonic conditions Figure 2.4 Flowfield at Mach Figure 2.5 Flowfield at Mach Figure 2.6 Flowfield at Mach Figure 2.7 Flowfield at Mach Figure 2.8 Flowfield at Mach Figure 2.9 Flowfield at Mach Figure 2.10 Re at Mach Figure 2.11 Re at Mach Figure 2.12 Re at Mach Figure 2.13 Re at Mach Figure 2.14 Re at Mach Figure 2.15 Re at Mach Figure 2.16 Predicted flowfield for Apollo free-flying model, Mach 10 Helium, α=0. 19 Figure 2.17 Effect of grid size on predicted aftbody pressure, Mach 10 Helium, α=0 20 Figure 2.18 Comparison of predictions with measured pressures, Mach 10 Helium.. 21 Figure 2.19 Variation of base pressure coefficient with Mach number for blunt base capsule at Mars Figure 2.20 Variation of free-stream, dynamic and base pressure (base pressure constructed using curve in previous figure Figure 2.21 Representative nozzle flow, Mach countours (LAURA) Figure 2.22 Effect of scarf on predicted performance Figure 2.23 Nozzle shape used for calculations at 65 degree scarf angle (showing complimentary 25 degree angle labeled) Figure 2.24 Fine (.84M points) grid used for nozzle calculations. Unscarfed grid is shown Figure 2.25 Representative pressure along the nozzle wall Figure 2.26 Regular (top) and Mach (bottom) reflections within the under-expanded jet Figure 2.27 Plume at Mach Figure 2.28 Plume at Mach Figure 2.29 Plume at Mach vii

9 Figure 2.30 Plume at Mach Figure 2.31 Plume at Mach Figure 2.32 Partial schematic of dual thruster flow field Figure 2.33 Flight Enthalpies, red line - jet enthalpy, black - free-stream Figure 2.34 Predicted effect of enthalpy ratio on heating due to roll thruster Figure 2.35 Predicted effect of enthalpy ratio on heating due to yaw thruster Figure 2.36 Predicted effect of enthalpy ratio on heating due to yaw thruster, local effects Figure 3.1 Phoenix capsule geometry Figure 3.2 Illustration of the flowfield around Phoenix capsule Figure 3.3 Phoenix capsule features, image borrowed from project material Figure 3.4 Phoenix RCS layout, image borrowed from project material Figure 3.5 Pitch firing configuration Figure 3.6 Yaw firing configuration Figure 3.7 Moments about X-axis Figure 3.8 Moments about Y-axis Figure 3.9 Investigated conditions Figure 3.10 Variation of dynamic pressure and basecover pressure Figure 3.11 Iteration history of aftshell moment Figure 3.12 Iteration history of control gain Figure 3.13 Mach 27.2 and 6 deg sideslip without the thruster Figure 3.14 Mach 27.2 and 6 deg sideslip with the thruster Figure 3.15 Mach 27.2 and 6 deg sideslip without the thruster Figure 3.16 Mach 27.2 and 10 deg sideslip with the thruster Figure 3.17 Yaw interaction at Mach 27.2 for 6, 10 deg sideslip Figure 3.18 Pitch control and aerodynamics at Mach Figure 3.19 Yaw control and aerodynamics at Mach Figure 3.20 Pitch interference at Mach Figure 3.21 Yaw interference at Mach Figure 3.22 Flow around MSL capsule at Mach Figure 3.23 Jet-wake interaction Figure 3.24 Layout of OML 6 RCS model showing roll and pitch-yaw jets Figure 3.25 Detail of the grid near nozzle (every other gridpoint shown) Figure 3.26 Mach 2.5 comparison of LAURA predictions with data Figure 3.27 Mach 3.5 moment coefficient with and without jets over a range of angles of attack Figure 3.28 Mach 3.5 flowfield with the sting Figure 3.29 Mach 3.5 flowfield without the sting Figure 3.30 Mach 3.5 flowfield pressure with the sting Figure 3.31 Mach 3.5 flowfield pressure without the sting Figure 3.32 Mach 3.5 comparison of moments, effects of support sting Figure 3.33 Slice of flowfield showing structure at range of nozzle pressures Figure 3.34 Surface pressures from solutions with different nozzle pressures Figure 3.35 Effect of nozzle pressure on interaction moment viii

10 Figure 3.36 Viking RCS[35] Figure 3.37 Second iteration of RCS Figure 3.38 Third iteration of RCS Figure 3.39 Flow environment Figure 3.40 RCS cover heating, qualitative Figure 3.41 Thruster effluent mixing with the capsule s wake Figure 3.42 Final thruster arrangement of MSL RCS Figure 3.43 MSL aftbody pressure, yaw jets, candidate RCS layout, computed for Mach Figure 3.44 Aftshell surface yaw-moment arm distribution Figure 3.45 MSL aftbody pressure, yaw jets, computed for Mach Figure 3.46 MSL aftbody pressure, yaw jets, computed for Mach Figure 3.47 Nozzles in grid, configuration Figure 3.48 Nozzles in grid, final configuration Figure 3.49 Mach 18 high incidence baseline solution Figure 3.50 Mach 18 high incidence yaw solution Figure 3.51 Mach 18 LAURA-DPLR comparisons with and without RCS, from [33]. 77 Figure 3.52 X-moment arm lengths for each point on the MSL backshell w.r.t. the CG 78 Figure 3.53 X-moment arm lengths for each point on the MSL backshell w.r.t. the CG Figure 3.54 Thrust direction options Figure 3.55 Predicted surface pressures before redesign Figure 3.56 Predicted surface pressures after redesign Figure 3.57 Layout of Orion RCS Figure 3.58 Schematic of Orion hypersonic flowfield Figure 3.59 Representative laminar aftbody heating distribution, Run 18 condition. 86 Figure 3.60 Windside aftbody comparison, run 18, from [40] Figure 3.61 Single yaw jet, TSP Figure 3.62 Single yaw jet, LAURA Figure 3.63 Dual yaw jet, TSP Figure 3.64 Dual yaw jet, LAURA Figure 3.65 Entry trajectory profiles Figure km/s heatrate (roll RCS) Figure km/s heatrate (roll RCS) Figure km/s heatrate (roll RCS) Figure km/s heatrate (roll RCS) Figure km/s heatrate (single roll RCS) ix

11 Chapter 1 Introduction 1.1 Introduction As a capsule enters an atmosphere, it interacts with the surrounding gas. This interaction produces aerodynamic forces and moments that act on the capsule during entry, and in the process, reduce its kinetic energy to an acceptable level for the deployment of a decelerator or to start the powered descent. Interactions between the vehicle and the surrounding flow, which are of importance to this paper, occur during hypersonic and supersonic flight. In these regimes, flow around the capsule is characterized by the presence of a bow shock ahead of the capsule, multiple expansion waves around the forebody shoulder, a massively separated wake flow field, and a complex recompression shock system behind the vehicle as shown in Figure 1.1. Depending on the capsule shape and size and the free stream conditions, the flow around it may be laminar, transitional, or turbulent. Because of a large amount of energy that must be dissipated during entry, capsules are shaped to produce large amounts of drag for their volume. Typically a small amount of lift is required for aeroassist maneuvers, therefore low lift-to-drag ratios are used. While drag, experienced by the vehicle during entry reduces the vehicles total energy, lift can be used to alter its course through the atmosphere. Lift can be obtained by flying an axisymmetrically-shaped capsule at an angle-of-attack, either by CG offset or by using a hypersonic trim tab. Use of lift for guidance during entry can help reduce peak deceleration loads and aeroheating as compared to a ballistic entry, all other things being equal. Reduction of the landing ellipse and a greater altitude at parachute deployment are typically sited as additional benefits of guided lifting entry. Viking landers flew lifting unguided entry with the lift vector pointed straight up to maximize the altitude at parachute deployment for safety, because of concerns over very high uncertainty in Mars atmosphere. Large fraction of current and future entry vehicles are designed to use lift for guidance. Mars 1

12 Figure 1.1: Computed(LAURA) MSL capsule flowfield at Mach 18 hypersonic condition, streamlines, contours of pressure Science Laboratory entry vehicle has hypersonic L/D of.24 and will perform bank reversals during entry to mitigate heating, minimize landing dispersion and to achieve required altitude at parachute deployment [1]. Orion entry vehicle will carry a human crew, so the use of guided lifting entry is especially critical to reduce deceleration loads to acceptable levels. Mars Phoenix lander, conceived as a low L/D vehicle, has abandoned lifting entry partly because of the limited resources, in lieu of a three-axis stabilized controlled ballistic entry. Phoenix will use its control system to dampen out attitude rates, but not for guidance during entry. Concept vehicles, including future Mars architectures to deliver large payloads (>20 metric tons) to surface, all employ various guided entry technologies. Reaction Control System (RCS) is an enabling technology for entry vehicle guidance and control. RCS is composed of rocket nozzles typically installed on the back wall of a capsule, some form of fuel storage and supply and valves that are operated through some electronic interface by the flight control program. In the course of atmospheric entry these rocket nozzles are fired to induce torques on the capsule (Figure 1.2) such as to control attitude and attitude rates and to impart moments, necessary for bank angle modulation commanded by guidance. 2

13 Guidance maneuvers may be commanded at any time during entry. Figure 1.2: Layout of Apollo CM RCS, borrowed from [2] Properties of the local flow in the proximity of the nozzle depend on the free-stream parameters, atmospheric composition, capsule size, shape and attitude and the location of the nozzle. Local flow can be attached and supersonic, or it can be separated. Generally, attached flow is more energetic, and interactions between it and the nozzle effluent can produce shock structures, sometimes referred to as horse-shoe shocks. Such structures develop a quazi-nozzlelike flow directed toward the surface, essentially creating a high energy stagnation flow at the surface of the capsule upstream of the rocket nozzle exit. This type of an interaction can result in a significant increase over baseline surface heating, pressure and shear. Irrespectively of the character of the local flow, any interaction between the rocket nozzle effluent and the local flow will result in changes to the wake flow. This is due to the fact that much of wake is subsonic, and any changes in a given location affect any other location that is within the elliptic boundary. The result of this kind of dependence is that changes in surface environments can occur over most of the rear wall of the capsule when an RCS nozzle is fired. Most of the environmental changes, that occur within the separated part of the wake are low in magnitude, however any interaction with energetic flow outside of the wake shear layer, like the kind that will happen if the jet from the rocket nozzle punches through the separated zone and into the more energetic flow, can result in significant changes in surface environments. In addition to the changes in aeroheating environment which influences TPS response, a change in surface pressure distribution will produce moments on the capsule, which can interfere with the native authority of RCS. Possible effects of RCS jets on gasdynamic environments should be understood for sound design of capsule s TPS and control system. Ground testing plays a key part 3

14 in these analyses, however it has several important limitations. It is impossible to test a live size capsule at conditions, representing flight. Because of this a scaled model must be tested at the conditions, approximating flight. Flowfield past the model, installed in a wind tunnel is not similar, strictly speaking, to that around the flight vehicle at hypersonic speed for a number of reasons. These reasons include the support interference, lack of real gas effects in most facilities and a frequent lack of availability of relevant gas compositions (for example, Mars) specific to the given mission. Because most hypersonic facilities operate at temperatures, where real gas effects are not present, the tunnel simulation has an additional handicap. These limitations aside, scaling of nozzles presents another challenge. When designing a test to look for effects of RCS interference the nozzles must be scaled appropriately to ensure relevance of the jet - flow interference phenomena. Current thinking on the subject of scaling is to duplicate flight jet exit pressure ratio and momentum ratio expected in flight. These parameters have been successfully used to perform Shuttle and MSL analyses. Their adoption is based on empirical evidence of applicability, and not on any rigorous analyses. Even if all of these limitations and challenges can be overcome there remains a question of measurement technique. Heatfluxes on the back cover of a blunt body model in the test are frequently so small that the measurement uncertainty and variability of the signal (noise) may outweigh the signal itself. Similar difficulty exists in the force-moment testing for aero-rcs interference effects: a balance, sized to handle forces and moments, developed on the forebody is typically too insensitive to measure small moments on the back cover due to unbalanced pressure forces. Measurement of forces and moments can not, strictly speaking, be viewed as a validation of an RCS-aero interference numerical model. The reason is simple: an infinite number of pressure distributions may achieve the same total torque. From this point of view the agreement between computed and measured moments is not all that is required. It is beneficial to have test data on a model, whose back cover is instrumented with multiple pressure ports, so that a direct comparison of computed pressures and pressures measured in the test can be made. Current approach to analysis of RCS-induced aeroheating augmentation and control interference is to use state of the art numerical techniques for flight predictions at supersonic and hypersonic speeds, and to use ground test facilities to validate these numerical methods at flight-like conditions, achievable in those facilities. Ground testing is typically limited to perfect gas air supersonic and hypersonic facilities, in which a cold gas (N2) is used as working fluid to produce jets. To capture aeroheating augmentation temperature sensitive paint (TSP) has been used so far, some researchers are developing plans to use thin film gauges along with TSP. RCS aerodynamic interference has been measured using a force-moment balance with mixed success, and plans are being developed to test a model with a large number of aftbody pressure ports. In this dissertation analysis of experimental, computational and engineering data sources with respect to RCS is presented. The author s work is summarized to formulate several 4

