design and testing of a large composite asymmetric payload fairing

American Institute of Aeronautics and Astronautics

1

Design and Testing of a Large Composite Asymmetric Payload Fairing

Tomoya Ochinero, Ph.D.,1 and Thomas Deiters, P.E.2 ATA Engineering, Inc., Los Angeles, CA, 90245

John Higgins, Ph.D., P.E.3 and Brandon Arritt4 AFRL Space Vehicles Directorate, Kirtland AFB, NM, 87117

and

Eric Blades, Ph.D.,5 and James Newman, III, Ph.D.6 Mississippi State University, Mississippi State, MS, 39762

A very large asymmetric composite payload fairing (PLF) was developed for launching payloads of unconventional geometry and size on the Atlas V Heavy Lift vehicle. Currently, no launch system exists that can accommodate these payloads without requiring a redesign of the launch vehicle and/or integration facility. The asymmetric design was tailored to accommodate very large payloads while maintaining structural requirements and control authority limits of the launch vehicle as it currently stands. An optimal design was achieved through the use of an innovative computational fluid dynamics (CFD)-based geometric optimization, composite structural tailoring, and novel manufacturing methods. The design was validated through correlation with subscale wind tunnel testing, and extremely close agreement between the analysis and test was achieved. The final design resulted in a composite sandwich structure that meets or exceeds strength, buckling, flutter, thermal, and acoustic requirements and does not require significant modifications to existing launch pad integration facilities. The geometry, methods, and processes demonstrated here have wider applicability to the whole range of launch vehicle sizes and can increase the payload capabilities of each by offering fairings which are tailored specifically to existing control capability and payloads of nonstandard geometry.

Nomenclature α = pitch angle β = yaw angle Cp = pressure coefficient s = search vector β = design variable

1 Project Engineer, Design and Analysis, 1960 East Grand Avenue, Suite 560, Member. 2 Manager, Design and Analysis, 11955 El Camino Real, Suite 200, San Diego, California. 3 Deputy Program Manager, Space Situational Awareness Integration Group, 3550 Aberdeen Ave, Member. 4 Acting Manager, Integrated Structural Systems, 3550 Aberdeen Ave, Senior Member. 5 Assistant Research Professor, Computational Simulation and Design, P.O. Box 9627, Senior Member. 6 Associate Professor, Department of Aerospace Engineering, 312C Walker Engineering Building, Member.


2

I. Introduction HE Air Force Research Laboratory (AFRL) began development of a larger fairing for the Minuteman III-based Minotaur series of launch vehicles in the year 2000, using composite technology and an advanced-grid system

concept to demonstrate alternate design and fabrication concepts in order to use composites in large space structures. The successful use of this fairing design for two launches provided encouragement for bolder applications seeking to not only build on strength, stiffness, and lightweight benefits of composites, but also to demonstrate the versatility of the material. AFRL wants to use composites in a broad array of large space structures. They began to develop appli-cations such as a payload attachment fitting (PAF) for Delta IV, multi-payload adapters for the Peacekeeper-based Minotaur launch vehicle series, and eventually even provided advice and support for composite options for future NASA manned spacecraft. All of these applications were designed and built and are in some stage of test and vali-dation. Major goals of AFRL in all of these studies were to demonstrate the great degree of conservatism imposed on composites based on failure theories and data from test coupons of small structural parts, to demonstrate the ver-satility of the material for unusual applications and geometries, and to help the space community overcome concerns for supposed vulnerabilities of large composite structures.

