Correlation of HIFiRE-5b Flight Data With Computed

Pressure and Heat Transfer for Attitude Determination

Joseph S. Jewell*, Roger L. Kimmel�, and David W. Adamczak�

U.S. Air Force Research Laboratory, Wright-Patterson AFB, OH 45433, USA

Jonathan Poggie§ and Kevin M. Porter¶

Purdue University, West Lafayette, IN 47907, USA

Thomas J. Juliano�

University of Notre Dame, Notre Dame, IN 46556, USA

Flight pressure has been compared to angle-of-attack- and yaw-dependent computa-tional fluid dynamics results for pressure distribution at three time points in the descenttrajectory of the HIFiRE-5b. Computed pressures, normalized by freestream pressure,were interpolated to the flight Mach numbers at each time point throughout the descenttrajectory, and angle-of-attack and yaw were estimated from measured pressure by deter-mining the combination minimizing the difference between the measured and computedpressures. The resulting vehicle attitude was compared to the vehicle attitude derivedfrom inertial measurement unit and GPS results from the flight. The two methods showedexcellent agreement for the entirety of the analyzed reentry portion of the trajectory.

I. Introduction

The Hypersonic International Flight Research Experimentation (HIFiRE) program is a hypersonic flighttest program executed by the Air Force Research Laboratory (AFRL) and the Australian Defence Scienceand Technology Group (DSTG). Its purpose is to develop and validate technologies critical to next genera-tion hypersonic aerospace systems. Candidate technology areas include, but are not limited to, propulsion,propulsion-airframe integration, aerodynamics and aerothermodynamics, high temperature materials andstructures, thermal management strategies, guidance, navigation, and control, sensors, and system com-ponents. The HIFiRE program consists of extensive ground tests and computation focused on specifichypersonic flight technologies. Each technology program is designed to culminate in a flight test.

The philosophy of the HIFiRE program has been to identify and attack specific technology gaps in hypersonicflight. Preference is given to phenomena that are difficult to analyze computationally or with ground test.The intent of the program is to characterize the technology as fully as possible with a program of coordinatedground test and computation, culminating in a test flight for emphasis.

The HIFiRE-5b vehicle flew a ballistic trajectory, with no active attitude control.1 The elliptic cone testarticle remained attached to the second stage booster at all times, and relied on aerodynamic stability tominimize angle of attack. The payload spun to minimize trajectory dispersions. Since the payload was

*Research Aerospace Engineer (NRC Research Associate), AFRL/RQHF, 2145 5th St., Room 202F, Building 24B.jjewell@alumni.caltech.edu. AIAA Senior Member.

�Principal Aerospace Engineer, AFRL/RQHF, Wright-Patterson AFB, OH 45433. AIAA Associate Fellow.�Senior Aerospace Engineer, AFRL/RQHV, Wright-Patterson AFB, OH 45433. AIAA Senior Member.§Associate Professor, School of Aeronautical and Astronautical Engineering. AIAA Senior Member.¶Graduate Student, School of Aeronautical and Astronautical Engineering. AIAA Student Member.�Assistant Professor, Department of Aerospace and Mechanical Engineering. AIAA Senior Member.

1 of 7

American Institute of Aeronautics and AstronauticsCleared for Public Release, Case Number: 88ABW-2017-2317.

47th AIAA Fluid Dynamics Conference

5-9 June 2017, Denver, Colorado

10.2514/6.2017-3133

This material is declared a work of the U.

S. Government and is not subject to copyright protection in the United States.

AIAA AVIATION Forum

generally at some small angle of attack and spinning, any given point on the payload showed an oscillatoryangle of attack and yaw (or equivalently, total angle of attack and roll) relative to the wind. Since thetransition location is a function of vehicle attitude, it is important to determine accurately both the attitudeand the time-dependent transition location.2 The analysis in the present work closely follows methodsdeveloped for attitude determination of the HIFiRE-5a flight.3,4

Calculations provide both an assessment of measured and computed quantities, and a means of reconstructingthe flight. Analysis of the HIFiRE-5a flight compared measured data to heating and pressure estimates, aswell as computational fluid dynamics (CFD) results for attitude determination3–5 and similar work hasbeen used to derive flight attitude where Global Positioning System (GPS) information is unavailable, suchas during Martian atmospheric entry,6 and combined with inertial measurements to improve accuracy.7

Computed pressures may then be used to back-calculate the vehicle attitude to establish a check of theattitude measured by the on-board inertial measurement unit (IMU) and GPS. The method for determiningvehicle attitude from GPS and IMU was described in a prior paper.8 Flight angle of attack and yawinformation is also used to inform predicted laminar and turbulent heat transfer levels for boundary layertransition determination.2 Since both the GPS and IMU suffered minor anomalies during the HIFiRE-5bflight, the pressure-derived vehicle attitudes took on added significance.1 Detailed CFD calculations of theHIFiRE-5b flight at reentry flight conditions between Mach 7.7 and Mach 8 have been performed as partof this effort9 and are used in subsequent sections of this paper. Complementary wind-tunnel tests10 andboundary layer stability analysis11 have also been performed.