15 conclusions: 1. Effect of attitude control thruster firings on aftbody forces and aftbody aeroheating can be significant for blunt capsules; the aftbody is an aerodynamic surface and the forces and moments, generated on it can compete with RCS authority 2. In order to minimize RCS interaction with aerodynamics and aeroheating it is best to use smallest thrusters with largest moment arm; thrusters should be pointed as close to external flow (free-stream) as feasible; this design approach allows to minimize strong interactions 3. CFD is generally uncertain in predicting RCS interactions; requiring that experimental techniques be used whenever possible; 4. RCS design cycle must consider aerodynamics 1.2 Literature This section contains an overview of available relevant literature on the subject of supersonic jets, their interaction with exterior flow, and the existing ideas on the simulation of these processes, both analytically, and with ground test facilities. The analysis of the interaction of an under-expanded jet and exterior flow underwent significant development over much of the last century to present. Experimental facilities can now do unsteady heat flux and pressure measurements, high speed flow visualization, hot wire techniques to determine unsteady content of the flowfield. Analysis tools went from something as simple as fitting a constant radius arc as an approximate plume boundary, to the method of characteristics, to the CFD techniques in use today. Spaid and Cassel [3] report on aerodynamic interference due to RCS jets. This report is particularly useful for high performance vehicles. The report is a compilation of experiments with sonic and supersonic jets in mainly a supersonic cross-flow. Some attempts are made to develop analysis methods for a general jet in a cross-flow problem by constructing correlations between various parameters of the flowfield. These parameters include combinations of jet and ambient flow pressures, jet thrust, ambient flow dynamic pressure, jet exit Mach number, cross-flow Mach number among others. The report is published in 1974 before the widespread acceptance of discretized numerical methods, and the authors make an effort to realistically assess validity of the developed correlational models to predict other data. Authors underscore the complexity of the flowfield and the inadequacy of the simplified models for general analysis. They, however, point out that analyses should be reasonably valid for geometrically similar 5

16 flowfields. Authors treat what is a relatively generic problem of interaction of a jet with crossflow. They make no attempt to investigate the interaction of jets in flows of mixed character (separated flows) where a jet may penetrate a shear layer. This type of a situation occurs in the case of blunt capsules, and is not typical for high performance vehicles that appear to be the focus of the report. In general most jet-crossflow environments result in flow separation some distance upstream of the jet, however this is not the same as the jet exhausting into the separated flow and intersecting a supersonic shear layer. A report on simulation of jets in ground test facilities by Pindzola [4] treats the problem of ground facility simulation of under-expanded jets in either quiescent medium or in supersonic flow aligned with the jet. This situation arises when a nozzle flow exits from the base of a rocket. Pindzola starts by describing in detail the flowfield, associated with the under-expanded plume in quiescent and moving streams. The assumption is made that the internal flow is well characterized, and directs effort to the plumes. Detailed discussion of pertinent physics and experimental observations allows to make determination of applicable analysis techniques. Initial inclination of the jet boundary and the shape of the jet boundary are discussed at length, and experimental correlation are shown. Through the paper an assertion is made that plume boundary shape is of paramount importance to jet simulations. To simplify simulation and analysis the initial turn is treated as a two-dimensional expansion, governed by Prandtl-Meyer equations. As the distance from nozzle lip increases the 3D flow aspects become important, and author references the method of characteristics and other inviscid techniques to define the shape of plume boundary. Further from nozzle, the transition to turbulence within the plume boundary is possible. Transition to turbulence makes inviscid techniques less relevant, and instead experimental data is required. For some jet pressure ratios a Mach reflection on the axis, frequently referred to as the termination shock, will develop, and its positioning and strength is of significance to ground simulation, which requires a match of static pressure ratio and the Kawamura parameter. Systems with multiple disk shocks are discussed in the context of the length of the periodic structure but are of little relevance to the present topic of plumewake interaction for blunt capsules. It should be pointed out that most of the report appears focused on the type of the flow that exists in case of a rocket motor installed in a base of a slender body that moves axially at some velocity, spanning subsonic to hypersonic regimes. This flow has different scales then the present problem. For example, the size of the plume is comparable if not greater then the size of the undisturbed wake, whereas this isn t usually the case for blunt capsules and their RCS. 6

17 1.3 Data from past missions There is a significant generation gap in NASA s use of blunt capsules for space exploration. Capsules of the 60 s and 70 s (Mercury/Gemini, Apollo and Viking) have flown successful controlled unguided and guided entries into Earth and Mars atmospheres. In the early 80 s the Space Shuttle took over the task of flying people and cargo into Earth s orbit and it was viewed as more versatile and capable, thus ending the use by NASA of capsules for manned flights near Earth. Crew Exploration Vehicle (CEV) is slated to be a replacement for the Space Shuttle orbiter and will return a capsule - based architecture to human flights near Earth and to the Moon. Since the two successful landings of Viking 1 and 2 in 1976, landings on Mars have seized for twenty years. A small and a comparatively simple mission, Mars Pathfinder, landed on Mars in 1997 and marked the renewed interest in Mars surface exploration. MPF was followed by larger and more complex MER Spirit and Opportunity in 2004 which are operational today. Capsules of these relatively recent missions were spin-stabilized and did not use controlled entry. Two other recent missions to Mars (Mars Polar Lander and Mars Phoenix) have used RCS for attitude rate control and azimuth alignment. It is unclear what happened to Mars Polar Lander which was due to land in December of 1999, but its sister ship Phoenix underwent significant system-wide changes, as compared to MPL, and has been launched in 2007 for a successful landing in For the purposes of this report Phenix is viewed as a current mission and not a prior mission. Analysis of Phoenix RCS is presented in this dissertation along with respective analyses for other current missions. Because none of the recently flown 3-axis stabilized spacecraft (MPL and Phoenix) had any instrumentation to help measure RCS effects in flight, and all other recent flights were not controlled, any inquiry into heritage data on RCS aeroheating and control interference for blunt capsules has to be addressed to prior missions, such as Apollo and Viking, which used RCS during entry and do have some data Apollo Program Apollo program carried out ground testing of aeroheating augmentation due to RCS jets using the then-new phase change coating technique. Results of that work are summarized by Jones and Hunt [5]. Authors found that interference heating on the Apollo shape was significant for yaw and roll jets. Augmentation factor of 4 due to yaw jet, factor of 11 due to forward-firing roll jets and a factor of 3 due to aft-firing roll jets. No appreciable augmentation was found due to pitch jets, which are located on the lee-side of the aftshell, and interact with separated subsonic flow. Interference heating in the case of the yaw and roll jets covered significant acreage of the backshell. In particular, forward-firing roll jets produced the most energetic interaction with the shear layer, coming off of the capsule s shoulder, and this yielded the greatest heating 7

18 augmentation over the largest area. Figure 1.3: View of Apollo 8 capsule showing effects of aeroheating due to RCS, source - JSC photoarchives Apollo entry capsules were instrumented to measure surface heating. Heating rate spikes on the lee-side of the spacecraft during entry were found to correspond to RCS jet firings and amounted to about a factor of 5 times the nominal measurement [6]. While nominally there was no tests to look at the effect of RCS jets on aerodynamics of Apollo capsules, the flight of Apollo 7 saw considerable pitch and yaw control activity in the transonic region during the final 2 min before drogue deployment, which was attributed to winds and thruster-flow interference [2] Viking Program Viking program has made an attempt to measure experimentally the magnitude of jet-aerodynamic interference. A test was conducted in Mach 20 wind tunnel, where thruster plumes were simulated as solid bodies. This test did not net any significant insight into the jet-wake interference because of the insufficient accuracy of the low AOA data [7]. It was suggested, that the test be 8

19 repeated with a balance, designed to measure smaller moments, but this was never carried out. No attempts were made by Viking Project to measure aeroheating augmentation due to jets because aftbody heating was not expected to be significant [8]. Viking was entered into Mars atmosphere from circular orbit at a relative velocity of about 4.6 km/sec [9]. Low speed entry of a capsule with a relatively low ballistic coefficient (m/c D A=63.7) resulted in very low heat fluxes on the aft-cover. Because these heatfluxes were low, on the order of 1 Watt/cm 2, it was possible to make the aft-cover of aluminum, and not cover it with thermal protection material. Use of the small 8lbf thrusters for rate damping and for lift vector alignment would not produce the heat fluxes and heat loads much beyond the baseline. Therefore, it was not essential for Viking to test RCS aeroheating augmentation Winged Vehicles, Space Shuttle Orbiter Space Shuttle Orbiter is a lifting body winged vehicle and does not share aerodynamic characteristics with blunt entry capsules. Aerothermodynamics and aerodynamics programs have conducted experimental and computational investigations to investigate some effects of nozzles, but only aerodynamic effects of blowing nozzles were investigated, see for example, Scallion [10]. 9

20 Chapter 2 Description of the Problem 2.1 Summary This section describes in detail the relevant phenomena and outlines the basic analysis tools and the state of knowledge with respect to these phenomena. By very nature of RCS-gasdynamic interaction, the process of interest occurs in the aft portion of the capsule flowfield. Because of this, the fidelity of the analysis depends on the fidelity of capture of individual processes, which occurr upstream of the interaction. While analysis of uncertainty is not in the scope of this dissertation (largely due to insufficient existing data) it is important to recognize the role of uncertainty in capture of various areas of the flowfield to the problem at hand. Because of the diversity of flow environment this section is separated by type of flow process: 1. Shocklayer 2. Shoulder expansion 3. Separation 4. Recirculated air zone 5. Thruster nozzle-internal flow 6. Thruster plume 7. Shear layer Shocklayer Shocklayer on the forebody of the capsule is composed of a chemically reacting and sometimes non-equilibrium gas mixture. Most of the shocklayer is subsonic, with parts of lee-side forebody 10

21 flow becoming low-supersonic. The current state of the art numerical tools LAURA [11] and DPLR [14] have been shown to accurately capture the bow shock shape and distance from a blunt body at hypervelocity in perfect gas (see, for example Horvath[15]), pressure and heating on the surface for both laminar and turbulent boundary layers (Hollis [16]). While both codes are capable of treating non-equilibrium dissociated mixtures, this capability is not entirely validated. Specific questions exist about modeling of thermal energy modes of polyatomic molecules, ex.: CO2. The author of this dissertation found an unexpected sensitivity to the two temperature model currently in use in the CO2 environment at Mars. Because of the low atmospheric density there, the shocklayer is in non-equilibrium through much of entry, and the shock stand-off can vary dramatically as a function of the model that s used, as illustrated by figure 2.1. This figure is generated by use of CAMAC and Millican-White CO 2 vibrational relaxation models. Stand-off and shape has a first order effect on shocklayer pressure, and can impact accuracy of analysis of any area downstream. This issue is of main concern in the low-density atmospheres made up of polyatomic gases (i.e. Mars), and is generally not a driver for flights into Earth atmosphere, where the pressures are sufficiently high to keep the shocklayer gas in thermal equilibrium. It is, therefore, reasonable to rely on LAURA and DPLR to solve this segment of the flowfield and to expect accurate results, keeping in mind possible uncertainty at Mars Shoulder Expansion The expansion fan at the blunt capsule s shoulder is formed as flow turns to align with the windside aftshell on the windside and with the lee-side mixing layer on the leeside. Generally the expanded flow reaches Mach number around 4 for much of hypersonic flight. Fidelity of computations through the expansion is difficult to quantify. The more practical parameter to consider is the heat flux, predicted on the windside aftbody surface. Comparisons between computation and wind tunnel measurements, carried out for that area by Hollis [16] show that the heat flux predictions in both laminar and turbulent flowfields are simulated accurately on the Orion shape with the use of aforementioned numerical tools. This provides some level of confidence that the parameters of the flow immediately after the expansion are correct Mixing layer and the recirculating zone The mixing layer divides the supersonic high energy flow from the recirculating zone of low energy air near capsule s leeside aftbody. The mixing layer determines the balance of mass, momentum and energy transport between these two highly dissimilar regions. Accurate prediction of the transport across the mixing layer and its geometry is vital to pressure and enthalpy of the air inside the recirculation zone, as well as to the accurate solution of expansion in the 11