This paper focuses on the specific objective of optimizing the maximum size payload that could be launched by an Atlas V launch vehicle. In 2001, AFRL hosted a two-day Large Fairing Workshop supported by major U.S. launch vehicle providers, composites innovators, and related academic and government agencies. In this workshop, AFRL outlined a five-year strategy for sponsoring innovation in composites and related technologies needed to develop a fairing capable of supporting a nominal ten-meter-diameter optical system without folding the primary mirror. This was a substantial goal, as the current technology would not support lateral dimensions exceeding five meters. The workshop generated enthusiasm for the adventure, a broad sharing of new composite technology con-cepts in the launch vehicle community, and a sound basis for a requirements and planning document for such a large fairing. In the course of further consultations with this community, AFRL became aware of major restrictions on fairing diameters for these large launch vehicles based on infrastructure constraints of existing civil structures sup-porting manufacture, transportation, and, most importantly, spacecraft and fairing installation at the launch pad. Typically, there is an erection tower that secures operations about the spacecraft during launch preparation. For one of the major launch vehicle providers, the tower cross section is only about seven meters square and would be costly to replace. One notion that was considered to accommodate this constraint was to fit a payload (and fairing) with a maximum lateral diameter aligned along the diagonal of the tower cross section. This could only work geometrically for a non-cylindrical fairing cross section. If successful, such a design would also reduce launch drag through reduced frontal cross sectional area. The concept for an asymmetric fairing shape was born.

While the original intent was to showcase composite technology development, the new asymmetric fairing con-figuration introduced much larger questions about launch vehicle global stability, flutter, control of shock detach-ment/reattachment, and unstable downstream flow vorticity. A quick concept study was completed by Boeing, which provided some initial comfort that at least the Delta IV could potentially provide adequate control authority for maintaining launch stability. AFRL decided to address the need for more definitive design, analysis, and testing by awarding two Small Business Innovative Research (SBIR) contracts. Both contractors provided good initial suc-cess, resulting in the award of two Phase II SBIR contracts in 2003. During the transition from Phase I to Phase II for these contracts, a second Large Fairing Workshop was conducted by AFRL with similar industry and govern-ment agency participation. By the time of the award of Phase II, each contract was associated with a heavy lift launch vehicle provider. To maximize the utility of these awards, each contract was given a somewhat different payload and requirements package to ensure that truly alternative designs would emerge.

The rewards for the Air Force from this program have been substantial. They have been able to generate produc-tion techniques for extremely large composite space structures by combining unique design criteria and iterative optimization techniques for structural and CFD analysis, and they have multiple demonstrations of successful pur-pose-built fairings. Preliminary testing and analysis have also shown that using composite materials in the structural design of asymmetric payload fairings will not negatively impact their ability to meet strength, stiffness, and weight requirements. This has since been fully substantiated for a variety of composite structures tested by AFRL under qualification and failure load tests to be addressed in other papers. With the tools generated by this program, AFRL can expect to support a much wider range of space experiments, extracting optimal launch vehicle capacity as needed to support unique payloads. Payload developers can turn to AFRL to develop unique fairings that adapt to the constraints of the launch vehicle and the needs of the payload without introducing unwanted packaging con-straints on the payload. There is now a new way of looking at how payload accommodations will be provided in the future.

T


3

II. Asymmetric Payload Fairings Tailored asymmetric payload fairings, such as the one shown in Figure 1, are advantageous compared to tradi-

tional axisymmetric payload fairings because they can be designed so that they “shrink-wrap” around the geometry of the payload. This attribute is beneficial when launching relatively light but awkwardly shaped payloads such as unmanned aerial vehicles, parabolic antennas, etc. Currently, such payloads are launched using vehicles capable of lifting much heavier payloads than necessary because of the volume constraint. By using tailored asymmetric PLFs, such payloads may be launched using smaller vehicles, thereby dramatically reducing costs.

One of the difficulties of asymmetric PLFs is the increased aerodynamic loads. Figure 2 shows the field pressure contours on an axisymmetric PLF and an asymmetric PLF at Mach 1.1, 450 psf

dynamic pressure, 4° angle of attack, and at an altitude of 33,500 feet. The higher field pressures acting on an asymmetric PLF translate to higher interface loads into the launch vehicle. Table 1 shows the interface loads. A significant increase in side force and bending moment is observed for the asymmetric PLF. Therefore, the challenge is to design and optimize the PLF geometry to minimize these aerodynamic loads so that it does not exceed the limits and control authority of the launch vehicle.