II. Pressure Distribution RMS Analysis and Comparison with CFD

Computations9 were performed at five values of α and β (-2.0°, -1.0°, 0.0°, 1.0°and 2.0°) for flight Machnumbers between 7.7 and 8.0, which correspond to the period at t=513–518 s during the reentry portionof the HIFiRE-5b flight.1 These conditions were chosen, on the basis of preliminary trajectory estimates,to bracket the attitude and Mach number range during transitional flow during reentry. Both laminar andturbulent conditions were calculated. Surface pressures did not show tangible differences between laminarand turbulent cases. The two planes of symmetry of the vehicle were exploited where possible to reduce thenumber of necessary calculations.

Figure 1, describes the angle of attack and yaw sign conventions used in both HIFiRE-5a and the presentHIFiRE-5b analysis, as well as the definition of the circumferential angular coordinate.

Figure 1. α, β, and φ sign conventions used in the present analysis.

Comparison of measured to computed pressures provided a method for assessment of vehicle attitude in-dependent of the IMU and GPS. While 15 flight pressures at multiple locations were used for attitudedetermination in HIFiRE-5a,4 an attitude assessment scheme involving 4 transducers equally spaced by90° in φ was found to be essentially equivalent and computationally more efficient, and is used in the presentanalysis. The method relied on finding the vehicle attitude that minimized the RMS (root mean squared)difference between measured and computed pressures. A similar approach, utilizing a matrix of CFD solu-

2 of 7

American Institute of Aeronautics and AstronauticsCleared for Public Release, Case Number: 88ABW-2017-2317.

tion points, has recently been used for the implementation of Flush Air Data Sensing (FADS) algorithms forreconstructing the Mars Science Laboratory entry, descent, and landing trajectory.6 The RMS difference isdefined as

p̃ =1

p∞

144∑

i=1

(pCFD − pF)2i

12

(1)

Here, p∞ is the freestreamm pressure, pCFD is the computed pressure at sensor location, and pF is the pressuremeasured by the transducer. To reduce the effect of transducer drift, all pressure signals are re-zeroed duringpost-processing for the exoatmospheric portion of the flight, prior to reentry.5

Kulite pressure transducers measured local static pressures. Additionally, several pressure transducers wereoperated in differential mode to measure differential pressures 180 degrees apart on the vehicle to aid inattitude determination. The differential transducers are examined separately in Section III. Details of theHIFiRE-5b pressure transducers are found in Kimmel et al.1

Since it was infeasible to perform multiple CFD solutions at various α, β combinations, discrete combinationsof α and β were calculated at freestream Mach numbers between 7.7 and 8.0. These Mach numbers boundedthe reentry flight case of interest. These CFD cases were interpolated in Mach number to estimate computedpressures (and heat transfer) at intermediate Mach numbers using the Matlab griddedInterpolant withsix control variables (M, Re, α, β, φ and x) as described in Porter et al.9

The process as described in Jewell et al.4 was repeated at 0.005 s intervals for 508 ≤ t ≤ 520.5 s. Figure 2shows vehicle attitude derived from pressure and IMU/GPS, and the minimum RMS between measured andcomputed pressures for the ascent case. Although the amplitudes of dominant IMU/GPS-derived cyclic αand β oscillations early in the descent are larger than those derived from pressure data, the mean values agreeextremely well. A higher variation was expected at early times in the reentry trajectory, when pressureswere low. Prior to t=510 s, atmospheric pressure was less than 1% of the payload pressure transducerrange. The oscillation frequency agrees well between both data sets. Both analysis methods show attitudevariations damping with time. Figure 3 is a detail of the portion of the reentry trajectory analyzed, simulated,or computed in the accompanying papers,1,2, 10,11 where the best agreement is obtained. In all cases,results were linearly interpolated from 0.2° intervals output from the Matlab griddedInterpolant, and nosignificant difference was found in the angles computed from pressure signals smoothed over moving 0.005,0.01, or 0.02 s windows.

The RMS deviation between measured and computed pressures is less than 8% for t ≥ 512 s.

