Multi-Disciplinary Design of a Grid Fin for a Generic High Speed Missile
June 2020
Report By: Shakunt Tambe CID: 01196160
Supervisor: Dr. Paul Bruce Imperial College London
Second Marker: Dr. David Hayes MBDA Missile Systems, UK
Submitted to Imperial College London in partial fulfilment of the requirements for the Degree of Master of Engineering in Aeronautical Engineering Department of Aeronautics Imperial College London South Kensington SW7 2AZ Blank Page Abstract
The use of grid fins for high speed missiles has recently seen a resurgence in interest. This project investigates the aerodynamic performance and the inevitable trade-offs that must be made for producibility of a low drag grid fin that is more competitive with traditional planar fins. By adopting a three-stage computational approach, CFD studies were carried out to characterise the effect on aerodynamic performance from altering local geometrical parameters such as the leading and trailing edge sharpness, thickness to chord ratio, chord to span ratio and 3D modifications such as the effect of brazed joints and tapered trailing edges. For the geometry of a Vympel R-77 grid fin, the importance of the leading edge sharpness and overall thickness to chord ratio was significant in achieving an optimum lift to drag ratio. Flow choking at the supersonic design point (Mach 3) was not found to be of concern and a periodicity in performance is analysed by developing a generalised model of the shock structure and underlying flow physics. Superior performance was typically observed at smaller angles of incidence resulting from greater pressure differences between the upper and lower surfaces of the grid fin cell. The effects of 3D modifications, particularly brazing at the joints, resulted in up to a 33% performance degradation, prompting suggestions for alternative approaches. Lastly, preliminary heating estimates from theory, empirical methods and CFD were compared which can additionally inform future material choices and producibility decisions.
i Acknowledgements
Dr Paul Bruce has not only provided an outstanding level of project supervision and guidance but also an impeccable standard of teaching during my time at Imperial; his lectures have been some of the most enjoyable, clearly delivered and of course engaging topics that I have studied. I am grateful to Dr Bruce for firstly taking on this project, and then sharing his incredible knowledge, passion and invaluable advice throughout, without which completing this project would never have been possible.
I would also like to express a huge thank you to Dr David Hayes, Jeff Thornton, Jeremy Murgatroyd and the rest of the team at MBDA who have allowed me the unique opportunity to complete this project and provided extremely helpful feedback and guidance.
Thanks also to Prof. Denis Doorly for providing an amazing level of support through personal tutorial sessions over the past four years at Imperial which have certainly helped me to overcome the challenges of this degree.
ii Contents
List of Figures ...... v
List of Tables ...... vi
Nomenclature ...... vii
1 Introduction ...... 1 1.1 Motivation...... 1
2 Background and Literature Review ...... 2 2.1 Missile Configurations ...... 2 2.2 Applications of Grid Fins...... 3 2.2.1 Vympel R-77 ...... 3 2.2.2 SpaceX Falcon 9 ...... 4 2.2.3 Orion Launch Abort Vehicle...... 4 2.2.4 Producibility of Existing Designs...... 5 2.3 Drag Reduction through Geometry Modification...... 6 2.4 Leading and Trailing Edge Geometry...... 9 2.5 Full Body CFD Studies...... 9 2.6 Unit Grid Fin Method...... 10 2.7 Theoretical Methods ...... 11 2.8 Flowfield around Grid Fins...... 11
3 Computational Approach ...... 12 3.1 Key Objectives ...... 12 3.2 Flight Conditions ...... 12 3.3 Configuration and Vympel R-77 Geometry...... 12 3.4 Three Stage Approach...... 13 3.4.1 Stage 1 - Single Flat Plate...... 13 3.4.2 Stage 2 - Two Flat Plates...... 14 3.4.3 Stage 3 - Unit Grid Fin (3D) ...... 14 3.4.4 Benefits of the Three Stage Approach ...... 15
4 Single Flat Plate ...... 16 4.1 Simulation Setup ...... 16 4.2 Mesh Convergence ...... 17 4.3 Aerodynamic Forces ...... 17 4.4 Validating Results...... 18 4.4.1 Identifying the Sources of Drag ...... 19 4.5 Higher Angles of Incidence...... 20 4.6 Effect of Thickness to Chord Ratio...... 21 4.7 Effect of Removing Leading or Trailing Edge Taper...... 21 4.8 Summary of Findings from a Single Flat Plate...... 22
5 Two Flat Plates ...... 23 5.1 Simulation Setup ...... 23 5.2 Small Angles of Incidence ...... 23 5.2.1 Detailed Analysis of Varying Chord to Span Ratio...... 24 5.3 Higher Angles of Incidence...... 27 iii 5.4 Mach Number Variation...... 28 5.5 Summary and Limitations of the Two Flat Plates Analysis...... 29
6 Unit Grid Fin ...... 30 6.1 Simulation Setup ...... 30 6.2 Mesh Convergence ...... 31 6.3 Aerodynamic Forces ...... 31 6.3.1 Validation...... 33 6.3.2 Transonic Performance...... 34 6.3.3 Vortical Structures ...... 35 6.4 Effects of 3D Geometry Modifications...... 36 6.4.1 Aerodynamic Performance...... 36
7 Heating Estimates ...... 38 7.1 Theoretical and Empirical Methods...... 38 7.2 Results...... 38
8 Conclusion ...... 39 8.1 Summary ...... 39 8.2 Producibility and Further Work...... 40
References ...... 41
Appendices ...... 46
A Vympel R-77 Missile ...... 46
B Grid Fin Drawing (Vympel R-77) ...... 47
C Drag Reduction through Geometry Modification ...... 48
D Flight Conditions and Prism Layer Parameters ...... 49
E Shock Expansion Theory for a Single Flat Plate ...... 50
F STAR-CCM+ Settings for CFD Simulations ...... 52 F.1 Single Flat Plate...... 52 F.2 Two Flat Plates ...... 53 F.3 Unit Grid Fin ...... 54
G Wall Coordinates ...... 55 G.1 Single Flat Plate...... 55 G.2 Two Flat Plates ...... 56 G.3 Unit Grid Fin ...... 57
H UGF 3D Geometry Modifications ...... 58
I Calculation of Heating Estimates ...... 60
iv List of Figures