Figure 1. Tailored asymmetric PLF can accommodate unique geometry payloads.


4

III. Geometric Optimization The objective was to design a large payload fairing for use on the Atlas V Heavy Lift vehicle that can accommo-

date the payload shown in Figure 3. The geometry of the PLF is of primary importance because it needs to be large enough to accommodate the payload and be aerodynamic enough so that the loads into the launch vehicle do not exceed those of a conventional axisymmetric 5.2-meter PLF. In addition, the design must also meet various thermal, structural, and vibrational requirements.

Computational fluid dynamics (CFD)-based geometric optimization was performed to evolve the design as

shown in Figure 4. The CFD analysis was performed using both the UNCLE [1] and U2NCLE [2] codes. UNCLE is a structured-grid, multi-block solver, and U2NCLE is an unstructured, vertex-centered solver. Both are parallel flow simulation codes that solve the unsteady Reynolds-averaged Navier-Stokes equations for complex geometries. The solution algorithm for each consists of the following basic steps: reconstruction of the solution states at the control

Figure 3. Payload geometry.

Table 1. Comparison of loads on axisymmetric and asymmetric PLFs. Normal Force (lbs) Axial Force (lbs) Base Moment (in.lbs)

Axisymmetric -35,992 -85,217 1,041,744Asymmetric -40,952 -163,099 1,873,979

Figure 2. Field pressure contours on axisymmetric (left) and asymmetric (right) PLFs.


5

volume faces, evaluation of the flux integrals for each control volume, and evolution of the solution in each control volume in time. The inviscid fluxes are computed using the flux-difference splitting technique of Roe.[3] A second-order spatial method is used to reconstruct the inviscid flux terms at the faces of the control volume and Barth’s limiter is also applied to these terms.[4] The viscous terms are discretized using a directional derivative technique. The temporal discretization is done using a second-order (in time) backward Euler implicit scheme. The time evolu-tion is accomplished using a Newton relaxation scheme to advance the unsteady solution at each time step.[5] The one-equation turbulence model of Spalart and Allmaras is used for simulation of turbulent effects in high Reynolds number flows. [6]

The optimization was performed using the UNCLE code.[1] It incorporates both direct and adjoint techniques for

calculating the sensitivity derivatives. As previously mentioned, the governing fluid equations are the Navier-Stokes equations, which are a system of nonlinear partial differential equations (PDE) expressing the conservation of mass, momentum, and energy. The differentiation of this system can be accomplished using one of two methods. For the first approach, referred to as the continuous or variation approach, the PDEs are differentiated prior to discretization and these differentiated equations are subsequently solved. In the second approach, referred to as the discrete approach, the PDEs are differentiated after discretization. The direct approach is used here. A detailed review of the aforementioned techniques and formulations for sensitivity analysis as utilized for aerodynamic shape optimization may be found in Ref. [8].

Both the direct and adjoint techniques require the computation of Jacobians and other linearizations in their con-struction. There are numerous ways in which these can be computed or obtained. The derivatives can be constructed by hand, using automatic differentiation software, or using numerical techniques such as finite differences. The derivatives in this work are computed using the Complex Taylor Series Expansion (CTSE) method.[9-12] The advan-tage of the CTSE method is that it yields a second-order accurate approximation without suffering from the subtrac-tive cancellation errors which are inherent in finite-difference approximations.

The current design employs a gradient-based optimization algorithm, and as such utilizes sensitivity derivatives to compute an appropriate search direction, s . Given this search direction, a one-dimensional or line search is per-formed in order to find the minimum along this path. Updating of the design variables, β , is usually carried out via

mmm sαββ += −1 (1)

where α is a scalar representing the step-length along the search direction. This one-dimensional search is performed in a sequential manner where the design variables are updated through successive increments of the step-length, α. Once a minimum is found, the respective design variables are then adopted as the current design, and gradient-information is then recomputed for a new search direction evaluation. The process is repeated until optimality con-ditions or stopping criteria are satisfied.