22 Figure 2.1: Variance in shock stand-off and shape at Mars, U=2km/sec, density=3.483e-3 kg/m3 (LAURA) lee-side external flow, as it is influenced by the wake angle. Nominal operation of tools, such as LAURA and DPLR results in inadequate grid density in the wake and incorrect grid alignment with respect to the mixing layer. Significant additional grid density is required, together with some suitable grid alignment procedure, to provide the solution domain for the mixing layer (see, for example, Hollis [17]). Discussion of the simulation of transport phenomena across the mixing layer must include a discussion of turbulence. The mechanism, postulated by Chapman [18] required that the mass of recirculating air that is evacuated by the mixing layer is replaced through the layer itself near the trailing end of the recirculating zone. If all or part of the layer is turbulent - this balance will be affected, resulting in a different shape of the mixing layer and a new pressure of the recirculating air. Because of the extent to which turbulent transition in the mixing layer can affect the outcome of the wake analysis, it should be determined if turbulence in the layer is likely. It has been shown (for example, see von Doenhoff [19]) that low speed free shear layers are more susceptible to turbulent transition then attached boundary layers. In other words, for 12

23 the same Reynolds number, free shear layers transition earlier than attached boundary layers. However, at high Mach numbers, mixing layers have been shown to exhibit remarkable stability [18]. For example, the laminar limit of about Re=1.E6 was shown for Mach 4 flows for a wide range of flowfield geometries. This is particularly significant in the context of the present discussion involving a hypersonic free shear layer. As will be discussed later, in some cases of hypersonic capsules there appears to be a good reason to expect a laminar shear layer, which might transition only toward the end of the recirculating region. The mixing layer establishes the interface between the external flow and the recirculating zone. The recirculating zone contains a mass of air, whose total enthalpy is between percent of that in free-stream. The enthalpy and momentum of the recirculating air mass is provided from the exterior flow through diffusion across the mixing layer, and is, therefore, directly affected by whether the layer is laminar or turbulent. The state of the mixing layer is therefore a critical player in the pressure and enthalpy of the recirculating air. The above assessment of percent is based on the author s experience through numerous CFD calculations using a laminar model over a broad range of Mach numbers. The state of the flow inside the recirculating region is not necessarily coupled to the state of the mixing layer. Inside the recirculating zone, the predicted local Reynolds number is generally low due to low pressure there, and the flow scale length there is of the order of a meter. It appears unlikely that such a flow will develop and sustain turbulence, instead it should remain laminar, even if unsteady and vortical. Existing leeside aftbody heating data for CEV [16] suggests that the laminar models adequately capture heat rates in the separated area even when the attached flow elsewhere on the capsule is turbulent. Convective heating predicted by both laminar and turbulent models is also shown to be very close in this low density separated area. This is, however, not an indication of completeness of these models. MSL Entry Trajectory, Free Stream Profile and Flow Predictions Figure 2.2 shows variation of conditions along MSL design trajectory. Data in this plot is generated by POST[20], and is used here as a source of conditions for relevant analysis. The plot shows the free-stream Mach number, Reynolds number and the free-stream dynamic pressure along the trajectory. This plot is going to be a backdrop to following discussion of MSL wake environment. Table 2.1 shows free-stream conditions selected for analysis from figure 2.2. Preceding discussion of likelihood of turbulence is relevant here, and examining figures 2.4 through 2.9 and 2.10 through 2.15 shows why. The first set demonstrates the scale of the recirculating region predicted using laminar model at Mach 26, 22, 18, 14, 10 and 6 along the trajectory shown in the figure 2.2. The latter set of figures shows the predicted Reynolds number computed on a per-meter scale. The figures are extracted from LAURA solutions performed for this analysis using 8-specie two temperature gas model. Interrogation of these plots will 13

24 show that the lee-side shear layer Mach number at these hypersonic conditions falls between 3 and 6. Reynolds number above the shear layer does not exceed 45000, as shown in the figure 2.3. Error bars on Mach and Reynolds numbers indicate variance within the relevant portion of the flowfield, and are plotted simply to give a better indication of the flow parameters. For a recirculating region of the order of capsule diameter (here 4.5 meters), this mixing layer is likely to stay laminar for much of hypersonic entry, making laminar analysis appropriate. This doesn t necessarily mean that the flow should be expected to be steady, and at some of these conditions (particularly near maximum dynamic pressure) unsteadfy flow is predicted by the model, requiring time-consistent integration. Figure 2.2: MSL trajectory profile Table 2.1: MSL hypersonic conditions M ρ, kg/m 3 V, m/sec T, K E E E E E E

25 Figure 2.3: Properties above the wake shear layer for hypersonic conditions 15

26 Figure 2.4: Flowfield at Mach 6 Figure 2.5: Flowfield at Mach 10 Figure 2.6: Flowfield at Mach 14 Figure 2.7: Flowfield at Mach 18 16

27 Figure 2.8: Flowfield at Mach 22 Figure 2.9: Flowfield at Mach 26 Figure 2.10: Re at Mach 6 Figure 2.11: Re at Mach 10 Figure 2.12: Re at Mach 14 Figure 2.13: Re at Mach 18 17

28 Figure 2.14: Re at Mach 22 Figure 2.15: Re at Mach 26 18

29 CFD comparison with Free-flying Model Pressure Data Accurate prediction of pressure in the recirculating zone of a blunt capsule at hypervelocity remains a challenge. Validation of CFD codes for this task is held back for a number of reasons: It is difficult to provide good quality wind tunnel measurement in the wake of a capsule model, in part due to low pressure, that s being measured, and in part due to presence of the support and uncertain interaction between the support and the wake. Presence of the support complicates the CFD model by introducing shear layer impingement mechanics, that would not be as severe for a flight vehicle. Kemp [21] conducted a number of experiments on free-flying tunnel models with precise intent to avoid support interference while measuring aftbody pressure. Models were tested in NASA Ames hypersonic helium tunnels and used FM signal telemetry to communicate with their data acquisition system. Models were launched by a pneumatic launcher into the test section, such that the model velocity relative to the facility was a negligible fraction of the total free-stream velocity. Pressure was measured half-way down the aftbody cone, and it was done on the wind-side of the aftbody in those cases where a high angle of attack test was being conducted. Data in the report is sufficient to reconstruct all relevant conditions and scales and to set up a CFD simulation to approximate the experiment. Figure 2.16: Predicted flowfield for Apollo free-flying model, Mach 10 Helium, α=0 LAURA calculations were set up in an attempt to match experimental predictions of Kemp. Laminar LAURA simulations with no symmetry plane and with a time-consistent integration were used. Grids had sizes from 3.1M elements to 9.5 M elements. Perfect gas Helium model used molecular weight 4.003, Prandtl number and specific heats ratio LAURA Sutherland s viscocity law with coefficients v1gas= e-6 and v2gas=80.0 was used. Com- 19

30 puted flowfield at zero angle of attack is shown in the figure No symmetry is assumed in the solution, and at these conditions wake computation is steady and non-oscillating. Figure 2.17 shows the variation of base pressure predictions as a function of the volume grid size for a a model at Mach 10 in helium, again assuming zero angle of attack. The figure indicates that as the grid is refined to the upper limit, explored here, the aftbody pressure prediction is continuing to change. There is a practical limit to grid refinement, and while it is not reached in these perfect gas calculations, the present grids are past that point should a real gas non-equilibrium entry environment be analyzed on them. The purpose of the last comment is to illustrate that the CFD grid-independence is not always practical, nor it is reachable in some circumstances. Example of such a circumstance would be a wake simulation, in which successive refinement picks up finer flow scales, thus slightly changing the flow. Comparisons of predictions with Figure 2.17: Effect of grid size on predicted aftbody pressure, Mach 10 Helium, α=0 measured data at Mach 10 are shown in the figure CFD overpredicts data for very low angle of attack. At small non-zero angle of attack predicted pressure drops to a level, more consistent with the measurement (this trend is also consistent with, for example, the Mach 15 20

31 Figure 2.18: Comparison of predictions with measured pressures, Mach 10 Helium dataset from the same source), and at higher angles of attack where the flow becomes attached the CFD and data match very well. One possible explanation of the disagreement at zero degree angle of attack is that the physical model, with its pitching motion may never develop a truly axisymmetric wake when going through zero incidence, unlike the CFD model, where the capsule is held at a static angle to the stream, allowing the wake to settle into an axisymmetric configuration. Wether the disagreement is in fact a local phenomena and is centered at local angles of attack, thus traceable to the lack of symmetry in the flow, can not be answered with the present data, as little of it is collected at low angles of attack. The general outcome of this and other comparisons [17] of CFD against experimental hypersonic wake data is that lack of relevant physics models is in the way of widespread use of CFD for such problems. Credible turbulence modeling and time accuracy are some of the essential requirements. Aftbody Pressure on Blunt Capsules at Mars During most of the atmospheric entry the pressure on the aftbody of the capsule is significantly smaller then the forebody stagnation pressure. At hypersonic speeds aftbody pressure coefficient is essentially independent of the free-stream Mach number. The influence may come in due to vehicle passing through regions of equilbrium and non-equilibrium flow and due to change in ratio of specific heats in shocklayer, which can affect shocklayer geometry, and, consequently, aftbody flow. Because of this, hypersonic regime is characterized by aftbody pressure following the free-stream dynamic pressure. Free-stream pressure is of smaller significance in this context. 21

32 Because of the uncertainty, associated with the CFD prediction of the wake of a blunt capsule, Mitcheltree (see, for example, [22]) reconstructed aftbody pressure from Viking flight data and arrived at the base pressure coefficient for use for Mars Pathfinder. He fit a polynomial of the form C A,base = C p,base = a 0 + a 1 + a 2 M M 2 + a 3 M 3 (2.1) where a 0 = 8.325E 03, a 1 = 1.129E 01, a 2 = 1.801E + 00 and a 3 = 1.289E 00. Figure 2.19 shows the form of the Cp curve in the above equation. The curve indicates two regions of distinct wake behavior. At hypersonic speed, wake pressure coefficient is nearly constant with respect to Mach number. The two mechanisms that could influence Cp in that region (and, thus, possibly conflict with the shape of the curve in the figure 2.19) are shear layer turbulence and shocklayer non-equiolibrium. The former would change transport across the layer, thus influence pressure, the latter would change conditions within the external part of the flowfield, which will in tern drive separation line and angle. At hypersonic conditions the pressure on the base of the capsule at Mars will track dynamic pressure, and will exceed free-stream static pressure during part of the entry. At supersonic conditions, as the dynamic pressure drops, the pressure in the wake will drop below the free-stream static pressure and form what s typically thought as a low pressure wake. As the velocity further decreases the momentum deficit in the wake becomes less noticeable, and wake pressure will tend to free-stream static pressure. Velocities, at which this occurs are not interesting to the present work, because capsules typically deploy a parachute at low supersonic velocities, and control interaction under parachute is a mute point. Figure 2.20 shows the free-stream pressure, dynamic pressure, and aftbody, or base pressure, reconstructed from it and the Cp, shown in the figure This relation is currently applied to all Mars entry capsules to determine base pressure for aerodynamics. Its applicability across a range of entry system designs needs further research, as both the shape and entry profile vary significantly for Mars entry capsules. 22

33 Figure 2.19: at Mars Variation of base pressure coefficient with Mach number for blunt base capsule Thruster nozzle internal flow Typical RCS thruster is a mono-propellant hydrazine thruster where catalyst is used to start and maintain combustion. Because of the complexity of the interaction of thruster flow with the wake, it is important to understand operation of these thrusters. Hydrazine combustion is composed of two main processes: 1. 3N2H4 4NH3 + N2 +Q 2. 4NH3 + Q 2N2+6H2 Breakup of ammonia occurring in the second reaction takes the heat out of the system, reducing thruster performance. The existing test data indicates that about 50% of ammonia is consumed by the time flow exits the thruster. It can also be shown that the composition changes very little after the nozzle throat, and it is common to assume frozen flow through the diverging part of the nozzle. There are a number of ways to obtain the properties at the exit plane of the nozzle of a hydrazine thruster. One of them is to use the CEA (Computer Program for Calculation of Complex Chemical Equilibrium Compositions and Applications) code of Gordon 23