1 A Russian R-77 missile with tail mounted grid fins [1]...... 1 2 Canard, wing, tail and split-canard control missile configurations, each shown with grid fins.2 3 Falcon 9 with grid fins mounted at the top of the first stage [22]...... 4 4 Four, body folding titanium grid fins on the first stage of the SpaceX Falcon 9 launch vehicle [29]...... 5 5 Swept grid fins at the base of a scale model of the NASA Orion LAV [28]...... 5 6 Local sweep applied to individual grid fin cell members [31]...... 6 7 A folded sheet metal and mandrel curing approach to producing grid fins as given in [32].. 6 8 Flowfield structure around a single grid fin cell at different speeds [37]...... 7 9 A schematic of the flow approaching a blunt, unswept grid fin cell and a 30° sharp swept cell [41]...... 8 10 Section and edge variations in [48]...... 9 11 Flow structure and SBLI between a grid fin and the missile body at M = 3, α = 0° [40]. . 10 12 Surface pressure distribution showing the grid fin wake interacting with the aft missile body at M = 2.5, α = 5° [4]...... 10 13 Representation of a unit grid fin within a lattice framework [53]...... 10 14 A simplified drawing of the R-77 grid fin geometry showing a top and right view with key geometrical parameters labelled. A full drawing is provided in AppendixB...... 13 15 Stage 1 analysis...... 13 16 Stage 2 analysis...... 13 17 Stage 3 analysis...... 13 18 An illustration of the different LE and TE taper geometries used within the stage 1 analysis. All corners have a radius of 0.01t...... 14 19 An illustration of the UGF model showing the ‘tunnel’ domain and the pairs of periodic wall boundaries. Variable have the values presented in Table3...... 15 20 The mesh around a sharp single plate configuration...... 16 L 21 Plots of CL, CD and D convergence against varying number of cells within the domain. . . 17 22 Lift and drag coefficient against α for all four edge taper geometries. Linear theory (and skin friction corrected linear theory) also plotted for comparison...... 18 23 Shock expansion theory around the single flat plate (sharp LE and TE), shown at an α > 0. 18 24 Edge geometries used for verifying drag due to thickness. Dimensions in mm, not to scale. 19 25 CD against α for all taper geometries and the sharpened edge cases. Computed from CFD simulations and skin friction corrected linear theory...... 20 26 CL and CD against α for all four edge taper geometries, up to α = 20°. Linear theory (and skin friction corrected linear theory) also plotted for comparison at small angles...... 20 L t 27 D against varying c ...... 21 L 28 Plots of CL, CD and D against α with the effect of removing LE and/or TE taper...... 21 29 The simulation setup used for CFD analysis of the two flat plates...... 23 L c 30 Plots of CL, CD and D against varying s for small angles of incidence (α = 0, 3, 5°). . . . 24 31 An illustration of the flow structure between (a forward section of) two sharp edged flat plates c at an intermediate s (where flow choking has not occured) and α = 0°...... 25 32 Absolute pressure fields from CFD and corresponding CP distributions on the internal surfaces c of the two plates, shown for s = 1, 2, 4.5, 8, 30. All shown for M = 3, α = 5° ...... 26 L c 33 Plots of CL, CD and D against varying s for higher angles of incidence (α = 0, 3, 5, 10, 20°). 28 c 34 Absolute pressure for s = 2 at M = 3, α = 20°...... 28 L 35 Plots of CL, CD and D against varying Mach number (0.6 ≤ M ≤ 3) for a single flat plate and two flat plates in 2D...... 29
v 36 Volume mesh inside the tunnel domain, with a volume of refinement around the UGF. . . . 30 37 UGF surface mesh; oriented with flow left to right...... 30 L 38 Plots of CL, CD and D convergence against varying number of volume mesh cells within the ‘tunnel’ domain around the UGF. Simulated at M = 3, α = 10°...... 31 L 39 Plots of CL, CD and D against α for 0° ≤ α ≤ 20° from the 3D UGF simulations...... 32 40 Mach number contour plots for different free-stream Mach numbers at α = 5°. Shown in a lengthwise plane within the ‘tunnel’, 5mm offset from the centre of UGF with free-stream flow is left to right...... 33 41 Absolute pressure contours for M = 1.5, 2, 3 shown on a widthwise plane (s × s) centred within the UGF and tunnel domain. All at α = 5°...... 33 42 Validation of results against those presented by Miller et al. [48]...... 34 L 43 D against Mach number for the high subsonic, transonic and low supersonic regime. . . . 34 44 Mach number contours around the horizontal plate of the UGF at M∞ = 0.6, 0.7, 0.8, 1. . 35 45 Skin friction coefficient along the vertical UGF plate. Take at 5mm above the centre for M = 3, α = 5°...... 35 46 τw vectors at M = 3, α = 5°...... 36 47 Vortices from the Q-criterion at M = 3, α = 5°...... 36 48 Geometrical modifications applied to the TE of the UGF. Top view of TE shown...... 36 L 49 Percentage changes in CL, CD and D from an unmodified UGF at M = 3, α = 3°...... 37 L 50 Percentage changes in CL, CD and D from unmodified UGFs at M = 1.8, 2 and 3; all at α = 3°...... 37 51 Temperature distribution from CFD shown for a forward section of the horizontal UGF plate. 38 52 q˙w along the plate from using theoretical, CFD and empirical (graphical) methods...... 38 53 Alternative node locations for a UGF study [53]...... 40 A.1 A schematic diagram showing the layout of the Russian Vympel R-77 missile with tail mounted grid fins [69]. Parts are labelled in Table A.1...... 46 A.2 Close-up images of the R-77 grid fins [69]...... 46 B.1 An engineering drawing of the R-77 grid fin...... 47 C.1 A synopsis of the geometrical variations covered in several literature studies [47]...... 48 E.1 Identified oblique shocks and expansion fans used for application of the shock expansion theory to the single flat plate configuration at α = 3°...... 50 G.1 Histogram plots of wall y+ for different edge geometries at M = 3, α = 3°, over the whole single flat plate surface...... 55 + c G.2 Histogram plots of wall y at M = 3, α = 3° for the sharp edged two plates with s = 1, 3, 8, 20...... 56 G.3 Histogram plots of wall y+ at α = 5° and M = 0.9, 1.5, 2, 3 for the unit grid fin 3D simulation. 57 H.1 An exhaustive illustration of all geometrical modifications applied to the TE of the UGF. All views are isometric taken from the TE...... 58 ∗ I.1 τw, Cf and St found from theory, CFD and through the empirical method given in [68]. . 62 List of Tables