The initial design and geometric constraints placed on the fairing are shown in Figure 5. Many of the design con-straints were added to ensure that the fairing is able to accommodate the payload. However, other constraints were due to the launch vehicle itself, such as maintaining a sharp boat tail angle in order to ensure stable separation and to control shock waves during transonic flight. Constraints were also placed to maintain constant area sections in order

Figure 4. PLF geometric optimization flow diagram.


6

to reduce the effects of oscillating shock separation. The constraint on the maximum height of the fairing was placed to accommodate existing launch pad facilities. An additional constraint placed on the fairing design was that it had to maintain two planes of symmetry, both longitudinally and laterally.

The shape of the initial payload fairing design was parameterized with twelve Bezier surfaces. The parameteriza-tion is shown in Figure 6. The blue surface patches represent the forward section of the fairing and the green surface patches are constant straight section, i.e., the payload section. Due to the design being quarter-plane symmetric, the control points of the various Bezier surfaces were linked and thus this parameterization resulted in a total of 28 unique design variables. The design objective was to minimize a weighted function of the pitching moment and axial force at the base of the boat tail where it attaches to the Centaur launch vehicle. The weighting was biased toward the minimization of the pitching moment since this represented the most significant obstacle in producing a useable design. The single-point analysis was conducted at Mach number M = 1.0, angle of attack α = 4°, and a Reynolds number Re = 1.784x108 based on a reference length (the fairing height) of 75 ft. The design point was based on the launch vehicle maximum dynamic pressure, which corresponds to a dynamic pressure of 450 psf at an altitude of 33,800 feet.

Figure 7 illustrates the reduction in the bending moment with the line search history. An evolution of the payload fairing shape is shown in Figure 8. The final design, shown in Figure 9 and Figure 10, meets all of the geometric requirements. The most noticeable difference between the initial and final designs is the forward section of the fair-ing. The final design is considerably narrower, and the entire forward portion ahead of the constant straight section is concave whereas the initial design is slightly convex.

A comparison of the surface and field quantities for the initial and final fairing designs is illustrated in Figure 9. Figure 9 (a) shows a comparison of the windward side surface pressure distribution, and the skin friction coefficient distribution is shown in Figure 9 (b). As previously mentioned, the forward section of the final design is slightly concave. The concavity results in a gradual compression of the flow, which results in higher normal pressures and lower shear stresses (and consequently lower skin friction) on the forward section for the final design. The compres-sion also serves to decelerate the flow and results in a weaker oblique shock at the transition to the constant section. The weaker oblique shock results in lower normal pressure and skin friction in the regions downstream of the shock. The weaker oblique shock is evident by the Mach number field contours shown in Figure 9 (c).


7

Figure 5. Initial payload fairing design and constraints.

Figure 6. Surface parameterization for the initial payload fairing design.


8

Figure 7. Objective function history.

Figure 8. Payload fairing shape evolution.


9

(a) (b)

(c) Figure 9. Comparison of the (a) windward side surface pressure contours, (b) windward side skin friction contours, and (c) Mach number contours for the initial and final payload fairing designs.


10

IV. Structural Optimization Various structural analyses of the PLF were conducted to assess the strength, thermal, and dynamic responses of

the design, and to verify positive margins against the requirements. This section limits the discussion to the analyses utilized in the structural optimization of the fairing that was conducted to minimize the weight while meeting strength and stability requirements. Nastran Solution 200 Design Optimization [13-14] was used to achieve an optimal core thickness and faceskin layup.