34 Figure 2.20: Variation of free-stream, dynamic and base pressure (base pressure constructed using curve in previous figure and McBride [23] [24]. The included rocket problem allows a rough assessment of exit state and performance parameters. CEA is not able to account for non-equilibrium aspect of the hydrazine combustion, however it is possible to freeze the composition within some bounds. Generally, because freezing is only possible starting at throat, the properties at nozzle exit will not capture effect of non-equilibrium accurately. A better approach is to calculate gas properties based on assumed dissociation of ammonia. Because experimental data on composition of effluent exists this shouldn t be very difficult to do. A most rigorous approach is to develop a CFD simulation to include the above species, and reverse-engineer the composition in the chamber based on the relevant measurements of the composition at exit. In other words, several sets of chamber compositions should be used, and the exit of each simulation compared to the desired. Composition changes very little between subsonic chamber and exit plane for this type of thruster. This means that the iteration process can be bypassed in the interest of time. Table 2.2 shows mole fractions of ammonia, nitrogen and hydrogen as a function of completeness of the second reaction. 24

35 Table 2.2: Product moles as function of NH3 dissociation dissociation, % NH 3 N 2 H % water is assumed to be in fuel because N2H4 is a hygroscopic substance. In addition to that, most original development of these thrusters was done with traces of water in fuel, and it is believed that the removal of water might have unintended consequences [25]. Mole fractions of the products can be computed with correction for the presence of water as shown in the table 2.3. Values of γ are based on the ideal gas degrees of freedom for respective substances. Table 2.3: mass Combustion product as function of NH3 dissociation, assuming 0.25% water by dissociation, % NH 3 N 2 H 2 H 2 O M mixture γ mixture,ideal Range of static temperatures in a thruster is from near room temperature at exit to 1500K at the chamber (where static is the total temperature). Because of this, γ can vary significantly 25

36 along the nozzle for the same mixture of gases. The following example demonstrates. Calculations using CEA code were carried out for hydrazine fuel for the exit area ratio 26.2, chamber pressure bar and temperature 1364K. Unfortunately CEA doesn t allow composition at the chamber to be frozen (freezing possible only after the throat) so the resultant composition has very low content of NH3, though the example in the table 2.4 is still illustrative. It should Table 2.4: Values of γ at different stations within the nozzle for a frozen mixture of 87.11% N2, 12.54% H2 and.3% NH3 by mass location T,K p,bar γ chamber throat exit be noted that the computer programs that exist to determine thruster performance have a slightly different objective, from what s needed for the present analysis. Here, the main interest is in the gas properties and composition. The fact that the properties of the effluent will have some uncertainty associated with them, and that this uncertainty isn t directly related to the understanding of the general uncertainty in prediction of performance of the thruster should be understood. 26

37 Performance of the thruster for scaling purposes can be assessed in a number of ways. Figure 2.21 shows a representative nozzle flow, calculated using LAURA code for this analysis. To circumvent lack of characteristic inflow boundary condition in LAURA the author employed a technique of artificially freezing part of the domain to create an upstream boundary condition. This technique may result in around a 10% error in pressure, but the error can be easily quantified by integrating across the inlet, and corrected. Because there are no provisions in LAURA for subsonic inflow, this approach was developed and used in most of simulations in this dissertation. Thrust in vacuum can be found by integrating over the exit plane which is the same as T vac = For many thrusters T vac = inlet exit (ṁu)da + surf1 (ṁu)da + inlet exit P da (2.2) P da + F surf da (2.3) surf0,1 F surf da = 0.25T vac (2.4) which sets the practical upper limit on effects of nozzle scarfing on thrust. Figure 2.21: Representative nozzle flow, Mach countours (LAURA) The term scarfing refers to a nozzle exit, such that the exit plane is not perpendicular to the nozzle axis. Scarfing is done frequently to accommodate installation with non-orthogonal intersection between thruster axis and capsule s outer surface. Scarfing is done by one of three methods. Simplest and least effective is the addition of a cylindrical extension to the bell nozzle, and cutting the cylinder at an angle. Simple as it is, this approach generates internal shocks and performance losses. A better approach is to replace the cylinder with a cone, of half angle equal 27

38 to the wall angle of the bell nozzle near exit. This approach removes internal shocks, but yet a better approach is to extend the nozzle and to cut the bell at an angle. While this produces a more complex intersection with the capsule surface, possibly increasing the difficulty on the mechanical side, the overall best nozzle performance is achieved. For purposes of scaling flight nozzles to tunnel conditions it is sometimes important to know the nozzle s area ratio. While a very simple task for an axisymmetric nozzle, a case of a scarfed nozzle presents a challenge. Several approaches have been used to determine the effective area ratio. One approach is to determine the ratio as an average of the long side and short side area ratios. The long and short side refer to the longest and shortest distances along the nozzle to exit lip. Another approach is to determine area ratio based on the projection of the nozzle s exit plane onto the plane, orthogonal to thruster axis. This can be tedious, and fundamentally doesn t add any real value. Namely, while the estimate of the axial flow area is slightly better then the first approach, the relationship between the effective area ratio, effective exit Mach and real distribution of Mach is still the question. Either approach works, as long as its clearly understood which one is being used. Because RCS thruster flows are under-expanded in most applications, the exit Mach is not the most important parameter in of itself, instead the relation of nozzle geometry, exit and ambient conditions and resultant plume geometry are desired. Practical scarfing angles have moderate effect on thrust magnitude and direction. As figure 2.22 indicates there s a range of scarf angles for which axial thrust degradation is on the order of a couple of percent. The figure represents LAURA simulations set up for the thruster configuration shown in the figure Nozzle scarfing was performed as shown in the figure Several grids were used, the finest had.84m points, and its unscarfed version is shown in the figure There is a practical reason why even the large scarf angles result in moderate thrust penalty. As the figure 2.25 indicates, pressure in the divergent section of the nozzle drops very quickly, so that alterations of the nozzle, made far enough from throat do not change the integral in equation 2.4 significantly, which is reflected in the figure Another way to think of effect of scarfing is that its effect on thrust direction is mainly through the pressure term in the equation 2.2, as the direction of velocity doesn t change, but the pressure integral acquires a side projection. Notably, the scarf angle and the scarf location shown in the figure 2.23 are extreme, and it is common to avoid cutting so close to the throat. 28

39 Figure 2.22: Effect of scarf on predicted performance Figure 2.23: Nozzle shape used for calculations at 65 degree scarf angle (showing complimentary 25 degree angle labeled) Thruster plume Geometry of the thruster plume is cited by Pindzola as one of the most important parameters for simulation in ground test facilities. He stresses the initial turn angle and plume boundary shape, however its easy to see that his emphasis on the plume shape is driven by the specific 29

40 Figure 2.24: Fine (.84M points) grid used for nozzle calculations. Unscarfed grid is shown. Figure 2.25: Representative pressure along the nozzle wall application he appears to be considering. In a general case of a plume interacting with a wake of a blunt capsule the plume shape may not be as critical a parameter, as it is for a rocket. Analysis of plume geometry is not a new field. Manuel Salas [26] developed an inviscid method to calculate plumes in quiescent as well as supersonic exterior flowfield where flow is aligned with 30

41 the plume. The method can capture lip shocks, termination shocks, as well as treat periodic structures when they occur. This method doesn t answer the question of turbulent transition of the plume boundary, and effects that might have on downstream plume development, but its a quick and accurate tool for comparison of plume/wake scale between the flight vehicle and a wind tunnel model. Plumes of RCS thrusters are under-expanded for all trajectory conditions during entry due to comparatively low backpressure that s created in the capsule s wake. Geometry and the physics of an undisturbed thruster plume are explored in this section on the example of MSL thruster in as-built configuration. As the capsule decelerates through the atmosphere, pressure in the wake is changing continuously as was shown in the figure From near vacuum at the entry interface, the base pressure will build up to its maximum at near peak dynamic pressure, which is followed by a decrease to the ambient pressure at low speeds. Physics of the supersonic jet in an inviscid fluid are discussed in great detail in a classical work of Pai [27]. As the flow exits the thruster it undergoes a 2D initial turn, where the turn angle and properties immediately downstream are governed by a Prandtl-Meyer function. As the distance from nozzle lip increases, the 3D effects become important to the shape of the plume and an axisymmetric correction must be introduced to the inviscid technique used. Pai discusses the method of characteristics as a tool to determine some aspects of the jet flowfield, a method popular at the time of the work. Figure 2.26 shows possible plume structures, produced by under-expanded jets. The difference between the upper and lower schematic is the transition from the regular to the Mach reflection. Jet shock can undergo a regular reflection at the centerline, or it can transition to Mach reflection, where the oblique shock is connected to the centerline with a normal shock (also called termination shock, or Mach disk). The area behind this normal shock is a high pressure and high temperature zone, that forms a convergentdivergent nozzle flow further downstream (see, for example, [26]). The jet pressure ratio, gas properties (γ) and nozzle exit wall angle are some of the differentiators between the regular and Mach reflection. Generally, as the jet shock strength increases, Mach reflection is more likely. Figures 2.27 through 2.31 show the relative scale of RCS thruster plumes and the computed wake structure for MSL entry at Mach 22, 18, 14, 10 and 6. As the figures indicate the undisturbed plume dimensions follow aftshell pressure (figure 2.20) inversely, with the smallest plume occurring near maximum trajectory dynamic pressure. Large plume occurs at high altitude condition, where aftbody pressure is low, and at a supersonic condition, where pressure drops again. In the early hypersonic regime the aftbody pressure is low enough to not expect significant RCS-aero interaction. This will be demonstrated on the example of Phoenix capsule in the section on Phoenix analysis. Regardless of the conditions it is clear from the figures 2.27 through 2.31 that a thruster of adequate power produces a plume that will interfere with the supersonic shear layer for most practical orientations of the thruster. 31

42 Figure 2.26: Regular (top) and Mach (bottom) reflections within the under-expanded jet Figure 2.27: Plume at Mach 6 Figure 2.28: Plume at Mach 10 32

43 Figure 2.29: Plume at Mach 14 Figure 2.30: Plume at Mach 18 Figure 2.31: Plume at Mach 22 A particular challenge to analysis is presented by the dual thruster arrangement. It is not uncommon to have a dual-string thruster arrangement, where pairs of thrusters are installed such that the nozzles are parallel, and are placed very close to each other. The plume flowfield that is produced by such a system is characterized by a high pressure turbulent mixing zone, and pair of shocks that traverse the inviscid part of each plume. Figure 2.32 shows the partial schematic of such a flow. Part of the flow to the right of the shown structure is usually turbulent due to natural transition in the plume boundary and due to transition in the compression zone and interaction of the interference shock with plume boundary. 33

44 Figure 2.32: Partial schematic of dual thruster flow field 34

45 2.2 Plume and Nozzle Scaling Testing of RCS in ground facilities requires duplication of a number of relevant physical scales. External flow would typically require to some degree replication of the free-stream Mach, Reynolds numbers, enthalpy, momentum thickness Reynolds number, especially if this is an aeroheating test. Comparable properties of the gas mixture are also desired. In addition to that, the nozzles and nozzle flow should be scaled appropriately. Scaling parameters, presented by Pindzola [4] for this problem result in the attempt to match pressure, momentum and enthalpy ratio. Pressure ratio is defined as the ratio of the jet exit to local ambient 2.5. Here, some ambiguity exists in determining what the local ambient is, and whether it should include the dynamic component of pressure, in the case of strong crossflow. The former concern results in a appropriate uncertainty that s placed on the left and right hand side of the equation 2.5, while the latter concern leads to the equation 2.6. [ Plocalflow ] P jet,exit T est = [ Plocalflow ] P jet,exit F light Spaid and Cassel [3] indicate jet penetration into a supersonic stream related to a momentum ratio-type term. For the purposes of the RCS-gasdynamic interaction this quantity, as defined in the equation 2.6 should use momentum outside the capsule s shear layer. Flight-to-test scaling change the relation between that local momentum and free-stream momentum. Specifically, free-stream enthalpy, dissociation in the shocklayer, surface roughness and effect of wake turbulence on the separation location and wake closure angle all can influence the difference in the momentum ratio. [ ] Qexit A exit Q A ref [ ] HT H w H jet H w T est T est [ ] Qexit A exit = Q A ref [ ] HT H w = H jet H w F light F light (2.5) (2.6) (2.7) Equation 2.7 formulates scaling to enthalpy. Enthalpy ratio between free stream and the jet influences characteristics of RCS-induced aeroheating. This ratio can assume very high values at entry and low values at the end of the deceleration. For example, the enthalpy of the jet of a hydrazine thruster is of the order of 3MJ/kg. Lunar return free-stream enthalpy at entry is of the order 63 MJ/kg. By the time capsule s deceleration is over the free-stream enthalpy is of the order of 0.3 MJ/kg. Clearly, the ratio of jet and free-stream enthalpies can assume a range of values, and enthalpy potentials, referenced to the wall can be further altered due to variations of wall temperature in flight. Figure 2.33 shows the relevant enthalpies graphically. As the capsule slows down free-stream enthalpy departs from being a square of free-stream velocity, and assymptotically approaches the free-stream static enthalpy. Jet enthalpy doesn t change 35