1 A summary of the identified aerodynamic flow features through grid fins at different regimes; based on [37]...... 11 2 Conditions calculated using the ISA at an altitude of 10km above mean sea level...... 12 3 Values of the geometrical parameters shown in Fig 14...... 13 5 4 Comparison of CL for the sharp edged plate at M = 3, Re = 9.71 × 10 from CFD, linear theory and shock expansion theory...... 19 A.1 Identified components presented in Figure A.1...... 46 vi Nomenclature s Span between grid fin cells, m
2 M Mach number Sref Reference area, m
2 α Angle of incidence, ° Sw Wetted area, m
∆s First prism layer height St Stanton number
δ Boundary layer thickness, m T Temperature, K
2 q˙w Heat transfer rate, kW/m t Thickness of grid fin web element, m c T ∗ Reference temperature, K s Chord to span ratio L uτ Friction velocity, m/s D Lift to drag ratio t V c Thickness to chord ratio ∞ Free stream velocity, m/s
γ Specific heat ratio, γ = 1.4 for air xt Leading or trailing edge length, m
ν Kinematic viscosity, m2/s y+ Wall distance coordinate
3 ρ∞ Free stream density, kg/m ALM Additive Layer Manufacture
τw Wall shear stress, Pa CFD Computational Fluid Dynamics a Speed of sound, m/s CNC Computer Numerical Control c Chord length, m GF Grid fin
CD0 Zero lift drag coefficient ISA International Standard Atmosphere
CDi Induced drag coefficient LAV Launch Abort Vehicle
CD (Total) Drag coefficient LE Leading edge
Cf Skin friction coefficient RANS Reynolds-Averaged Navier Stokes
CL (Total) Lift coefficient SBLI Shock Boundary Layer Interaction
CP Pressure coefficient TE Trailing edge n Number of total prism layers UGF Unit grid fin
P Pressure, Pa USAF United States Air Force
Re Reynolds number VOR Volume of Refinement
vii 1 Introduction
Grid fins (also commonly referred to as lattice fins) are an alternative lift generation and control device for missiles and spacecraft as compared with traditional planar fins. Grid fins typically consist of an outer frame with an internal grid structure of thin, small chord sections as shown in Fig.1, aligned perpendicular to the oncoming flow.
Grid fins were commonly seen on several intermediate range missiles and rockets produced within the Soviet Union dur- ing the 1970s such as the SS-12, SS-20 and Soyuz N1 launch vehicle intended for the Soviet moon program. Driven by modern day requirements for missiles to have compact overall packaging for use in internal weapons bays of next gener- ation aircraft, there has been a renewed interest in grid fin design and develop- ment. More recently, notable uses of grid fins can be found in the American GBU-43/B (MOAB) missile and the Rus- Figure 1: A Russian R-77 missile with tail mounted grid fins [1]. sian Vympel R-77, which is a supersonic ramjet powered air-to-air missile. SpaceX has also had in interest in the use of grid fins for their hypersonic reusable launch vehicles such as the Falcon 9 booster, configured to have four tail mounted grid fins to control roll, pitch and yaw movement [2] and having the ability to be stowed when not in use. 1.1 Motivation The unique aerodynamic and structural characteristics of grid fin architectures bring several advantages over conventional planar fins. A near zero hinge moment is observed allowing downsizing of actuator components giving weight reductions and more compact packaging; a smaller centre of pressure variation over a wide Mach number range; a high strength to weight ratio and increased stall angle of attack [3–5].
Grid fin applications for control and stability during atmospheric re-entry such as those seen on the SpaceX Falcon 9, have primarily focused on material selection and structural design to ensure the grid fin would be capable of withstanding the thermal environment during re-entry and would be able to effectively translate control inputs to stabilise the vehicle. However, a major drawback of grid fins as compared with planar fins of similar lift characteristics, is the higher drag values; often up to four times that of a planar fin, especially at transonic and low supersonic speeds [4–6]. Minimising this drag has not been a key performance driver for applications such as SpaceX whereas, when designing a grid fin for a high speed missile, this would be crucial to maximise range and provide stability at supersonic speeds.
Given that minimising drag is a key performance driver for such missile applications, an ideal grid fin design would have infinitely sharp leading and trailing edges as well as the thinnest possible section [7]. However, in practice, manufacture of such a design would be impossible and so, this project investigates several design attributes and tradeoffs such as the variation of geometry, which would be valuable to the designer in order to balance the aerodynamic performance with producibility. Aerodynamic analysis conducted in this project will allow the best compromise to be made depending on the manufacturing methods, materials and processes available to achieve the desired performance and hence, assess the relative importance of specific geometry and aerodynamic variations in producing the optimal grid fin. 1 2 Background and Literature Review
Given the typical context of grid fin use in missiles and defence applications, especially within the Soviet Union and United States Air Force (USAF), the availability of model-specific technical and performance data within the public domain is limited. However, more recent commercial applications such as in the SpaceX Falcon 9 allow for investigation into grid fin manufacture methods, geometry variations and possible performance drivers for the design. In addition to this, generic grid fin performance has been explored through various experimental and CFD studies allowing for comparison and validation against results produced in this project. This also allows more fundamental aerodynamic theory to be applied and thus, provide more readily available methods for evaluating the performance and producibility tradeoffs in grid fin designs.
2.1 Missile Configurations Similar to aircraft, several variations of missile configurations and layouts exist. These variations are primar- ily driven by the performance, packaging and thermal requirements of the missile being designed. The positioning of the fins along the missile body and their function is commonly varied and is an important consideration regardless of the type of fin used.