A. Design Ultimate Load The structure was optimized to provide positive strength margins for the design ultimate load (DUL) defined in

Table 2. It is the combination of surface pressure loads for a flight condition of α = 4.0˚, β = 0.0˚, Mach = 1.075, Q = 450 psf, and inertial loads of -3.0 G axial (Y axis) and 0.3 G lateral (Z axis). See Figure 11 for definition of the global coordinate system. The DUL is considered an envelope of the worst-case aerodynamic loads combined with the highest expected inertial loads.

Figure 10. Asymmetric PLF mounted on launch vehicle.


11

B. Finite Element Model Various finite element models were created of the large payload fairing for structural analysis. A half-symmetry

model was used for optimization and strength calculations while a full model was developed for normal modes analysis. A full model was required for modes analysis to capture the primary bending mode about the symmetry plane. The half-symmetry model was simply reflected for this analysis. The base of the model was constrained in six degrees of freedom for all analyses.

C. Optimization of Sandwich Panel Layups As noted, Nastran Solution 200 was utilized to size the sandwich structure of the large payload fairing. The pri-

mary variables for optimization were faceskin ply thickness, faceskin ply orientations, and core thickness. The core thickness was forced to remain constant across all zones for fabrication reasons, but the thickness was optimized. The FEM was divided into three zones based on the results of initial stress analyses. Figure 12 shows the three lay-up zones in blue, red, and yellow. After these three zones were created, each was allowed to optimize to a unique layup that minimized overall weight while maximizing strength margins. The zone areas shown represent the fully optimized ideal design. The exact shapes of these zones will be approximated into rectangular regions for manufac-turability. Figure 13 and Figure 14 show the faceskin and core thicknesses and overall weight of the structure through the design iterations. The optimized Nastran layup output was manually adjusted to create a balanced and symmetric laminate for production reasons. This added one ply to zone 1 and two plies to zone 2. Further blending of the layup is anticipated to allow for the maximum continuous layers across the zones.

Table 2. Design ultimate load definition.

Gx[g]

Gy[g]

Gz[g]

α[deg]

β[deg] Mach No. Q

[psf]DUL 0.0 -3.0 0.3 4.0 0.0 1.075 450

Aerodynamic PressuresAccelerationsLoad Case

Figure 11. Design ultimate load combines worst-case pressure and inertial loads.


12

Figure 12. Identification of three distinct layup regions.


13

Ply and Core Thickness vs Design Iteration

0.0608

0.1216

0.1900

1.2943

0.0000

0.0500

0.1000

0.1500

0.2000

0.2500

0 1 2 3 4 5 6 7 8 9 10

Design Iteration

PID

Thi

ckne

ss [i

n]

0.0000

0.4000

0.8000

1.2000

1.6000

2.0000

Cor

e Th

ickn

ess

[in]

PID 100 PID 200 PID 300 CORE

Figure 13. Design variable iteration history.

Weight vs. Design Iteration

6267

4000

4500

5000

5500

6000

6500

7000

7500

0 1 2 3 4 5 6 7 8 9 10

Design Iteration

Stru

ctur

e M

ass

[lbm

]

4000

4500

5000

5500

6000

6500

7000

7500

Figure 14. Weight versus design iteration.


14

D. Strength Results The strength of the large payload fairing was assessed using Nastran. The composite failure capability of the

PCOMP card was utilized. The maximum strain failure theory was selected. The results shown in Table 3 are based on nonlinear quasi-static analysis. The strains were reviewed at the end

of each load step. When the failure index reaches unity at the end of a load step, the faceskin margin is zero. This occurred first in zone 1 at 1.3 times the DUL with a failure index of 1.032 (Figure 15 and 16), representing a margin of 0.26 at DUL [ MS = (1/1.032)*1.3*DUL – 1 = 0.26 ]. These results are considered more accurate than providing the strain margin at 1.0 times the DUL, which would show a margin of safety of 0.38 and which equates to a 12% overprediction.

The weight of each layup region is listed in both total weight (for each half of the fairing) and weight percentage. Although zone 3 (PID 300) has 25 plies, it represents only a very small portion of the fairing where stresses are highest.