46 during flight. Two types of interactions are likely: the first type is interaction of the jet with Figure 2.33: Flight Enthalpies, red line - jet enthalpy, black - free-stream a nominally separated flow, where mixing of the jet with exterior flow will produce a complex flow structure, and the aeroheating augmentation will be driven by some average of jet and free-stream enthalpy. The second case is when the jet interacts with a nominally attached flow. In this second case the aeroheating horseshoe will be driven solely by the free-stream enthalpy, and near-nozzle heating will be driven by the jet enthalpy. An example of the first type of interaction is shown in the figure 2.34 where roll thrusters of Orion, nominally in separated zone, are simulated in LAURA at simulated wind tunnel conditions. In this case enthalpy ratio is changed through the jet temperature. Evident effect on the horseshoe heating with increase in jet temperature. The second type of an interaction is shown in figures 2.35 and This example is of Orion yaw thrusters, which are nominally in attached supersonic crossflow. The exterior interaction (horseshoe) shows little if any sensitivity to the enthalpy of the jet, but near nozzle phenomena are driven by it, for example the compression heating along the line splitting the nozzles. The negative value of enthalpy ratio in one case is simply the manifestation of indeterminance in the equation 2.7 at near-equal values of jet and wall enthalpies. Generally, because of low free-stream enthalpies in ground facilities it is impossible to make meaningful use of this parameter for blunt capsules. Testing for wake environments is generally conducted at low enthalpies (even high enthalpy reflected shock facilities limit their envelope 36

47 Figure 2.34: Predicted effect of enthalpy ratio on heating due to roll thruster Figure 2.35: Predicted effect of enthalpy ratio on heating due to yaw thruster due to wake establishment considerations), so the driving enthalpy on the separated aftbody is disproportionally low. It is possible to get the ratio itself to be close to flight, but at that point the wall enthalpy and jet enthalpy may be very close in magnitude, making measurement challenging. Relevance of enthalpy matching is, therefore, a questionable proposition for RCS testing of blunt capsules in low enthalpy flows. Given the above relations the parameters of interest can be computed as shown in 2.8 and 2.9. P jet,exit T est = P localflow T est P jetexit F light P localflow F light (2.8) Q exit A exit T est = (Q A ref ) test (Q exit A exit ) F light (Q A ref ) F light (2.9) The exit diameter doesn t strictly have to be identically scaled to the flight vehicle. It can be varied to suit pressure and momentum scales. 37

48 Figure 2.36: Predicted effect of enthalpy ratio on heating due to yaw thruster, local effects Additional challenges to RCS scaling include, for example, the duplication of chemical composition of the flowfield. It is usually not possible to duplicate external flow (take, for example,the CO2 atmosphere), nor the RCS effluent (non-equilibrium flow of combustion products). Matching molecular weights will only partly fill the requirement, as in most reacting environments the composition changes with location in the flowfield. It is possible to compensate for incorrect gas properties by varying jet exit Mach number and nozzle exit cone angle to still match the undisturbed plume boundary considered important in jet interaction testing. 38

49 Chapter 3 Analysis 3.1 CFD Modeling Development of RCS modeling capability in LAURA and RCS modeling capability for blunt capsules in general was driven by the intent to assess aerodynamic-rcs interactions at supersonic speeds for MSL entry capsule. Aerothermal augmentation was not expected, nor was there any expectation of the significant aero-rcs interaction at hypersonic speeds, the latter partly due to a misunderstanding of aftbody pressure and aftbody contribution to aerodynamics at hypersonic speeds, in the context of the RCS authority. It was thought, that since the forces and moments on a hypersonic capsule are dominated by the forebody, no reasonable amount of change of aftbody environment can be significant. Later relating aftbody changes to the nominal thruster authority exposed the flaw in thinking, as for some capsules, thruster activity can trigger changes in aftbody flow that will generate moments comparable to the thruster authority. this last point was later supported with the idea, that the aftbody pressure can significantly exceed free-stream static pressure during hypersonic flight. Effect of RCS thrusters on the aerothermal environment was found later to be significant for MSL, as will be shown later. Analysis methodology developed over several years, and it benefited greatly from the increase in computer availability. Initially the method involved the basic LAURA code, where several cells on the surface were set up to allow the outflow to be specified (this was put into LAURA by Cheatwood [28]). Velocity, density and temperature were set consistent with expected thruster values, and the number of cells was set to mimic the exit area of the thruster. This approach allowed (crudely) to develop a very coarse plume as essentially a protuberance to the surrounding flow. This approach worked well enough to illustrate that aeroheating concerns may be warranted at hypersonic speeds, once a plume is introduced into the flowfield. The slightly more complex approach involved the use of MORPH tool [29] to modify the flight simulation grid in such a way that a circular grid patch was added at the thruster location. 39

50 This patch could be used to set up the flow conditions for the thruster exit, and, generally, this worked fairly well for quick analyses, RCS configuration screening, where the time to develop a topologically accurate model was not available. A topologically accurate grid would involve nozzle grid that reaches into the throat area, and into the convergent section in some cases. This grid has greater grid density in the nozzle area, and a more accurate representation of the geometry and the resultant plume interaction flowfield. This type of simulation is least flexible, due to the time required to construct such a mesh. Because LAURA was not intended for RCS simulations, there was not reason to include all species that are not met in flight but arize in RCS flows. One such specie is ammonia, NH3, a product of hydrazine combustion. Because it was not originally in the code, a non-reacting gas with transport properties of ammonia was added, to help simulation. 3.2 Analysis of Recent Flight Vehicles This section contains description of RCS-gasdynamic analyses, performed for three recent flight systems: Mars Phoenix, Mars Science Laboratory and Orion Crewed Exploration Vehicle. The work is presented for each system individually, and the conclusions are drawn at the end of each segment Mars Phoenix Capsule Aero-RCS Analysis On May 25, 2008 Phoenix successfully landed on Mars. Phoenix entered the Martian atmosphere directly from its interplanetary trajectory and executed a ballistic three-axis stabilized non-spinning entry into the atmosphere shedding its initial energy to levels appropriate for a safe landing. Phoenix is the first Mars mission to execute a non-spin-stabilized entry from such high velocity; all successful missions before Phoenix employed different strategies. In 1976, Vikings 1 and 2 performed controlled unguided entries from circular orbit, whereas more recent missions, namely Pathfinder (1997), and Mars Exploration Rovers (2004) entered directly on interplanetary trajectories but utilized spin-stabilization. Because the next Mars mission, Mars Science Laboratory (MSL), is designed to fly a guided lifting entry from an interplanetary approach, the flight experience of Phoenix was thought to be very valuable. Phoenix capsule is shown in the figure 3.1. While originally designed to perform a lifting guided entry, the final configuration of Phoenix was to fly a ballistic entry. Its Reaction Control System was intended to provide rate damping and roll control during entry. The RCS is composed of hydrazine thrusters capable of generating control torques, fuel tankage, and control valves operated by the flight control program. During the course of the atmospheric entry the RCS thrusters would be fired to produce torques about the pitch, yaw and roll axes, commanded by the control program. The control 40

51 program may issue commands at any time during entry. Therefore, the control system must be effective in all regimes from rarefied to supersonic, where parachute deployment occurs. The RCS thrusters fire into the capsule s wake. During operation, the thruster effluent interacts with the wake and alters it. Because pressure on the backshell of the capsule is not zero, interactions between thruster plumes and the capsule wake can cause a change in the aftbody pressure distribution. One possible result is the emergence of capsule moments that may compete with the native authority of the control system. It is possible to generate aftbody moments that create gain in a given channel or cross-coupling into other channels. An example of gain would be if in response to activation of pitch thrusters the capsule would develop an aerodynamic pitching moment, that would add or subtract from the native RCS moment. An example of cross-coupling would be if in response to activation of roll thrusters, the capsule would develop some pitch moment, such that pitch thrusters would have to be used to counter it. Effects of RCS-wake interaction depend greatly on the details of the local flow in the vicinity of the thruster exit. Properties of the local flow depend on the free-stream parameters, atmospheric composition, capsule size, shape and attitude and the location of the thruster. The local flow can be part of the attached wake, in which case its typically supersonic, or it can be a part of separated wake, in which case it is typically subsonic. Under the influence of some RCS thrusters, a separated wake may be forced to reattach, whereas an attached wake may be forced to separate. Generally, attached flow is more energetic, and interactions between it and the thruster effluent can produce shock structures referred to as horse-shoe shocks. Such structures develop a quazi-nozzle-like flow directed toward the surface, essentially creating a high energy stagnation flow at the surface of the capsule upstream of the thruster exit. This type of an interaction can result in a significant increase over the baseline in heating, pressure, and shear at the surface. Irrespective of the character of the local flow, any interaction between thruster effluent and local flow will result in changes to the wake. This is due to the fact that much of wake is subsonic, and changes in any given location affect any other location that is within the elliptical boundary. The result of this kind of dependence is that changes in surface environments can occur over most of the rear wall of the capsule when an RCS thruster is fired. Most of the environmental changes that occur within the separated part of the wake are small, however, any interaction with energetic flow outside of the wake shear layer, like the kind that will happen if the plume from the thruster nozzle punches through the separated zone and into more energetic flow, can result in significant changes in surface environments. Specifically, a change in surface pressure distribution will produce moments on the capsule, which can interfere with the native authority of the RCS. The current approach to analysis of RCS-induced control interference is to use state of the art numerical techniques for flight predictions at flight conditions, and to use ground test facilities for validation. Because of time constraints, it was not possible to develop an experi- 41

52 mental program to support the analysis of the efficacy of the Phoenix RCS. As such, the present analysis methodology relied entirely on computational techniques, while some validation was provided through ground testing of RCS effects by the Mars Science Laboratory (MSL) Project. Because RCS thrusters exit into complex wake flow (Fig. 3.2), they induce diverse flow interac- Figure 3.1: Phoenix capsule geometry tion phenomena. While these interactions may alter both the aerodynamic characteristics and aerothermodynamic environment of the capsule, this paper will specifically focus on analysis of the former (i.e. the induced aerodynamic moments). The objective of the present analysis is to determine the cumulative effect of changes in the basecover pressure distribution on the RCS control authority. Phoenix was designed to fly a ballistic three-axis stabilized trajectory. Trim angle of attack is near zero for most conditions, except when bounded instability occurs (see, for example, Gnoffo [30], Edquist [31]). The RCS in this scenario is used mainly as a rate damper. The Phoenix RCS consists of four 26.3 N thrusters (TCM 1, 2, 3 and 4), used for pitch and yaw corrections and four 5.7 N thrusters (RCS 1, 2, 3 and 4), providing roll authority. Layout is shown in Figs. 3.3 and 3.4. Figures 3.5 and 3.6 identify thruster firings to obtain positive pitching and yawing moment. Because of a small moment arm, the yaw thrusters (Figure 3.6) provide control authority of only 10.5 Nm as opposed to 57 Nm for the pitch thrusters (Fig. 3.5). 42

53 Figure 3.2: Illustration of the flowfield around Phoenix capsule The objective of the present analysis is to determine the magnitude of aerodynamic moments, developed on the aftbody of Phoenix entry capsule because of the interaction of RCS thruster plumes with the wake. The interference torque is defined as: M interference = Cm interference S ref L ref 1 2 ρv2 (3.1) where Cm interference = Cm T CM Cm Baseline (3.2) Cm T CM is the aerodynamic moment on the capsule, whose surface pressure distribution is perturbed by presence of the thruster plumes. Cm Baseline is the aerodynamic moment on the capsule in the baseline flow unperturbed by the thruster. Note, that torque due to thrust of the nozzles does not enter the definition of the interference moment. However, we can use it to define control gain: Gain = T T CM + M interference T T CM (3.3) When the gain is less then unity, interference torque is creating a deficit of authority. When the gain is greater then unity, a surplus of authority is caused by the interference torque. Figures 3.7 and 3.8 show acreage distribution of the available moment arm about yaw and pitch axes for the Phoenix aftshell. Moment arms are computed about the center of mass of the capsule. Plots indicate that there are areas, typically near capsule maximum diameter and 43