By varying fin position, size and function, a limitless number of designs are possible even within a specified Mach number or flight condition as presented in [8]. However, the most commonly seen layouts are wing control, canard control or tail control configurations [9–11]. Within each, the fin actuation is usually combined to give three degrees of freedom for the missile; roll, pitch and yaw (as well as in-flight longitudinal and lateral stability). The three common configurations, along with a novel split-canard configuration, are illustrated in Fig2.
Active grid fin(s)
(b) Wing control (a) Canard control
(c) Tail control (d) Split-canard control
Figure 2: Canard, wing, tail and split-canard control missile configurations, each shown with grid fins.
As discussed in-depth in [9, 12, 13], each of the configurations in Fig2 has advantages and drawbacks. Canard control which consists of active fins located on the aft section typically offers better manoeuvrability at small α but experiences large flow separation as α is increased to higher angles. Canard mounted grid fins would still require large tail fins (likely planar) to overcome the destabilising effect. Overall, the canard configuration is best suited to smaller missiles which do not require roll stabilisation [12] and will not experience large deflections of the grid fins. It is common to design canard controlled missiles with no roll control since roll stabilisation on planar tail fins adds complexity and cost [14]. An alternative is to place grid fins at the tail along with a planar finned canard as is investigated in [14], showing an improvement in roll effectiveness at supersonic speeds (M = 1.5, 3).
2 Wing control involves a large actuated fin mounted close to the center of gravity with fixed stabilising fins at the tail and aft sections. Despite being an early design, wing control still benefits from one of the fastest response characteristics as compared with canard or tail control [12], aiding seeker performance to minimise tracking error and be highly manoeuvrable [13]. A major disadvantage is packaging of the missile due to the large dimensions of the fin though use of a grid fin, as shown in Fig 2b, would help to mitigate against this by stowing against the body. The interaction of vortices shed from a grid fin with the tail mounted fins also needs to be considered carefully and often requires iterative optimisation of the fin location [12].
Tail control is the most commonly used configuration of modern missiles and benefits from good manoeuv- rability even at high α due to the large moment arm between the centre of gravity and the tail mounted fins. Due to tail deflection being in the opposite direction to the missile angle of attack, slower response is seen as compared with other configurations since the initial lift force is in a direction opposite to the desired one [12]. A common challenge with tail control layouts is the limited packaging space for control mechanisms and hence, employing grid fins rather than planar fins would mitigate against this since grid fins typically produce smaller hinge moments as compared with planar fins [3,5]. As shown in Fig1, the Vympel R-77 / AA-12 along with most other grid finned missiles of the past have utilised tail mounted grid fin control configurations. As a result of this, many existing CFD studies on grid fin performance, as subsequently discussed, assume a tail grid fin configuration.
In addition to these three commonly discussed configurations, many novel configurations also exist as given in [8]. One of these is a split-canard control layout as shown in Fig 2d. Such a configuration, albeit with planar fins, has been used on missiles such as the Kegler AS-12, Vympel R-73, Vympel R-60 and R550 Magic. A split canard layout overcomes some of the challenges of a single canard by generating vortices to accelerate the flow to the second canard and delay the onset of separation, enabling a higher α limit [13]. The downwash produced by the first canard postpones the leading-edge vortex breakdown to higher angles of attack [15]. Such a design may be of particular interest for a supersonic missile with grid fins in place of a second canard, perpendicular to the oncoming flow; grid fins perform best in high supersonic regimes rather than low supersonic or transonic [6] and hence this additional acceleration from the first canard would have potential for improving performance.
2.2 Applications of Grid Fins In order to appreciate the design considerations that have been made for grid fins and understand producibility motivations, it is useful to focus on specific applications of grid fins for missiles and spacecraft which have already been designed, manufactured, tested and are fully operational rather than purely conceptual studies. The grid fins used on the Russian R-77 Vympel missile are useful to analyse as they have been designed for operation in similar thermal and atmospheric conditions to those presented as requirements in [7]. Also, the application of grid fins to the SpaceX Falcon 9 booster provide a useful insight into how producibility decisions are altered for a commercial space application. Nevertheless, given the specialised defence and space related applications, material properties, specifications and design drawings are not readily available in the public domain. Therefore, model-specific details presented in this project have been cited where appropriate and are mostly a combination of technical specifications presented in various published sources.
2.2.1 Vympel R-77 The Vympel R-77 missile (also known as the AA-12 Adder) is designed as a medium range air-to-air missile, suitable for launch from under wing platforms such as that on the Russian MiG-29 aircraft [16]. The typical flight regime for the R-77 is supersonic, between Mach 2 and Mach 4, with up to a 40° angle of attack and a range of up to 70km [17]. As shown in Fig1, the cross (‘ ×’) configuration of the grid fins allows for roll, pitch and yaw motions with each fin being individually retractable by a motorised hinge [17]. The ability to retract 3 against the body of the missile is seen on several of the Soviet missiles presented earlier and allows for ease of transportation and storage; the fins are typically in the extended position when the missile is suspended from the aircraft wing as they form the primary control surfaces and hence are required at all phases of flight.
The grid fin design employed on the R-77 is typical of a generic design such that it consists of a straight edged frame, with a regularly spaced internal lattice structure. The overall chord length of the grid fin is constant throughout and each cell within the lattice has equal spacing. The cells are divided such that each grid division is diagonal across the rectangular frame, with a tapering seen towards the base where the span reduces and frame thickens for structural rigidity. In this diagonal orientation, the grid fin is composed of 2 cells (width) by 4 cells (height). The base of the grid fin is attached to the body of the missile using a single hinge, with a ramped fairing at the base covering the exposed hinge components [18]. A detailed labelled diagram of the R-77 layout as well as images of the grid fin are given in AppendixA. The geometry of the R-77 grid fins and its particular application to this project [7] is discussed in greater detail in Section 3.3.
Due to the generic nature of the R-77 grid fins, with no cell or chord-wise variations in geometry, it is often used as a basis for CFD studies and literature concerning grid fin aerodynamic analysis; The R-77 / AA-12 missile is specifically used for investigating aerodynamic performance in [18–20] and is discussed in [21] within a review of comparing approaches to CFD studies on grid finned missiles. A scale model of the R-77 missile is used in [19] in order to compare theoretical predictions (based on shock expansion theory) with prior experimental data from wind tunnel tests for the normal and lateral forces along the grid fin, pitching moments and vortex interactions of the missile body with the grid fin cells. Variations of this configuration that include altering the cell size and grid fin shape are also presented for comparison in [19].