The foam core represents 1954 lb of the 6562 lb per fairing half. This represents 29% of the weight. The core has high margins and could be substituted for lower density core materials such as 51 WF Rohacell to further reduce the mass. This has been shown to yield positive margins in all but small regions that would require a locally spliced higher density core.

Figure 15. Contour plot of failure indices at 1.3x DUL (isometric view).

Table 3. Fiber strength margins at DUL (nonlinear analysis results). PID MS Min Weight Weight [lbm] Weight [%] Faceskin Plies100 0.26 11.99745 4636 70.7% 9200 0.31 4.408225 1703 26.0% 18300 0.35 0.575777 222 3.4% 25

min → 0.26 total → 6562 100.0%


15

E. Stability Analysis Results Stability analysis of the large payload fairing was accomplished using two approaches with the FEM. Linear

buckling (Nastran Solution 105) was conducted to identify the maximum possible buckling load and buckling shape. The load values computed with linear analysis represent the theoretical maximum values for the structure and pro-vide a load regime to conduct nonlinear analysis. During nonlinear loading (Nastran Solution 106) with stiffness matrix updates, the buckling load can be considered after the nonpositive definite stiffness matrix is detected. In some cases, the collapse is progressive and does not occur suddenly but instead becomes more and more nonlinear. When this occurs, some criteria must be specified to identify the point at which unacceptable nonlinearity occurs. In the case of the large payload fairing, a snap-through event occurs at 1.925x DUL. At the next load step, the maxi-mum displacement increases from 17.98 inches to 81.93 inches. This snap-through event is considered the Pcrit load for collapse and is graphically depicted in Figure 17 and Figure 18. The faceskin fibers will reach zero margin at 1.26x DUL. The geometric stability of the structure is therefore separated from the strength by a factor of 1.53.

Figure 16. Contour plot of failure indices at 1.3x DUL (narrow view).


16

Displacement vs Ratio of DUL

y = 7.0037x - 0.2052R2 = 0.9991

0

2

4

6

8

10

12

14

16

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Design Ultimate Load Ratio[Load / DUL]

Max

Dis

plac

emen

t [in

]

0%

4%

8%

12%

16%

20%

24%

28%

32%

Dis

plac

emen

t Diff

eren

ce f

rom

Li

near

5 Zone Max Displacement5 Zone Max Displacement (perturbed)3 Zone Max Displacement5 Zone Displacement Difference from LinearLinear (Linear Region (to DUL))

Figure 18. Stability imperfection study of early design iterations (low sensitivity shown).

Load vs Displacement

1.925

1.3

0

0.5

1

1.5

2

2.5

3

0 5 10 15 20 25

Max Displacement [in]

Load

Rat

io [D

UL]

-1

-0.5

0

0.5

1

1.5

2

Fibr

e M

argi

n of

Saf

ety

Node 15307 Displacement Linear Buckling Zero Ply MarginLinear Regression MS First Ply Failure

Figure 17. Stability analysis results for final configuration.

Max Displacement in model

Linear Buckling

Fiber Margin

Linear Trend Projected From DUL to Linear

Buckling


17

F. Summary A composite sandwich panel fairing design was optimized to provide positive margins under DUL while

minimizing weight. The design is comprised of three distinct zones, each with a unique layup. All of the laminates are balanced and symmetric. The strength analysis showed a positive margin of 0.26. The buckling analysis (nonlin-ear collapse) showed that snap-through occurred at 1.925 times DUL. This indicates that the geometric stability of the structure is separated from the strength limit by a factor of 1.53. The optimized fairing design weighs 13,124 lb, meeting the weight target of 17,000 lb.

V. Wind Tunnel Testing Wind tunnel testing of a 2.5% scale fairing model was conducted at the Triumph Aerospace Trisonic Wind

Tunnel (TWT) facilities in El Segundo, California. All tests were completed as specified by the test plan [15]. This included 61 blows consisting of 71 runs.