54 Figure 3.3: Phoenix capsule features, image borrowed from project material. near parachute cone, where the moment arm takes on a large positive or negative value. If surface pressure in one of those areas was altered due to the interaction of thruster and wake, an appreciable moment would be developed. Present analysis of the RCS-induced interference effects covered the entire entry from rarefied to supersonic regime. Calculations were performed at a rarefied flow condition, corresponding to KnD = 0.1, at high hypersonic condition of Mach 27.2 (which corresponds to a KnD = ), at hypersonic Mach 18.8 (which is roughly peak dynamic pressure on Phoenix entry trajectory), and at supersonic Mach 3. These points are shown in the Fig Mach 18.8 was selected because the aftbody pressure goes through a maximum near peak free-stream dynamic pressure. Mach 3 condition was selected because of relatively high expected RCS activity during supersonic flight (due to a dynamic instability [31]) and because of a large contribution to the overall capsule moments expected from the aftbody. Aftbody pressure can be estimated from a base correction curve. Figure 3.10 shows the variation of dynamic pressure with Mach number and shows variation of basecover pressure, as computed from Eq. (2.1). The two points on the plot are CFD-predicted pressures. Generally, CFD follows the trend of the curve, but magnitudes don t always agree. Notably, the plot of base pressure indicates two regions where aftbody pressures peak. One of these regions is near peak dynamic pressure on the trajectory and another occurs during supersonic flight. These are the regions of most interest from the point of view of aero-rcs interactions. Analysis at supersonic and rarefied conditions is not 44

55 Figure 3.4: Phoenix RCS layout, image borrowed from project material. presented here, but is described in detail in [32]. A rarefied regime condition was selected to confirm that RCS interference at this early stage in entry should not be significant. The Mach 27.2 condition was selected to verify qualitative agreement with results of analysis in the rarefied regime. Neither analysis attempted to simulate pulsed operation of thrusters. Because of the limited time accuracy of codes and high computer cost this was thought impractical. It is believed that the current set of simulations with a continuous thruster firing should be bounding of the expected phenomena. Analysis in the hypersonic regime was carried out using Langley Aerothermodynamic Upwind Relaxation Algorithm (LAURA)[12], [13], [11] code. LAURA is a parallel three dimensional multi-block program that is extensively used in aerothermodynamic calculations of entry vehicles. LAURA can solve Euler, Thin Layer Navier-Stokes and full Navier-Stokes flow-fields using an upwind point- and line-implicit relaxation. The code can solve laminar and turbulent flows with and without thermo-chemical non-equilibrium. A wide range of chemical and thermal wall conditions are available. In the present study an eight species Mars gas model (CO2, CO, C, N2, N, NO, O2, O) allows thermo-chemical nonequilibrium. Flow is laminar and the radiative equilibrium wall temperature condition with ɛ =.89 (representative of charred Super Lightweight Ablator (SLA)) is specified. A super-catalytic wall boundary condition is used, such that species concentrations near the wall are set to their free-stream values of 97%CO2 and 3%N2 by mass. Computational 45

56 Throat Position (m) Thruster X C Y C Z C RCS RCS RCS RCS TCM TCM TCM TCM Scarfed Thrust Direction Vectors Thruster X C Y C Z C RCS RCS RCS RCS TCM TCM TCM TCM Figure 3.5: Pitch firing configuration grid for baseline (no thruster firings) calculations had 2.6 million points. This grid was modified to accommodate a thruster with use of Grid Morphing Software (MORPH) tool, developed at NASA Langley for the Shuttle Return to Flight (RTF) activities. The modified grid contained 5.9 million points. All solutions, generated for this analysis are symmetric about the α total plane. This approach reduced the computational requirements. The interior of the thruster was not modeled. Instead, Chemical Equilibrium and Applications (CEA) program [24] was used to determine nozzle exit conditions, given the nozzle area ratio and chamber pressure and temperature. The fuel is hydrazine (N2H4) and the products of its combustion are ammonia, hydrogen, nitrogen and traces of unspent hydrazine. CEA assumes equilibrium process through the convergent part of the nozzle, but from the throat to exit all reactions are frozen. Conditions computed at the nozzle exit are written into the modified CFD grid to emulate a thruster. In 46

57 Figure 3.6: Yaw firing configuration Figure 3.7: Moments about X-axis the CFD solution the thruster effluent is modeled as non-reacting ammonia. Because of the large disparity between forces on the forebody and aftbody of the capsule during hypersonic flight, it is possible that numerical errors in evaluation of forebody moments will be comparable in magnitude to the RCS thruster-induced interference moment that 47

58 Figure 3.8: Moments about Y-axis Figure 3.9: Investigated conditions is sought after. In other words, a small interference moment can be indistinguishable from numerical noise of the forebody solution. To isolate the desired information the moment summation is performed on the aftbody only. Algorithmically this means that the forebody solution is converged and frozen and becomes an input to the aftbody solution (wake). 48

59 Figure 3.10: Variation of dynamic pressure and basecover pressure The wake flow is unsteady, and the surface pressure distributions in the separated region oscillate with iteration. This oscillation results in an unsteady moment, which requires that the output be averaged over a number of iterations. The CFD iteration uses local time advance, but sometimes it is necessary to use global time advance to get through difficult transients in the wake. Physically, all of the wake should be advanced at the same rate using a global time step, but associated computational cost and a limited temporal accuracy of the code make this approach impractical. Instead, a local time step was used in this analysis with an understanding that unsteady processes within the wake evolve at their own rates, not necessarily consistent with each other. Figures 3.11 and 3.12 illustrate the unsteady moment output at a Mach 18.8 condition. The iteration history of the aftbody pitching moment about the center of gravity (CG) is shown. Figure 3.12 contains information from the Fig. 3.11, reduced by the Eq. (3.3). This step provides the direct measure of the effectiveness of the RCS thruster. Note that the scale is exaggerated, and a relatively small native RCS authority is illustrated. In the hypersonic regime calculations were performed at Mach 27.2 for yaw thrusters and Mach 18.8 for pitch and yaw thrusters. Mach 27.2 calculations included sideslip angles β=6 and β=10. Yaw thrusters were fired on the lee-side of the capsule. Solutions at Mach 18.8 were limited to angle of attack of α=10 for pitch and an angle of sideslip β=10 for yaw thrusters. Figures show computed surface pressures at Mach 27.2 with and without the thruster firings. Note the reduction of surface pressure and delayed separation when the thruster is on. It is hypothesized that this is happening due to entrainment of gas by the thruster plume, 49

60 Figure 3.11: Iteration history of aftshell moment Figure 3.12: Iteration history of control gain as it punches through the supersonic shear layer. Figure 3.17 shows the aftbody moment coefficients computed for the two angles of attack with and without the thruster firing. For reference, the nominal, or native yaw authority is plotted as a dashed line. The error bars do not indicate the level of uncertainty. Instead, they reflect only the variability of the moment output 50

61 with iteration. Uncertainty would have to be added on top of this variation. As seen in the figure, the predicted interaction, or Cm interference is significantly below the nominal authority. In other words the CFD solutions do not show a significant control degradation in yaw (or pitch, for that matter) at this high altitude flight condition. These results are in concurrence with the result of analysis in the rarefied regime, namely, unless aftbody pressure is high enough to produce an appreciable torque on the capsule, it is highly unlikely that disturbing the wake would result in significant levels of interference torque. It should be pointed out that Mach 27.2 condition produces partly rarefied wake, and pockets of rarefied flow can be found near shoulders. Figure 3.13: Mach 27.2 and 6 deg sideslip without the thruster Figures 3.18 and 3.19 show pitch and yaw control authority on top of the static stability curve at Mach 18.8, which corresponds to the peak of dynamic pressure on the trajectory. As the figures indicate, RCS is not effective at changing the attitude of the capsule at this Mach number. Pitch thrusters are capable of about 0.5 change in attitude. Yaw thrusters, due to lower native authority of 10.5 Nm, can only produce about 0.1 change in the sideslip angle. As Figs and 3.21 indicate, the native moment is comparable to the variability of aftshell CFD solutions. Plots show nominal authority, baseline aftshell moment and the aftshell moment, perturbed by the flow due to interaction with the thrusters. Figure 3.20 indicates that the thruster on the lee-side (α=-10 ) of the capsule results in favorable difference 51

62 Figure 3.14: Mach 27.2 and 6 deg sideslip with the thruster Figure 3.15: Mach 27.2 and 6 deg sideslip without the thruster in authority (i.e. Cm interference = Cm T CM Cm Baseline has the same sign as the nominal RCS thruster authority moment), while the thruster on the windside (α=10 ) results in an adverse effect (Cm interference is opposing nominal authority). Solution with the leeside yaw thruster 52

63 Figure 3.16: Mach 27.2 and 10 deg sideslip with the thruster Figure 3.17: Yaw interaction at Mach 27.2 for 6, 10 deg sideslip (Fig. 3.21) shows an adverse interference torque. Again, no uncertainties are applied to these calculations, and error bars only indicate the solution variability with iteration. Because yaw thruster authority is so small in comparison to both the solution variability and the interference 53

64 Figure 3.18: Pitch control and aerodynamics at Mach 18.8 Figure 3.19: Yaw control and aerodynamics at Mach 18.8 moment it can not be said with any certainty that the RCS will perform acceptably during flight near peak dynamic pressure. Results presented in the preceding sections suggest that the Phoenix RCS system may not be able to effectively control the capsule in every flight regime due 54

65 Figure 3.20: Pitch interference at Mach 18.8 Figure 3.21: Yaw interference at Mach 18.8 to sugnificant interaction between the aerodynamic flowfield and thruster plumes. Calculations in high hypersonic regime indicate that problems there are unlikely, mainly because of low aftbody pressures. There is simply not enough pressure to generate a significant interference 55

66 moment. Computations at a lower hypersonic point near peak dynamic pressure condition suggest significant control authority degradation in the pitch channel and a possible reversal in the yaw channel. These results are given prior to adding any uncertainty. Because of this, there is little confidence in the effectiveness of the Phoenix RCS during hypersonic and supersonic flight, moreover, a possibility exists of a control reversal. Numerical analyses of the efficacy of the Phoenix Reaction Control System (RCS) showed that the system might be inadequate to control the capsule during entry due to a significant interaction between the wake and the plumes of the RCS thrusters. Specifically, significant control degradation in pitch and control reversal in yaw may occur near peak dynamic pressure and at supersonic speeds. In addition, significant cross-coupling into pitch and yaw channels due to use of roll thrusters during supersonic flight is predicted. Based on the results it can not be said with any certainty that the Phoenix RCS system would perform adequately during most of the continuum regime of the atmospheric entry. The techniques utilized in these analyses have known issues. Calculations assume a thruster that is constantly on due to the limitations of current state-of-the-art techniques. However, it is believed that the present approach bounds the phenomena. Similarly, because of the complexity of the capsule wake it is not practical to attempt to achieve grid independence. Increasing the grid level was found to produce a non-monotonic response of the moment output. In summary, the control authority of the Phoenix RCS system is low, particularly in the yaw channel. At some flight conditions the RCS control moment is lower then the level of unsteadiness in a baseline Computational Fluid Dynamics (CFD) solution. This illustrates that present CFD techniques are at or beyond their limit when analyzing RCS control authority with such a low native moment. It also illustrates that it is beneficial to design an RCS system with greater native moment capability, so that these problems can be avoided. Because of the issues identified through the present analyses, the Phoenix Project changed its plan regarding the use of the RCS system during atmospheric entry. Specifically, the RCS system deadbands were sufficiently widened for flight through the continuum regime as to essentially eliminate any thruster firings. Consequently, Phoenix became the first ballistic uncontrolled non-spinning entry. 56

67 3.2.2 Mars Science Laboratory RCS-Aerodynamics and RCS-Aeroheating Analysis MSL capsule (figure 3.22) is equipped with Reaction Control System (RCS), which will enable rate damping and bank maneuvers during entry. Because RCS jets exit into the complex aftbody flow, they can induce diverse flow interaction phenomena, as shown for example in the Figure These interactions may alter aerodynamic characteristics and aerothermodynamic environment of the backshell of the capsule. This section will focus mainly on analysis of the former, i.e. the induced aerodynamic moments that impact RCS effectiveness. The objective is to determine the cumulative effect of changes in the basecover pressure distribution on RCS control authority so that the interference moment can be determined. Much of this work is described in greater detail in [33]. Figure 3.22: Flow around MSL capsule at Mach 18.1 where M interference = Cx interference S ref L ref 1 2 ρv2 (3.4) Cx interference = Cx RCS Cx Baseline (3.5) 57

68 Figure 3.23: Jet-wake interaction This interference moment can be in the axis of the active RCS thruster, in which case it contributes to RCS gain, or to one of the other two axes, in which case it produces cross coupling. 58