2.2.2 SpaceX Falcon 9 The integration of grid fins on the first stage of the SpaceX Falcon 9 reusable launch vehicle has the same purpose as on the R-77 missile, to provide a manoeuvrable control surface device [23]. However, the thermal and atmospheric environment that they have been designed for varies largely from the application to missile systems; SpaceX grid fins are designed for hypersonic flight speeds and for deployment during the re-entry phase [24] meaning that the fins must withstand much higher temperatures and multiple uses without need for replace- ment. The configuration employed on the Falcon 9 is shown in Fig 3 and Fig4 with four foldable heat-resistant grid fins mounted at the top of the first stage section [25, 26]. The folding design drives the requirement for the fins to have a sweep in the chord-wise direction allowing better integration around the Falcon 9 body and minimising drag (by minimising exposed frontal area) when folded. During re-entry, cold gas thrusters are used to correct the vehicle’s attitude Figure 3: Falcon 9 with grid fins before the fins are deployed; this is then followed by deployment of mounted at the top of the first stage the four composite landing legs as it approaches the touchdown point [22]. [24].
2.2.3 Orion Launch Abort Vehicle As presented in [27, 28] and shown in Fig5, grid fins were also designed for use on the NASA Orion Launch Abort Vehicle (LAV). An interesting aspect of this design is that the grid fins are optimised for small length : diameter ratios which is different to typical missile bodies. A wide Mach number range (0.5 < M < 2.5) is explored in [28] with a particular focus on transonic performance and the effect of sweep to improve pitch 4 stability of the LAV as compared to without grid fins. Both [27] and [28] present experimental results from wind tunnel tests (of a 6% scale model), while CFD results are also compared in [27] for a range of twelve different Mach numbers and for 0° < α < 15°. Although produciblity and integration with the LAV is not explored in depth, the results presented in [28] show the differences between a (‘+’) and (‘×’) configuration as well as geometrical sweep of the fins (over the whole Mach range) which would provide starting points when considering producibility and manufacture. Based upon the CFD data generated from [27] and the experimental data in [28], an aerodynamic database is now available to the designer of a generic missile, if meeting the length : diameter ratios of the LAV and at flight conditions compatible with those in [7].
Figure 4: Four, body folding titanium grid Figure 5: Swept grid fins at the base of a fins on the first stage of the SpaceX Falcon scale model of the NASA Orion LAV [28]. 9 launch vehicle [29].
2.2.4 Producibility of Existing Designs An important motivation for the use of grid fins on SpaceX Falcon 9 is the ability for them to be reused on multiple flights and hence be non-ablative. During atmospheric re-entry, the exterior surface of the Falcon 9 experiences temperatures up to 1850°C[30] and hence stagnation points on the grid fins would likely be much hotter. Therefore, a single piece cast and cut titanium alloy is used for each fin [26, 29]. The high strength : weight ratio, corrosion resistance and high melting point allow titanium to be the ideal choice for grid fins on space vehicles despite the relatively high cost. Typically, casting allows for benefits such as a consistent quality cross section, high dimensional accuracy and a superior surface finish even when thinner sections are used, ideal for optimising grid fin aerodynamic performance which is sensitive to surface geometry and dimensions. More recently, SpaceX has announced the move to welded steel grid fins on future projects such as the Starship Super Heavy. This decision is motivated by the fact that the Starship’s grid fins would be much larger (8-10 times the surface area) and hence the cost and feasibility of casting titanium at that scale is a challenge. The stainless steel alloy has been designed to have similar structural properties despite the reduced heat capacity [29]. The construction of a welded grid fin does however reduce the possibility of variations between cells as new tooling would be necessary (which is often at a high cost) and would ultimately produce a surface finish, particularly at joins, inferior to that of a casting process.
While limited information is available regarding the manufacturing process used on the Vympel R-77 grid fins, a common challenge (also a key objective of this project [7]) is the trade off between produciblity and aerody- namic performance. The desire for thin surface elements with perfectly sharp leading edges and the associated high manufacturing costs for a grid fin based on the R-77 is highlighted in [31]. Further, the possibility of using thin metal sheets within each cell is discarded as this would likely not withstand the thermal and mechanical loads present at supersonic flight speeds and would be difficult to produce accurate leading and trailing edges on.
5 Therefore, [31] explores various methods of reducing wave drag (which would be inherent of more typical production methods) such as through local sweep of fin cells, altering intersection geometries and varying sharpness of the leading edges based on available man- ufacturing techniques; results from experimental tests at 4 ≤ M ≤ 6 are presented for three different geo- Figure 6: Local sweep applied to individual grid metries. Most notably, up to a 20% improvement in lift fin cell members [31]. to drag ratio and 38% in zero lift drag is reported for local sweep applied to the grid fin [31] (shown in Fig6) which informs the impact of producibility on performance, depending on the chosen process.
Modern day capabilities with composite materials and additive layer manufacturing (ALM) means that new methods of producing grid fins are becoming possible and hence, would be able to better meet the geometrical demands to achieve optimum aerodynamic performance. Although patented, ideas presented in [32] and shown in Fig7 include a metallic origami-like structure for the internal framework produced from thin metal sheets cut by laser or CNC water jet methods. Also presented is a method to produce a composite grid fin from a mandrel based curing process giving elongated extrusions from which the final fin is sliced. An alternative, economical moulding process to re- duce cost of manufacture as compared with water jet cutting or machining from solid metal billet is discussed in [33] which still allows for complex geo- metries and the use of composites to achieve the optimum performing design. ALM, such as by the use of metal 3D printing is also becoming a popular option to produce the intricate geometries of grid fins [34]. ALM also has the key benefits of reduced need for quality control, superior surface finishes, being highly accurate and being able to alter the infill density to reduce overall weight of the fin, fur- ther minimising hinge moments. ALM for grid fins may also have benefits in improving the heat transfer Figure 7: A folded sheet metal and mandrel curing within the component as compared to brazed joints approach to producing grid fins as given in [32]. which is explored through simulation in [35].