The test plan outlined three types of tests: • Pitch-Pause: Steady blow (constant Mach number) having stepped variation in angle of attack (Test Matrix

1). • Mach Sweep: Unsteady blow having constantly increasing Mach number at constant angle of attack (Test

Matrices 2 and 3). • Flow Visualization (Flow Viz): Using flow visualization techniques such as oil flow, shadowgraphs, high-

speed video, fluorescent mini-tufts, to gather qualitative flow data (Test Matrix 4). The test plan was designed in a tiered manner such that each test matrix became increasingly comprehensive. The specific test matrices described in the test plan are summarized in Table 4.

A comparison of the static pressure test data to CFD predictions is presented in Figure 19. The CFD predictions

using the final fairing design were obtained using the U2NCLE unstructured flow solver. These plots show the nor-malized pressure coefficients (Cp) with respect to the pressure sensor location. The yellow bars represent the pres-sure coefficients predicted by the U2NCLE CFD code. The red bars represent the data taken from a pitch-pause test (TM1) where the Mach number was held constant while the fairing orientations were changed. The purple bars rep-resent the data taken from a Mach sweep test (TM2) where the fairing orientation was held constant while the Mach number was swept up to cover a prescribed range. The results show good agreement (differences of approximately 0.03% to 8.00%) between the two tests and CFD predictions. Considering the fact that the conditions during the test (TM1 and TM2) and the CFD analyses are not identical (i.e., the wind tunnel-scale model Reynolds number was approximately 106 versus the CFD full-scale Reynolds number, which was approximately 108), the agreement is very good. This agreement is a good indication that the CFD model is well correlated with the wind tunnel tests and, by extension, the actual fairing during flight conditions. Furthermore, the agreement between the two test results shows that the data quality is good and tunnel conditions predictable. Although the Mach sweep performed in TM2 is inherently nonstationary, its agreement with the stationary pitch-pause blows of TM1 indicates that the dwell time and sweep rates used for TM2 resulted in well-settled flow conditions that appear to be free of nonstationary effects. Some of the discrepancies between TM1 and TM2 results are due to the tighter tolerances on the fairing orientation angles than were achieved in TM2. Due to the nature of the pitch-pause technique, and the fact that TM1 was per-formed during early testing before the tunnel settings were optimized, larger errors were observed on the pitch and yaw controls compared to TM2.

Table 4. Test matrices. Test

Matrix Blow Type Mach Alpha (deg) Beta (deg)

1 Pitch-Pause 1.075 0, 1, 2, 3, 4 0, 1, 2, 3, 4 2 Mach Sweep 0.6 – 1.25 0 0 3 Mach Sweep 0.6 – 1.25 1, 2, 3, 4 1, 2, 3, 4 4 Flow Viz 0.7 and 1.1 0 0


18

The most successful flow visualization test was the oil flow imagery. This was performed by placing viscous

fluorescent oil drops on the fairing and photographing the resulting flow of the oil after a blow. Pathlines that were created by the blow provide a description of the flow patterns within the boundary layer around the body. Two such tests were performed at Mach 0.70 and 1.10 for α = 0˚ and β = 0˚. Figure 20 shows the results compared to CFD predictions. Despite the fact that the conditions between the two are not identical (differences in Reynolds numbers), the results show very good qualitative agreement.

VI. Conclusion A very novel and innovative design optimization methodology was developed to “shrink wrap” non-axisymmetric

fairings around large payloads which otherwise would require entirely new launch support facilities to be designed and built in order to be launched into space. Through this program, the feasibility of this methodology was proven by the development of an asymmetric payload fairing capable of accommodating a payload larger than the standard axisymmetric fairing for use on the Atlas V Heavy Lift vehicle without significant modification to the existing launch facilities. The geometry of the fairing was optimized using novel iterative CFD techniques. The composite sandwich panel comprising the fairing was optimized to reduce weight while meeting strength require-

Figure 19. Static pressure comparison at Mach 1.075 α=0° and β=0°.