69 MSL Unitary Tunnel Test A test of aero-rcs interaction was carried out at NASA Langley Research Center s Unitary Plan Wind Tunnel (UPWT). The first round of tests focused on the supersonic (below Mach 4.5) performance, and data was collected at Mach 2.5, 3.5 and 4.5 for a range of Reynolds numbers and jet pressure ratios. Computational analysis, performed by the author in support of this test and comparison with some of the data is presented in this section. Ambient gas was air and unheated nitrogen was used as jet effluent. Data was collected for the individual firings of pitch, yaw and roll jets, and combination firings. The model design for this test (see figure 3.24) reflected the then current MSL entry vehicle (outer mold-line designation 6) and the then current RCS system layout. The data, collected in this test series was compared with CFD predictions and the results were used as an estimate of confidence in numerical predictions [34]. Model nozzles were conical with a 10-degree half-angle of the divergent section. Throat diameter, nozzle half-angle and nozzle chamber pressures were selected to produce jet plumes as closely scaled to those in flight as possible. The model was mounted on a five-component balance. The sixth component was sacrificed to make room for nitrogen supply into the model. LAURA CFD calculations were performed with perfect gas air laminar model, and nozzles included convergent section as shown in the figure Comparison of predicted and measured interaction moments for pitch thruster firing at Mach 2.5 condition is shown in the figure The error bars on the experimental data indicate the uncertainty in moment measurement, and not the uncertainty in the interaction coefficient itself, which can be significantly lower. Generally CFD predictions describe a comparable picture as the data does, but there are clearly disagreements in both the trends and magnitudes. Moment integration for these comparisons includes only the aftbody surfaces not including the sting, and not including nozzles. Nozzles are excluded because thrust is not a part of the difference in the definition of the interference moment. Forebody of the capsule is excluded because the round-off error in force-moment summation there can overshadow the aero-rcs interference on the aftshell. Figure 3.27 shows the LAURA moment output for several angles of attack at Mach 3.5. In the figure an uninterrupted string of data is shown as the angle of attack in the LAURA simulation is set at 5, 10 and 15 degrees. Each angle of attack range starts with baseline (no jet flow) run, which is allowed to settle, and then thruster is turned on). Change in the magnitude is due to thruster-wake interaction, and essentially defines the aero-rcs interaction magnitude. As the angle of attack increases, the oscillations in the moment output reduce. 59

70 Figure 3.24: Layout of OML 6 RCS model showing roll and pitch-yaw jets Figure 3.25: Detail of the grid near nozzle (every other gridpoint shown) 60

71 Figure 3.26: Mach 2.5 comparison of LAURA predictions with data Figure 3.27: Mach 3.5 moment coefficient with and without jets over a range of angles of attack 61

72 Additional aspect of relating experimental data to flight is the model support interference. In the case of RCS testing it is difficult to reduce sting diameter due to the space occupied by the force-moment balance, and the required gas supply. Figures 3.28 and 3.29 show centerline cut through the flowfield with and without the sting. At angle of attack the interaction of the shear layer with the sting produces a different flowfield from the case where the sting was removed from the computation. It is reasonable to expect difference in aerodynamics as a result. Figures 3.30 and 3.31 show pressure slice through the thruster location at 0.8 inches from the centerline for the same flowfields, showing high pressure nozzle flow. Forebody flow is not shown as it is unaffected by the sting. Reduction of the surface pressures from these simulations (10 degrees angle of attack shown, 5 and 15 degrees are included in comparison) reveals the effect of the sting on predicted aero-rcs interference. Results are shown in the figure As the figure indicates the difference between the two sets of solutions is significant. As the angle of attack increases (in this case, nozzles move more and more to wind) there s less and less interaction. Because the shear layer is so close to the surface in absence of the sting, most of the plume gets cut off by the shear layer and its influence on the surface pressure is minimal, as compared to the case with the sting, where more volume is available for plume diffusion into the recirculated wake. Figure 3.28: Mach 3.5 flowfield with the sting 62

73 Figure 3.29: Mach 3.5 flowfield without the sting Figure 3.30: Mach 3.5 flowfield pressure with the sting 63

74 Figure 3.31: Mach 3.5 flowfield pressure without the sting Figure 3.32: Mach 3.5 comparison of moments, effects of support sting 64

75 Effect of nozzle pressure is frequently of interest, as the nozzle pressure is related to the momentum, and hence helps understand relevance to the flight environment. Over narrow range of pressures it is possible to obtain a relatively linear response, but in general, the effect of nozzle pressure on interaction coefficient is not linear, when pressure is varied over a broad range). Figures 3.33 and 3.34 show effect on flow structure and surface pressure. Integrated interaction pitch moment is shown in the figure The figure indicates areas of distinctly different behavior that appear as nozzle pressure is varied across the considered range. As the pressure is increased the aftbody is evacuated, resulting in more influence to the interaction coefficient coming from areas further from the nozzle. High pressures, considered here, are not realistic, and only included to illustrate the non-linear nature of aero-rcs interaction. In the smaller range of pressures the results are linearizable, as shown in the figure 3.35 for pressures between 0 and 500 psi. Figure 3.33: Slice of flowfield showing structure at range of nozzle pressures 65

76 Figure 3.34: Surface pressures from solutions with different nozzle pressures Figure 3.35: Effect of nozzle pressure on interaction moment Evolution of MSL RCS Configuration Mars Science Laboratory RCS has evolved due to a number of constraints. The initial configuration had thrusters arranged as shown in the Figure Thrusters provided nearly independent torques about all three axes and their layout idea was fairly analogous to the one, used by the Viking landers shown for reference in Figure In an effort to avoid interaction of RCS plumes with the parachute risers, the thrusters were moved to the rear of the capsule and oriented away from the capsule s centerline (Figure 3.37). At about the same time the capsule had grown an added rear volume to cover parachute 66

77 Figure 3.36: Viking RCS[35] mortar, also shown in the figure. This configuration was extensively studied for aerothermal and aerodynamic RCS interactions. Of primary concern was the enhanced heating due to the windward thruster[36]. Flowfield interaction of the windward thruster plume is shown in Figure Additionally, because of the relatively high angle of attack of MSL capsule much of the windside aftbody sees attached hypersonic flow, as shown in the Figure Figure 3.39 shows the local flow environment, indicating possibility of attached approaching flow. Figure 3.40 shows a 2nd iteration RCS cover in a representative local flow (rotated so the flow is top-to-bottom). As the figure indicates, this thruster configuration left the exposed nozzle openings in the path of high-energy attached flow and no solution could be found to protect nozzles from overheating. An alternative solution was to swap the places of the windside and leeside thrusters, such that covers could be used to protect the windward thrusters as shown in the Figure This configuration suffered from excessive RCS-induced aerodynamic moment when yaw thrusters were fired. CFD calculations also showed that the RCS effluent mixes with the entire recirculating wake regardless of thruster orientation and it is not possible to avoid contact between RCS effluent and parachute hardware. Figure 3.41 shows a computed boundary of the RCS effluent. Effluent fills all of the recirculation zone and gets convected aft as it mixes with the rest of the ambient gas. Because of this finding and due to the significant aerothermal and aerodynamic challenges posed by the RCS layouts in Figures 3.37 and 3.38 the thruster layout was modified so as to minimize aero-rcs interference and aerothermal load on thrusters. The principle differences 67

78 Figure 3.37: Second iteration of RCS Figure 3.38: Third iteration of RCS 68

79 Figure 3.39: Flow environment Figure 3.40: RCS cover heating, qualitative. 69

80 Figure 3.41: Thruster effluent mixing with the capsule s wake between this final configuration and the one in the Figure 3.38 are that the thrusters were rotated out of the base plane of the capsule, alleviating thruster interaction, and were moved outward from the capsule s centerline so as to allow thrust direction to cross the pitch plane ahead of the CG with significant moment arm. The benefits of this configuration are discussed in the following section. Sources of RCS interference There are two main sources of RCS-Aero interaction. First, the under-expanded jet produces a change in the near-exit flowfield, causing entrainment and reduction in pressure in the nearfield. If this jet impinges onto the surface, or collides with another jet, a local increase in surface pressure may result. This type of a near field interaction is relatively invarient with the trajectory condition, and is fairly easy to analyse. The second type of RCS interaction occurs due to the jet influencing the rest of the capsule wake flowfield, causing global changes in the capsule pressure field. This change is much more complex and it depends on the flight conditions. Figure 3.43 shows the pressure distribution and a plume outline for one of candidate RCS systems for MSL. In this system, pairs of jets were mounted as shown (although only the thrusters of the right side are included in the figure) and the four right jets would be fired to achieve yaw torque. The four jets formed a complex interaction flowfield with an impingement footprint in the near field that was shown through CFD to be invariant with free-stream conditions. The four jets formed a larger plume that proceeded laterally to the side, which is near the bottom in the figure and interacted with the capsule s shear layer. The result was additional pressurization of 70

81 Figure 3.42: Final thruster arrangement of MSL RCS the backshell surface near the shoulder. Increased pressure forces over broad acreage produced a moment that countered the intended yaw authority. Most of the interaction torque came from the pressurization near the shoulder. In this case all of the yaw authority was negated by the aerodynamic torques. Figure 3.44 shows the distribution over the aftshell of the yaw-axis moment arm. This shows regions where pressure changes have a large effect on the moments about CG. Because some RCS activity produces unbalanced changes in the pressure distribution (like the example in the figure 3.43) it is beneficial to design the RCS such that the areas of high sensitivity are not affected by thrusters. Effect of aftbody pressure on RCS interference Aftbody pressure during flight was discussed in detail in the section Description of the Problem. Local pressure in the vicinity of the RCS jet is believed to play an important role in determining the extent to which surface environment will be affected by jet-flow interaction. Calculations indicate that the maximum interference effects occur at flight conditions near peak dynamic 71

82 Figure 3.43: MSL aftbody pressure, yaw jets, candidate RCS layout, computed for Mach 18.1 Figure 3.44: Aftshell surface yaw-moment arm distribution pressure on the trajectory, where aftbody pressure is maximum. Earlier in the trajectory, when the dynamic pressure is low there is simply no sufficient gas pressure on the aftbody to generate a significant interaction with a jet. A somewhat similar effect occurs at much lower speeds, where a reduction in aftbody pressure on the left side of the dynamic pressure curve (see Figures 72

83 3.45 and 3.46) reduces the effectiveness of any wake-rcs interactions. As the figures indicate, lower dynamic pressure of Mach 5 flight condition is not sufficient to produce any of the shock interaction phenomena, present in the Mach 18 solution. What remains a question is the fidelity of calculated trends in pressure over the rest of the separated flowfield. Figure 3.45: MSL aftbody pressure, yaw jets, computed for Mach 18.1 Figure 3.45 illustrates typical interference pattern in presence of a jet: there s a compression front formed by the interaction of the underexpanded jet pair and the supersonic crossflow, and a shadow region, formed downstream of the jets. This pattern occurs in the area of the windside (lower) RCS jet pair, and it is not present in the area of the lee-side (upper) RCS jet pair. This should be the case, as the upper jets are in the area of fully separated flow (see, for example, figure 3.22, except that graphic is inverted in relation to 3.45 and 3.46). Heat flux is affected by the jets in a similar way as pressure, and generally most of the qualitative trends translate between heating and pressure. Computational Analysis of RCS-Aerodynamic Effects in Flight An analysis of RCS-induced aerodynamic effects for the final iteration of thruster layout has been conducted at hypersonic and supersonic conditions. In this section some details of the methodology for flight calculations in the hypersonic regime and the main findings are outlined. Laminar Navier-Stokes calculations were carried out using LAURA [11] code. Calculations 73

84 Figure 3.46: MSL aftbody pressure, yaw jets, computed for Mach 5 included 8-species non-equilibrium Mars gas (CO 2, CO, N 2, O 2, NO, C, N, O) plus ammonia (NH 3 ) as a non-reacting RCS surrogate gas. Thruster effluent is composed of a mixture of H 2, N 2 and NH 3, but the makeup is simplified. Two types of LAURA calculations were carried out: in the preliminary calculations the representative nozzle exit plane gas properties were obtained from CEA[23] code and written into LAURA as a representative exit boundary condition. This type of a solution allowed for a quick turnaround necessary for configuration screening. The second round of calculations included an internal CFD domain that extended into the nozzle until the throat. Gas properties at the throat were evaluated using the CEA code and superimposed onto CFD solution. Grid complexity for the second type of calculations makes for significantly longer runtimes, rendering these solutions suitable only for the analyses of what s thought to be a reasonably established system. Figures 3.47 and 3.48 show CFD grid layouts for the third RCS layout and for the final layout. CFD grids are complicated because LAURA requires one-to-one block connectivity, which means that a boundary surface of one block can only be connected to one boundary surface of another block. Calculations were carried out at Mach 18 flight conditions for pitch-up, pitch-down, yaw and roll thrusters at an angle of attack of 17 degrees, which is representative of the expected MSL trim incidence during hypersonic flight). All solutions indicated induced aerodynamic torques on the order of 5-10% of nominal RCS authority. An additional solution for yaw thrusters was performed at the higher angle of attack of 27 degrees and a sideslip angle of 10 degrees. This solution indicated aerodynamic coupling into the pitch axis of about 10% of nominal authority. 74