2.3 Drag Reduction through Geometry Modification Based on linear aerodynamic theory, drag forces acting upon the complete missile can be decomposed into zero lift drag and induced drag [36].
CD = CD0 + CDi (1) Induced drag is often the largest component and is affected by design characteristics, such as the variation of leading and trailing edge geometries or the sectional dimensions of the finite lifting surface. The zero lift drag can be further separated as presented in [11]; the empirical relationship given in Eq2 is used for zero lift drag computation to account for interference effects of missile components and the variation in skin roughness over the body.
CD0 = 1.25 CD0 + CD0 + CD0 + CD0 (2) nose body fins tail
6 The contribution of zero lift drag from the (grid) fins can then be further separated into wave drag and skin friction drag.
CD0 = CDskin friction + CDwave (3) fins fins fins Many empirical relationships exist for predicting skin friction drag at supersonic speeds based on length scaled Re. Assuming turbulent flow, an (incompressible) skin friction coefficient relationship given in [11] is −2.58 CDskin friction = 0.455 (log10 Re) , to which a compressiblity correction and reference area scaling is then applied. Most importantly, [11] gives the relationships in Eq4 for the wave drag arising from a leading edge for a double wedge supersonic aerofoil (assuming a perfectly sharp edge), which acts as the grid fin section.
B t 2 S C = w Dwave (4) fins β c Sref √ B = c/xt β = M 2 − 1 where 1−xt/c ,
Several studies have been performed on the effects of varying grid fin cell geometry in order to min- imise the total drag and make it more comparable to that of planar fins. These have not only been limited to variations in leading and trailing edge sharpness but have also focused on the local and global sweep of lattice cells and variations of cell size, which would ultimately drive producibility decisions. A combination of experimental and CFD results for subsonic [3, 38–40], transonic [3, 40–46] and supersonic regimes [3–5, 40–43, 46] have been conducted. Typically, grid fins have the poorest performance at transonic flight speeds due to a large standoff bow shock occurring from chok- ing of the individual grid fin cells; the flow field at subsonic, transonic (choked), low supersonic and Figure 8: Flowfield structure around a single grid fin high supersonic speeds for a single cell is shown cell at different speeds [37]. in Fig8.
Similar to the application of SpaceX in Fig4, adding sweep to the grid fin aids packaging and minimises drag. By assuming a sharp leading edge swept fin (30° sweep) and a blunt edged unswept fin, [41] presents the experimental results from wind tunnel testing at transonic and supersonic speeds (0.75 ≤ M ≤ 1.7). As shown in Fig9, a standoff bow shock is observed for both, with the swept configuration reducing the strength and causing two, offset shocks rather than one; the overall CD saw a reduction of up to 30% by adding sweep. Similar studies by Zeng [43] and Wang et al. [42] use a CFD approach to investigate a sharp edge sweep for 0.905 ≤ M ≤ 2 with similar results showing a CD reduction, especially at transonic choking cases. Out of the eight geometries presented in [42], the largest CD reduction of 54.8% against a blunt unswept fin is seen at M = 1.7. However, [46] shows the results of the above but with the addition of accounting for the flow interaction with the full missile body. The full missile body is simulated along with the sweep effects of the grid fin at M = 0.9, 1.2, 2 and a smaller overall drag reduction is observed (up to 13%), indicating that flow interaction with the body is significant. Interactions of the boundary layer on the body with the incident shock waves from the grid fins (SBLI) likely cause a notable increase in drag which is not explored in [42].
7 Figure 9: A schematic of the flow approaching a blunt, unswept grid fin cell and a 30° sharp swept cell [41].
Existing studies [23, 38, 39, 44] on variation of cell geometry are mostly in the subsonic or transonic regime. However, [3] explores cell geometry variation up to M = 3.5 which is more suitable for the requirements of the project [7]. A more generic, octagonal shaped fin configuration is employed in [3, 39] whereas a shape more similar to the R-77 grid fin is used in [23, 44] albeit with scaling and simplifications applied where experimental tunnel testing has occured. As presented in [39], using smaller cells within the grid fin frame (i.e more cells and more surfaces), leads to an increase in normal and axial forces, however, the increase in drag coefficient is better explored in [3]. At M = 3, Washington et al. [3] shows that a coarse-celled grid fin has a CD twice that of a planar fin whereas a fine cell density causes a CD four times that of a planar fin. This is particularly significant as despite resulting in increased normal and axial forces (hence being more efficient as control devices), a trade-off with increased drag is observed as the number of cells and hence number of surfaces inside the grid fin frame is increased. A ‘transonic bucket’ is also reported in [3] which correlates to the drag increase at transonic regimes and a corresponding drop in normal force produced by the grid fin. This transonic regime (with choking in [3] occurring at M ≈ 0.75) is therefore of great concern when designing grid fins at high speeds since they will often transition through these transonic Mach numbers. The critical Mach number (Mcr) for the onset of transonic choking with varying cell geometries, web thicknesses and area ratios is reported in [44] and is in the range of 0.65 < M < 0.87; a more coarse fin or increasing thickness of the section results in increase of the observed Mcr. Nevertheless, the effect of changing α on Mcr L is not explored in [44]. Despite varying the cell geometry, the reported CL, CD and D shows similar trends in all cases in [23, 38] at the subsonic speeds tested. It is observed that CL shows a linear behaviour with α, up to a maximum α after which it begins to fall; CD shows a rapid increase at higher α (above ≈ 10°) and hence L there is a peak observed in the D plots indicating an optimum α for the given grid fin geometry and flight conditions. The CFD analysis in [23] also explores the variation in cell geometry through approximating each cell as a NACA 0012 symmetric aerofoil as well as then as a rectangular flat plate. It is important to note however, the analogy to a flat plate is more appropriate to a supersonic high speed simulation than is the (subsonic optimised) NACA aerofoil due to it is thickness and large nose radius. Pressure coefficient plots in [23] show good agreement in trends by using a blunt flat plate aerofoil as an approximation to the real grid fin section and this is most likely due to the lack of an oblique shock from the leading edge (at the subsonic speeds tested) meaning that a leading or trailing edge sharpness is not necessary for adequate performance.