Figure 20. Oil flow test results (left) compared to CFD predictions (right).


19

ments. Subscale wind tunnel testing using a 2.5% model was performed, and the results were found to agree extremely well with CFD predictions. This successful grounding of the newly developed analytical tools and meth-ods to test data provides the next step forward in the development of large composite asymmetric payload fairings.

Acknowledgments Tomoya Ochinero and Thomas Deiters would like to thank the Air Force Research Lab for their financial and

technical support, which made this research and development possible and resulted in a very novel and innovative fairing design process that has led to a new way of looking at how payload accommodations will be provided in the future.

References 1Pankajashan, R., Taylor, L.K., Sheng, C., Briley, W.R., and Whitfield, D.L., “Scalable Parallel Implicit Multigrid Solution

of Unsteady Incompressible Flows,” Frontiers of Computational Fluid Dynamics 2002, (Eds. D.A. Caughey and M.M. Hafez), World Scientific Publishing Co., 2002, pp. 181-195.

2Blades, E.L., and Marcum, D.L., "Numerical Simulation of a Spinning Missile with Dithering Canards Using Unstructured Grids," Journal of Spacecraft and Rockets, Vol. 41, 2004, pp. 248-256.

3Roe, P., “Approximate Riemann Solvers, Parameter Vectors, and Difference Schemes,” J. of Comp. Physics, Vol. 43, No. 2, 1981, pp. 357-372.

4Barth, T.J., and Jespersen, D. C., “The Design and Application of Upwind Schemes on Unstructured Meshes,” AIAA-89-0366, January 1989.

5Whitfield, D.L., and Taylor, L., “Discretized Newton-Relaxation Solution of High Resolution Flux-Difference Split Schemes,” AIAA Paper 91-1529, June 1991.

6Spalart, P.R., and Allmaras, S.R., “A One-Equation Turbulence Model for Aerodynamic Flows,” AIAA Paper 92-0439, January 1992.

7Briley, W.R., Taylor, L.K., and Whitfield, D.L., “High-Resolution Viscous Flow Simulations at Arbitrary Mach Number,” J. Comp. Physics, Vol. 184, No. 1, January 2003, pp. 79-105.

8Newman III, J.C., Taylor III, A.C., Barnwell, R.W, Newman, P.A., and Hou, G.J-.W., “Overview of Sensitivity Analysis and Shape Optimization for Complex Aerodynamic Configurations,” J. Aircraft, Vol. 36, No. 1, 1999, pp. 87-96.

9Newman III, J.C., Anderson, W.K., and Whitfield, D.L., “Multidisciplinary Sensitivity Derivatives Using Complex Variables,” MSU Report MSSU-EIRS-ERC-98-08, July 1998.

10Newman III, J.C., Whitfield, D.L., and Anderson, W.K., “A Step-Size Independent Approach for Multidisciplinary Sensitivity Derivatives,” J. Aircraft, Vol. 40, No. 3, May-June 2003, pp. 566-573.

11Anderson, W.K., Newman III, J.C., Whitfield, D.L., and Nielsen, E.J., "Sensitivity Analysis for the Navier-Stokes Equations on Unstructured Meshes Using Complex Variables," AIAA Journal, Vol. 39, No. 1, 2000, pp. 56-63.

12Burg, C.O.E., and Newman III, J.C., “Computationally Efficient, Numerically Exact Design Space Derivatives via the Complex Taylor’s Series Expansion Method,” Computers and Fluids, Vol. 32, No. 3, 2003, pp. 373-383.

13UGS, NX/Nastran Basic Nonlinear Analysis User’s Guide. 2004. 14UGS, NX/Nastran Design Sensitivity and Optimization User’s Guide. 2004. 15Ochinero, Tomoya T., “Asymmetric Payload Fairing Wind Tunnel Test Plan,” May 2007.

design and testing of a large composite asymmetric payload fairing

Documents