The oscillation frequencies were examined by means of fast Fourier transforms performed using the Matlabfft function. The descent signals were analyzed for t = 514 − 518 s, the range containing the target Machand Reynolds numbers, and the results are depicted in Figure 4). While both the pressure-difference RMSand IMU/GPS signals have a peak just below 7 Hz, and a smaller peak between 3 and 4 Hz, the IMU/GPSsignal also has a large peak near 5 Hz which is not seen in the pressure-difference RMS signal. This isqualitatively similar to the spectra for the descent of HIFiRE-5a,4 where both the pressure-difference RMSand IMU signals had a peak just below 1 Hz and near 3 Hz, but only the IMU signal had a strong peak at2 Hz. The source of the persistent ∼5 Hz peak in the IMU/GPS signals is unknown, as was the source ofthe persistent ∼2 Hz peak in the IMU/GPS signals from HIFiRE-5a, but neither was observed on any of thedifferential pressure traces examined for HIFiRE-5a or HIFiRE-5b.

The oscillation frequencies were also examined with FFTs performed over 2 s moving windows, and theseresults are presented in Figure 5. The oscillation frequencies of both α and β roughly double over theexamined period, and the α and β peaks are identical within the uncertainty of the FFT, which is 0.5° dueto the relatively small window size.

III. Comparison with Differential Pressures

Two differential pressure transducers were examined to qualitatively confirm the surface-pressure RMS com-puted angles of attack and yaw from t = 514 − 518 s. Transducer PLBW29D is located on the centerline(major axis) at φ = 0°, and primarily sensitive to angle of attack, α, and Transducer PLBW25D is located

3 of 7

American Institute of Aeronautics and AstronauticsCleared for Public Release, Case Number: 88ABW-2017-2317.

505 510 515 520−4

−2

0

2

4

α

(d

eg

)

505 510 515 520−4

−2

0

2

4

β

(de

g)

505 510 515 5200

5

10

15

20

25

t (sec)

Min

% R

MS

err

or

4 sens., 0.005 sec., 0.2 deg.

4 sens., 0.01 sec., 0.2 deg.

4 sens., 0.02 sec., 0.2 deg.

IMU

Figure 2. Angle of attack (α) and yaw (β) results for the ascent portion of the trajectory from the interpolation/RMSminimization routine compared with the IMU/GPS values, from the HIFiRE-5b flight. Trajectory times approximatelycorresponding to the computed CFD cases (514 ≤ t ≤ 518 s) are indicated by vertical dashed lines. See Figure 3 for adetail of this region only.

4 of 7

American Institute of Aeronautics and AstronauticsCleared for Public Release, Case Number: 88ABW-2017-2317.

513 514 515 516 517 518 519−4

−2

0

2

4α

(

deg)

513 514 515 516 517 518 519−4

−2

0

2

4

β (d

eg)

513 514 515 516 517 518 5190

2

4

6

8

10

t (sec)

Min

% R

MS

err

or

4 sens., 0.005 sec., 0.2 deg.

4 sens., 0.01 sec., 0.2 deg.

4 sens., 0.02 sec., 0.2 deg.

IMU

Figure 3. Detail of angle of attack (α) and yaw (β) results for the ascent portion of the trajectory from the interpo-lation/RMS minimization routine compared with the IMU/GPS values, from the HIFiRE-5b flight. Trajectory timesapproximately corresponding to the computed CFD cases (514 ≤ t ≤ 518 s) are indicated by vertical dashed lines.

0 1 2 3 4 5 6 7 8 9 100

0.05

0.1

0.15

0.2

0.25

Frequency (Hz)

FF

T M

ag

nitu

de

Re

sp

on

se

α RMS

β RMS

α IMU

β IMU

Figure 4. FFT magnitude response for descent α and β derived from IMU and RMS, t = 514 − 518 s.

5 of 7

American Institute of Aeronautics and AstronauticsCleared for Public Release, Case Number: 88ABW-2017-2317.

505 510 515 5203

4

5

6

7

8

Time (s)

Fre

quency (

Hz)

α frequency peak (2s moving window)

β frequency peak (2s moving window)

Figure 5. Peak FFT magnitude response for descent α and β derived from pressure-difference RMS in 2 s movingwindows, t = 505 − 519 s.

on the leading edge (minor axis) at φ = 90°, and primarily sensitive to yaw, β. Figure 6 presents differentialpressure measurements, normalized by freestream pressure from the best estimated trajectory, from the de-scent phase of the HIFiRE-5b flight. Axis scales were chosen to make the normalized differential pressuretraces approximately coincident with the angle traces. The assumption is that the differential pressure islinearly correlated with alpha and beta, which is reasonable over the small range of α, β and M considered.The normalized differential pressures agree substantially with the RMS-derived attitude results in terms offrequency and relative magnitude in both phases of the flight, which further confirms the reliability of thepressure-RMS method.