Altering geometric parameters (frame shape, frame thickness, web thickness, cell size) provides a reliable and consistent drag reduction technique and enables grid fins to exceed planar fin performance [47]; a summary of many of the studies discussed above is provided in [47]. An image showing geometrical variations used within these is given in AppendixC for a range of missile and spacecraft based applications.
8 2.4 Leading and Trailing Edge Geometry Miller et al. [48] explore several different combinations of grid fin sections, leading edges and trailing edges, shown in Fig. 10 as techniques of drag reduction. These are investigated through wind tunnel tests for 0.5 ≤ M ≤ 2.5 and −8° ≤ α ≤ 20° with a length averaged Re between 5.35×10−6/ft and 8.05×10−6/ft depending on the tunnel speed. Chord lengths for the grid fin sections are kept constant whilst the leading and trailing edge angles are varied between 17° and 90° (blunt edge) as well as the length of the edge along the chord. The chord sections are investigated independently as well with a grid fin frame attached; Figure 8 in [48] shows
that the highest CD0 is observed for a sharp edge with a frame, followed by a blunt edge without a frame. The lowest drag is seen for a half diamond wedge aerofoil. This indicates that the effect of the frame is also an Figure 10: Section and edge variations in [48]. key consideration for grid fin drag and hence may also benefit from a producibility analysis to minimise the thickness and minimise drag.
2.5 Full Body CFD Studies When considering the aerodynamic performance of grid fins, especially at supersonic speeds, SBLIs are present along each grid fin cell as well as along the missile body. Predictions for the drag reduction perform- ance of geometrical modifications are often overestimated when this interaction is not considered. Burkhalter et al. [49] presents a non-linear theoretical method for evaluating grid fin performance and many of the methods used, including an extension of the vortex lattice solution and prediction of forces accounting for the contribution from the missile body as well as the grid fin. The interaction of the missile body is also considered in [4, 18] such as through the distribution of surface pressure contours at the junction of the grid fin attachment and the skin on the missile body. Additionally, body mounted ramp fairings are investigated in [18] for up to Mach 2, as a successful method of reducing body interaction and blockage effects. Furthermore, the SBLIs inside the cells were more evident as the thickness of the web was increased. It is worth noting that the L3 configuration of the grid fin in [18] is geometrically similar to that of the Vympel R-77. Both a cruciform type (‘+’) and cross type (‘×’) configuration of grid fins is investigated in [4] through full body CFD sim- ulations at M = 2.5. The SBLI between the missile body and a mounted grid fin at M = 3 is shown in Fig. 11
Some of the first viscous full body CFD simulations on grid finned missiles were conducted by DeSpirito et al. [50, 51]. Both of these studies, at M = 2.5, showed good agreement with wind tunnel experiment and employed symmetry planes on a cruciform (‘+’) configuration grid finned tail control missile. Both also clearly show the complex shock structure occurring through the grid fin cells via the pressure coefficient contours. The ogival shaped nosecone also produces oblique shocks and the interaction of these with the fins can be seen in [51].
As seen in Fig. 11, the wake region of the grid fin can produce complex flow features including strong vortices and a recirculating region which eventually interacts with the body of the missile. CFD simulations of this wake interaction at M = 1.5, 2.5, 3.0 are included in [14] and [52]. As discussed in [14], the strength of the vortical structures in the wake grows rapidly, especially when the grid fins are deflected, causing a large increase in drag. Therefore, it may be argued that aside from optimising the grid fin geometry and balancing 9 produciblity to reduce drag, the grid fins should be placed as far aft as possible to minimise the effect of these vortices on the overall drag force, especially when a ramped fairing is not present [18]. Fig. 12 shows the interaction of the grid fin wake with the aft missile body.
web bow shock web wake vortices
flow direction base recirculating region recompression FIN near wake recompression shock shock boundary layer
BODY separation recirculation area reattachment
Figure 11: Flow structure and SBLI between a grid Figure 12: Surface pressure distribution showing fin and the missile body at M = 3, α = 0° [40]. the grid fin wake interacting with the aft missile body at M = 2.5, α = 5° [4].
2.6 Unit Grid Fin Method Whilst a full body simulation is valuable in understanding the interaction of flow features, especially in the boundary layer and the wake, it is often too computationally expensive especially at a preliminary design stage [53]. Additionally, the corresponding mesh required to be able to fully capture the flow across the whole missile requires several iterations to optimise. As proposed by Dikbas et al. [53], the unit grid fin (UGF) approach offers a suitable alternative to analysing and predicting the aerodynamic forces around grid fins by simplifying the problem to only a portion of the fin. The validation of this approach is presented in [53] against a cru- ciform (‘+’) configuration before the idea is extended to a planarform configuration at Mach numbers across the tran- sonic and supersonic regime ( M = 0.7, 1.2, 1.5). As found by Dikbas et al., the method is most accurate for supersonic regimes with reduced accuracy at transonic speeds due to the lack of ability to fully capture bow shock effects. The unit Figure 13: Representation of a unit grid fin grid fin approach is shown in Fig. 13 where a central region within a lattice framework [53]. of the grid fin lattice is isolated for the computation. Hence, the effect of the frame, connecting rods and body is not included so a correction can then be employed to calculate the effective angle of attack as given in Eq5. 2 r0 αeff = α 1 + (5) r1r2 where r0, r1, r2 are the radii to missile body surface, top of connecting rod and top of frame respectively. A comparison of Mach contours of a UGF against a full grid fin at M = 2.5 are shown in [53] and show good agreement of flow features allowing this to be a faster, more readily accessible alternative to full body CFD simulations. 10 2.7 Theoretical Methods Although the flow features of grid fins are extremely complex and sensitive to free-stream Mach number and conditions, theoretical prediction methods are available as a method of calculating global forces and moment coefficients by simplifying the grid fin. A modified Evvard’s theory for supersonic flight regimes is presented in [37]. This relies upon modifying a linear aerodynamic panel theory to account for grid fin frames and end plates. The grid fin element is separated into regions by Mach lines and then discretised into smaller rectangular elements such that the pressure distribution and hence normal and axial forces can be computed. While this theory is not able to account for complex geometries and complex flow features such as SBLIs, Kretzchmar et al. [37] found an acceptable comparison in the values calculated for horizontally placed fins with no deflection or angle of attack. Therefore, the theory has not been validated for non-zero α and it cannot fully capture flow physics inside each grid fin cell. Actuator disk theory has also been applied to grid fins by Reynier et al. [54]. Whilst this also is able to compute global forces and moments, even at small angles of attack, Reynier et al. concludes that the theory is limited as soon as vortical flow develops along the body and hence local Mach number variations occur. These theoretical methods are therefore unsuitable for reliable prediction of lift and drag forces as well as capturing the flow features around grid fins on a supersonic missile. CFD and experimental testing on scale models remain the most reliable methods of obtaining aerodynamic forces.