85 Figure 3.47: Nozzles in grid, configuration 3 Nozzles in grid, final configu- Figure 3.48: ration From calculations the main contributor to the moment is the impingement of the thruster-wake interaction feature onto the bottom of the parachute closeout cone (PCC). This moment is in the direction such as to reduce the vehicle s angle of attack. Figures 3.49 and 3.50 show the baseline surface pressure and the surface pressure with yaw RCS thrusters active. Figure 3.49: Mach 18 high incidence baseline solution 75

86 Figure 3.50: Mach 18 high incidence yaw solution In the context of the aerothermal analysis of MSL RCS[41] several calculations on higher fidelity grids at matching Mach 18.1 conditions and with an analogous model were carried out in LAURA and DPLR[14]. Resultant pressure maps are shown in the Figure 3.51 borrowed from [33]. Pressure in the strong interaction features agreed fairly well between the codes. The pressure distribution in the leeside separated zone had greater variability. LAURA calculation showed 20% increment in yaw due to aerodynamic interaction, and DPLR calculation indicated 10% increment. Both codes predicted some aerodynamic coupling into pitch axis due to pressurization of the PCC and change of pressure on the lee side of the capsule. In the case of the LAURA solution, pressurization of the leeside and PCC was more significant, predicting near 15% pitch torque in the direction of reducing angle of attack, which is consistent with the results of lower-fidelity grid calculations above. DPLR solution predicts 5% torque in the opposing direction. Variability between codes is one of the measures of uncertainty in these predictions. It should be noted, however, that the codes are solving essentially an analogous model of the flow field on very similar grids, and any such comparisons do not capture limitations of the model itself. In other words, the same assumptions of laminar flow, grid resolution and gas model are made for both codes, and the differences in the solution represent mainly the differences between algorithms. 76

87 Figure 3.51: Mach 18 LAURA-DPLR comparisons with and without RCS, from [33] RCS Design Philosophy Because of the high drag requirement, entry capsules typically have a high ratio of projected area to volume. This results in a broad subsonic shocklayer that is joined to the massively separated wake by rapidly expanding supersonic shoulder flow as shown in the Figure The influence of the jets on the aftbody surface pressures depends on the local and RCS flow and the thruster size, placement and orientation. Because the aft-cover is shaped with the primary goal of accommodating the payload, structure, cruise stage mounting etc. its role as an aerodynamic surface is frequently viewed as secondary. The resulting aftshell shapes can have regions with large moment arms about the CG as shown in the Figures 3.52 and The effect that the changes in pressure distribution due to the RCS activity may have on the capsule moments is difficult to anticipate due to the complexity of the interaction and the uncertainty in analysis tools. Experience with the analysis of the aerodynamic and aerothermal RCS effects has yielded several working paradigms that have been applied to the MSL RCS design. Because the interactions are strongest when the jet is aimed against the oncoming supersonic flow, it is preferred to direct RCS engines with the oncoming flow, or to place them in such a way, that the jet plumes would be contained entirely within the re-circulating region. The latter may not be possible, as the re-circulating regions shape and size may not be adequate. If strong 77

88 interactions between the jet and surrounding flow are unavoidable, it may be possible to have such interactions that result in favorable capsule moments (ex. Figure 3.54), or almost no moments. Mapping the surface moment arms can help understand where the interactions can be favorable. Figure 3.56 illustrates the recent MSL RCS design that follows this philosophy with good success. When compared to the one, shown in the Figure 3.55, the reduced effects of the jets on the surface pressures are evident. Generally, achieving the same ideal control torque by a smaller engine with a larger moment arm should reduce the interactions, and should be pursued if possible. Figure 3.52: X-moment arm lengths for each point on the MSL backshell w.r.t. the CG 78

89 Figure 3.53: X-moment arm lengths for each point on the MSL backshell w.r.t. the CG Figure 3.54: Thrust direction options Conclusions Design concerns and considerations for Mars Science Laboratory entry vehicle reaction control system (RCS) are presented. It is demonstrated through analysis that RCS-aerodynamic interactions during entry can be significant. The flight conditions during which the largest interactions may be expected were identified. It is determined that large aero-rcs interference torques can be developed during hypersonic flight where dynamic pressure is high. Numerical methods used to assess these effects are being developed and tested. Accurate characterization of RCS interference aerodynamics through experiments is challenging partly because the regime of interest is hypersonic flight. MSL RCS has evolved under the influence of mechanical, aerodynamic, aerothermodynamic and other constraints. In the course of the analyses of the RCS effects several off-designs were 79

90 Figure 3.55: Predicted surface pressures before redesign Figure 3.56: Predicted surface pressures after redesign explored to gain understanding of the design space. Based on this understanding, paradigms for RCS design were formed. These paradigms are consistent with the layout philosophy of the control system of the Viking landers. It is shown through CFD analysis that the RCS aerodynamic interaction effects should depend greatly on the jet location and direction. Understanding gained from this work has been applied to MSL flight vehicle. All analyses of the final MSL RCS indicate low to moderate levels of aero-rcs interference. Wind tunnel tests are needed to increase understanding of the phenomena and develop and validate CFD tools. In this regard, the benefits from an instrumented flight are especially evident. In addition, CFD has been 80

91 shown to be a tool that can be used in mapping out the aerodynamic and aerothernodynamic design space for RCS interactions on Mars entry vehicles. 81

92 3.2.3 Analysis of Orion CEV RCS aeroheating The Orion Earth entry capsule is being designed to be a part of the Constellation architecture, the next generation of NASA s human space exploration systems. The Orion capsule will carry crew to and from space, and its mission calls for a lifting actively guided profile of its reentry into the Earth atmosphere. During atmospheric entry Orion will use aerodynamic lift for guidance maneuvers and a reaction control system (RCS) will be used to alter direction of the lift vector and capsule s attitude rates in all flight regimes as required. Because the RCS will produce under-expanded jets of thruster exhaust, the effect of the interaction of these jets with the capsule s gas-dynamic environment must be understood. Namely, effects of jets on the capsule s aeroheating is of interest insofar as it can impact sizing and design of the Thermal Protection System (TPS), and the effect of jets on the pressure field around the capsule needs to be assessed to determine if any significant interference with RCS control authority might be expected. While there is a significant overlap in the means to achieve the two objectives, the present report will focus only on the aerothermal environment. Present section is an abbreviated version of an AIAA paper [40]. Description of Orion Reaction Control System Because Orion is designed to fly a lifting actively guided trajectory, the capsule is equipped with a Reaction Control System (RCS) composed of 12 hydrazine (N2H4) thrusters, placed on the aftshell in an arrangement somewhat similar to that of Apollo system. Thrusters are configured to produce pitch and yaw torques to help dampen attitude oscillations and to control the bank angle. In the off-nominal situations roll thrusters may be used to take out persistent aerodynamic torques that might originate from uneven heatshield ablation or other aerodynamic issues that can not be mitigated otherwise due to lack of aerodynamic stiffness in roll axis [34]. As configured, the thrusters will be used at various times during entry as needed and as commanded by guidance. Thruster layout is shown in the Figure Current analysis uses a chamber pressure p = 200 psia, a chamber temperature T = 2000 F and a supersonic area ratio AR = 25. Notably most of the nozzles on the vehicle are scarfed because of the angle between the thruster axis and the vehicle s outer mold line (OML) introducing added complexity in the jet flow. Thrusters are arranged in pairs as shown. Flight Conditions Several hypersonic conditions have been identified for this analysis. Conditions are shown in the table and represent a skip entry which is currently a baseline. 82

93 Figure 3.57: Layout of Orion RCS. Free stream conditions Mach V, m/sec ρ, kg/m 3 T, K Re D α, degrees E E E E E E E E E E E E E E E E E E E E E E E E6 18 Flow Characterization In the hypersonic regime the flow around the Orion capsule is characterized by the presence of a bow shock, supersonic expansion fan, located at the shoulder, followed by separation of the supersonic flow, forming a region of low energy recirculating air, and a recompression shock (Figure 3.58). 83

94 Figure 3.58: Schematic of Orion hypersonic flowfield. Aftbody heating in presence of RCS jets Because of the long break in the use of actively controlled entry capsules, Orion is one of the first recent projects to revisit aftbody aeroheating in presence of control system jets. Two series of tests have been conducted by Buck [38] [39] in the Langley 31-inch Mach 10 air tunnel. Tests were carried out with a 5 inch diameter scale model of Orion capsule. There are differences between the jet location, orientation and flow parameters between the first and second test series. Model aftbody was constructed of plastic and a steel forebody was used. Temperature sensitive paint (TSP) was selected as a global measurement method of choice. All current tests incorporate to some degree the scaling methodology outlined by Pindzola [4]. Laminar CFD predictions have been made for the second test series. Analysis of data is in progress Flight CFD Model In order to quantify effects of RCS jets on the aeroheating environment of the Orion entry capsule a numerical model was developed. In the present model air is a 5-specie non-equilibrium mixture (below 9 km/sec) or an 11-specie non-equilibrium mixture (above 9 km/sec). The flow is laminar and the CFD domain assumes no pitch-plane symmetry, to allow for three- 84

95 dimensionality of the wake in presence of asymmetrically activated jets. Additional calculations were performed on the geometry, consistent with a 5-inch wind tunnel model as tested in Langley Mach 10 tunnel. These calculations used air as perfect gas. Mach 10 Perfect Gas Calculations, Comparisons With TSP Data Calculations have been performed for a range of wind tunnel conditions and RCS nozzle flow conditions. Figure 3.60 from [40] shows a comparison of LAURA and DPLR predictions at the conditions of LaRC Mach 10 test. Difference in predictions falls within.3% of the reference Fay-Riddell heat flux. Figures 3.62 and 3.61 show a comparison of laminar LAURA predictions to TSP data for the same conditions with a single active yaw jet. Figures 3.64 and 3.63 show the comparison of LAURA with TSP data. The comparison shows multiple regions where CFD underpredicts data. Differences in modeling features of the physical model surface, gaps and nozzles may account for some of these differences. The metal forebody used in the experiment (forebody heating data and windside aftbody heating data were not acquired for the RCS test model) has been shown to influence the heating values on the aftbody as compared to a plastic forebody. The differences and accuracy of the acquired RCS interaction data are currently being analyzed. Note that the data are only acquired on the aftbody, and colors on the forebody are not valid. Heating predictions have shown to be sensitive to surface and volume grid refinement. Solutions so far don t show monotonic convergence with grid refinement. The range of grid sizes in this study was between 6.5M and 52M points. While it may be possible to obtain a grid level that will afford grid independence, it may also be possible that the wake will not reach a steady behavior and will continue to change with every grid refinement for any reasonable grid size. It may, therefore, be reasonable to compute on lower grid levels and mind grid sensitivity as a term in the uncertainty of the method. 85

96 Figure 3.59: Representative laminar aftbody heating distribution, Run 18 condition Figure 3.60: Windside aftbody comparison, run 18, from [40] 86

97 Figure 3.61: Single yaw jet, TSP Figure 3.62: Single yaw jet, LAURA 87

98 Figure 3.63: Dual yaw jet, TSP Figure 3.64: Dual yaw jet, LAURA 88

99 Flight trajectory Conditions Current analysis of flight environments is being carried out for a series of entry trajectories. Skip and non-skip trajectories are used to define trajectory space and provide free-stream conditions of interest to the CFD analysis as shown in the Figure The red line represents a design skip trajectory and blue curve is representative of the design direct entry. The intent is to isolate effects of dynamic pressure and Mach number across various segments of the hypersonic flight regime so an engineering model of the phenomena can be constructed. Figure 3.65: Entry trajectory profiles Analysis at hypersonic conditions Several conditions have been evaluated. Figures show convective heatrates predicted by LAURA on the aftshell in presence of roll RCS jets at the conditions of 9.8, 5 km/sec, 4 km/sec and 2 km/sec in flight. Calculations are laminar and show an elevated heatrate due to RCS jet displacing the free mixing layer and recovering some of the energy of the external flow. Notably, the heatrate is expected to vary with some free-stream parameters, and additional calculations are in progress with the intent to help make assertions regarding any functional dependencies. In the mean time the information is not sufficient to determine the functionality. 89

100 Figure 3.66: 9.8 km/s heatrate (roll RCS) Figure 3.67: 5 km/s heatrate (roll RCS) Discussion of the Limitations of the Current CFD Model So far the model has several shortcomings, that are discussed here. As mentioned in the section on the mixing layer the standard practice grid alignment in LAURA and DPLR results in a 90