514 514.5 515 515.5 516 516.5 517 517.5 518−2

−1

0

1

2

t (sec)

α °

(d

eg

)

514 514.5 515 515.5 516 516.5 517 517.5 518−1.6

−0.8

0

0.8

1.6

∆ P

CL /

P∞

α

∆ PCL

/ P∞

514 514.5 515 515.5 516 516.5 517 517.5 518−2

−1

0

1

2

t (sec)

β ° (

deg)

514 514.5 515 515.5 516 516.5 517 517.5 518−1.5

−0.75

0

0.75

1.5

∆ P

LE / P

∞

β

∆ PLE

/ P∞

Figure 6. Descent α and β derived from RMS pressure compared with normalized centerline and leading edge differentialpressures.

6 of 7

American Institute of Aeronautics and AstronauticsCleared for Public Release, Case Number: 88ABW-2017-2317.

IV. Conclusions

It has been demonstrated that normalized pressure CFD results may be used to infer angle of attack andyaw from a set of four pressure transducers distributed over the body of the HIFiRE-5b flight test article.Interpolations in Mach number have been correlated with the inertial measurement unit results for α and βfrom the flight, with excellent agreement for the reentry portion of the trajectory. Although attitudes derivedfrom both methods agree well, the pressure-derived attitudes are preferred for data analysis, since both theIMU and GPS suffered anomalies during flight. The calculated α and β also agree well with differentialpressure transducer measurements. The vehicle was found to increase in oscillation rate over the course ofits descent from about 4 Hz to 7.5 Hz. Further work to characterize the effect of the attitude on heat transferrates both laminar and turbulent and comparison with measured heat transfer rates will be pursued basedon these results. One recommendation is that to extend the method’s utility to a wider range of altitudes,high and low-range pressure transducers could be incorporated on the payload.

Acknowledgments

Dr. Jewell holds a National Research Council Research Associateship Award at the U.S. Air Force ResearchLaboratory. Dr. Juliano, Dr. Poggie, and Mr. Porter were supported by the U.S. Air Force Summer FacultyFellowship Program.

References

1Kimmel, R. L., Adamczak, D. W., and DSTG AVD Brisbane Team, “ HIFiRE-5b Flight Overview,” AIAA Aviation 2017 ,AIAA-2017-XXXX, Denver, CO, 2017.

2Juliano, T. J., Porter, K. M., Poggie, J., Kimmel, R. L., Jewell, J. S., and Adamczak, D. W., “HIFiRE-5b Heat Flux andBoundary-Layer Transition,” AIAA Aviation 2017 , AIAA-2017-XXXX, Denver, CO, 2017.

3Jewell, J. S., Miller, J. H., and Kimmel, R. L., “Correlation of HIFiRE-5 Flight Data With Computed Pressure and HeatTransfer,” AIAA Aviation 2015 , AIAA-2015-2319, Grapevine, TX, 2015.

4Jewell, J. S., Miller, J. H., and Kimmel, R. L., “Correlation of HIFiRE-5a Flight Data with Computed Pressure and HeatTransfer,” Journal of Spacecraft and Rockets, 2017, In Press.

5Juliano, T. J., Adamczak, D., and Kimmel, R. L., “HIFiRE-5 Flight Test Results,” Journal of Spacecraft and Rockets, Vol. 52,No. 3, 2015, pp. 650–663.

6Karlgaard, C. D., Kutty, P., Schoenberger, M., Munk, M. N., Little, A., and Kuhl, C., “Mars Science Laboratory EntryAtmospheric Data System Trajectory and Atmosphere Reconstruction,” Journal of Spacecraft and Rockets, Vol. 51, No. 4,2014, pp. 1029–1047.

7Karlgaard, C. D., Kutty, P., and Schoenberger, M., “Coupled Inertial Navigation and Flush Air Data Sensing Algorithm forAtmosphere Estimation,” AIAA SciTech 2015 , AIAA-2015-0526, Kissimmee, FL, 2015.

8Kimmel, R. L., Adamczak, D. W., Juliano, T. J., and DSTG AVD Brisbane Team, “ HIFiRE-5 Flight Test PreliminaryResults,” AIAA SciTech 2013 , AIAA-2013-0377, Grapevine, TX, 2013.

9Porter, K. M., Poggie, J., and Kimmel, R. L., “Laminar and Turbulent Flow Calculations for the HIFiRE-5b Flight Test,”AIAA Aviation 2017 , AIAA-2017-XXXX, Denver, CO, 2017.10Borg, M. P. and Kimmel, R. L., “Ground Test Measurements of Boundary-Layer Instabilities and Transition for HIFiRE-5at Flight-Relevant Attitudes,” AIAA Aviation 2017 , AIAA-2017-XXXX, Denver, CO, 2017.11Tufts, M. W., Gosse, R., and Kimmel, R. L., “PSE Analysis of Crossflow Instability on HIFiRE 5b Flight Test,” AIAAAviation 2017 , AIAA-2017-XXXX, Denver, CO, 2017.

7 of 7

American Institute of Aeronautics and AstronauticsCleared for Public Release, Case Number: 88ABW-2017-2317.