2.8 Flowfield around Grid Fins Based upon the literature reviewed in 2.3 and 2.5, the expected flow field around a typical grid fin can be summarised as given in Table1. It must be noted that the Mach numbers identified in Table1 are only an indication based on previous studies and hence would be highly dependant on the chosen grid fin geometry, thermal environment and atmospheric conditions of flight.
Table 1: A summary of the identified aerodynamic flow features through grid fins at different regimes; based on [37].
Mach number Flow features M<0.7 • No supersonic regions in local flow. Subsonic • Lift and drag show little sensitivity to geometry or cell pattern variation. 0.7
11 3 Computational Approach
3.1 Key Objectives Based upon the requirements and work packages given in [7], the literature reviewed in Chapter2 and the timeline for the project, the key objectives for aerodynamic and producibility analyses are summarised as follows.
• To understand and calculate aerodynamic forces and flow features around missile mounted grid fins at the specified flight conditions primarily through the use of CFD. • To synthesise the effects of grid fin geometry parameters (edge sharpness, span, chord, taper / sweep) on aerodynamic performance across a range of flight conditions. • To inform producibility decisions based upon the aerodynamic analysis that will ultimately make grid fins more competitive to traditional planar fins. 3.2 Flight Conditions The baseline conditions considered for a generic high speed missile were at an altitude of 10km above mean sea level with a cruising speed of M = 3 [7]. From this, the complete atmospheric and thermal environment was computed using the assumption of an International Standard Atmosphere (ISA) model which was a reasonable approximation to exact conditions given that the flight would still be within the troposphere. The conditions at M = 3 are given in Table2 and were calculated using an implementation of the ISA in MATLAB. The MATLAB code is provided for reference in AppendixD and has been used for all subsequent calculations of flight conditions.
Table 2: Conditions calculated using the ISA at an altitude of 10km above mean sea level.
Parameter Value Ambient temperature, T (K) 223.15 Speed of sound, a (m/s) 299.46 Pressure, P (Pa) 26436.30 Density, ρ (kg/m3) 0.4127
3.3 Configuration and Vympel R-77 Geometry As explored in Chapter2, several configurations for missile fin layouts exist depending on the application and performance requirements. Although novel concepts such as the use of grid fins on a split-canard layout have never previously been explored in literature and offer unique benefits to exploit the advantage of grid fins, a more conventional tail control model was ultimately chosen. The tail control configuration shown in Fig. 2c comprises of grid fins mounted aft of the missile body with additional planar fins mounted forward for stability. A tail control layout is most commonly observed in existing literature and hence would allow for a means of validating CFD results and making them reliable. Since the focus of the project is based upon a generic missile and that the key objectives relate to the grid fin performance rather than control or overall missile performance, a tail control layout, with a large number of experimental and CFD studies available for validation, was used. The baseline geometry for the grid fin was based on that of the Vympel R-77 grid fin [7] which also features grid fins mounted in the tail control configuration. Given the limited availability of model-specific engineering drawings to capture dimensions and detailed features, a number of sources were used to generate a compiled data-set of dimensions which would be most similar to that of the R-77. Initially, photographs of the R-77 missile were studied to gain an appreciation for the relative scale of the grid fin elements and cell layout. Following this it was noted that [19–21, 55] all presented dimensions for the R-77 12 grid fins (albeit some were scaled models). Nevertheless, slight differences in dimensions were seen and this is likely attributed to the authors also making best estimates for dimensions rather than having a verified source. Therefore, the dimensions from these were averaged to generate a final set of suitable geometrical dimensions. As shown in Fig. 14 the geometry can be summarised to have a rectangular frame, with a 4 × 2 cell layout and a tapered 2 cell base. The cells are square with a span s and material thickness between cells t; the overall chord length of the fin, c is also identified. Thes values of theset parameters are given in Table3. c
Table 3: Values of the geomet- rical parameters shown in Fig 14.
Geometrical Parameter (mm) Chord length, c 38.50 Span, s 45.00 Thickness, t 0.675 Figure 14: A simplified drawing of the R-77 grid fin geometry showing a top and right view with key geometrical parameters labelled. A full drawing is provided in AppendixB.
A full engineering drawing showing a complete set of dimensions for the R-77 grid fin is provided in Appendix B. 3.4 Three Stage Approach Although a single, full body or full fin CFD model is a potential method for obtaining the relevant aerodynamic forces and effects of geometry changes, it would lead to a high computational cost, increased dependence on the mesh quality and risk being too heavy a CFD exercise for the time available. Also, this would loose the importance of understanding the flow physics and produciblity trade-offs [7]. Therefore, an alternative three stage approach, as shown in Figs. 15, 16 and 17 was taken.
3D effects c t M M ∞ LE/TE ∞ s c M ∞
Figure 15: Stage 1 analysis. Figure 16: Stage 2 analysis. Figure 17: Stage 3 analysis.
3.4.1 Stage 1 - Single Flat Plate The first stage of the aerodynamic analysis involved a two dimensional model with a single flat plate placed inside of a circular domain of radius 10c. The single flat plate is representative of a single grid fin web element. This allowed a detailed analysis of the following effects: 13 1. The effect of varying the leading edge (LE) and trailing edge (TE) taper such that the effective sharpness of the edges was altered. The geometrical variations used for the LE and TE are given in Fig. 18.