9953056974 Call Girls In South Ex, Escorts (Delhi) NCR.pdf
Sanny omar esa_presentation_no_video
1. Sanny Omar, PhD
University of Florida
ADAMUS lab
ESA Presentation
1
A Drag Device and Control Algorithm for
Spacecraft Attitude Stabilization and De-Orbit
Point Targeting using Aerodynamic Drag
2. Outline
Research objectives
Drag De-Orbit Device (D3) hardware and
attitude stability
Guidance trajectory generation algorithm
Navigation and trajectory tracking algorithms
D3 CubeSat mission
Conclusions and future work
2
3. Research Overview and Objectives
Most Low Earth Orbit (LEO) spacecraft do not have thrusters and
re-enter atmosphere in random locations at uncertain times
Objects pose a risk to persons, property, or other satellites
Skylab de-orbit over Australia is an example
Has become a larger concern with the recent increase in small satellites
Retractable drag de-orbit device (D3) to expedite de-orbit and
facilitate orbital maneuvering, collision avoidance, and re-entry
point targeting
“Targeting Algorithm” which utilizes aerodynamic drag modulation
to lead a satellite to a desired de-orbit location
Guidance trajectory generation, navigation, and guidance tracking
components
2U CubeSat to test the drag device and control algorithms in orbit
3
5. Drag De-Orbit Device (D3) Overview
Drag De-Orbit Device (D3) attaches to
existing CubeSats to facilitate de-orbit
of a 12U, 15 kg satellite in 25 years
from a 700 km circular orbit
.5 𝑚2
drag area
D3 is retractable and facilitates re-entry
point targeting, orbital maneuvering,
and collision avoidance
D3 Installed on CubeSat
5
Orbit Lifetime vs. Initial Circular Orbit Altitude with D3 and in Max Drag without D3
6. Drag De-Orbit Device (D3) Design
Consists of four retractable, independently
actuated tape spring booms
4 cm wide and up to 3.7 m long
Inclined at 20 degree angle relative to x-y plane
Boom angle provides passive aerodynamic
attitude stability
Booms collectively deployed or retracted to
modulate drag for orbital maneuvering
Swarm maintenance and controlled re-entry
capabilities
Two opposing booms can be partially
retracted while other two remain fully
deployed for gravity gradient stabilization
Embedded magnetorquers used for detumble
and damping steady state attitude oscillations
x
z
y
6
7. D3 Interfacing
D3 adapter stage houses magnetorquers and interfaces
with standard CubeSat structure
Each D3 deployer requires two wires for motor and four
wires for rotary encoder that measures boom
deployment
24 wire ribbon cable for deployers
6 wire ribbon cable for magnetorquers
7
x
y
z
9. Maximizing Miss Distances using Aerodynamic Drag for 400 and 600 km Circular Orbits
400 km orbit 600 km orbit
9
Product of orbit lifetime and surface area unchanged before and after drag device deployment
If collision risk directly related to cross-sectional area, drag device may not reduce collision risk
Ability to maneuver away from impending collisions makes D3 a more appealing de-orbit mechanism than
static sails
10. Power-limited BDot De-Tumble Law
Conventional B-Dot: 𝝁 𝑑𝑒𝑠 = −𝐾 𝑩
𝑩 =
𝑩2 − 𝑩1
Δ𝑡
𝝉 𝑚𝑎𝑔 = 𝝁 × 𝑩
If 𝑩 is large, achieving 𝝁 𝑑𝑒𝑠 might
require excessive power
𝝁 = 𝐼𝐴𝑛 𝒏
𝑃 = 𝐼𝑉 = 𝐼2 𝑅
Scale magnitude of 𝝁 𝑑𝑒𝑠 such the
required power ≤ 𝑃𝑚𝑎𝑥
Required 𝑃𝑟𝑒𝑞 calculated
If P𝑟𝑒𝑞 ≥ 𝑃𝑚𝑎𝑥
𝝁 𝑑𝑒𝑠 = 𝝁 𝑑𝑒𝑠0
𝑃 𝑚𝑎𝑥
𝑃𝑟𝑒𝑞
10
11. Attitude Simulation
Spacecraft z-axis should align with LVLH
y-axis (along-track) while spacecraft x-axis
aligned with LVLH x-axis (radial)
Less than 5 degree pointing error in ISS
orbit
Magnetorquers utilized for rate damping
and ensuring gravity gradient stabilization
with x-axis zenith instead of nadir pointing
11
x
y
z
14. Algorithms Overview
“Targeting Algorithm” has three components
Guidance Trajectory Generation Algorithm
Computes the ballistic coefficient (Cb) over time profile and corresponding
trajectory that a satellite must follow to de-orbit in a desired location
Navigation Algorithm with Kalman Filtering
Given noisy GPS measurements, estimates the position and velocity of the
spacecraft relative to the guidance trajectory
Guidance Trajectory Tracking Algorithm
Based on the relative position and velocity, computes the ballistic coefficient that
spacecraft must maintain to return to the guidance trajectory
Continues LQR-based full state feedback
14
15. Orbit Simulation Framework
State is cartesian position (𝒓) and velocity (𝒗) relative to Earth Centered
Inertial (ECI) frame
ECI frame defined as aligned with ECEF frame at epoch
Equations of motions written in state space form and numerically integrated
using MATLAB ode113 or in-house RKF7(8) numerical integrator
𝒓
𝒗
=
𝒗
∑𝑭
𝑚
Earth gravitational and aerodynamic forces included in simulator
Relativity, solar gravity, lunar gravity, tidal effects, solar pressure, variable
winds, and precession/nutation of Earth’s rotation axis ignored
Accuracy sufficient for guidance trajectory generation
15
16. Guidance Trajectory Generation Algorithm
Given a numerically propagated decay trajectory, it is possible to analytically estimate the Cb
profile needed to de-orbit in a desired location 𝐶 𝑏 =
𝐶 𝑑 𝐴
2𝑚
Shrinking horizon with drag-work enforcement guidance generation strategy
Trajectory propagated with analytical Cb profile. First tg seconds comprises first part of guidance trajectory
tg is 1/10 of orbit life on each step
𝐶 𝑏 varied during 𝑡 𝑔 seconds of propagation to ensure total work done by drag matches analytical solution
New Cb profile analytically calculated, orbit propagated, and first tg seconds of resulting trajectory appended to
guidance
Procedure continues until trajectory found that yields low enough guidance error or less than certain amount (1
day) of orbit life remaining
16
17. Guidance Trajectory Generation Analytical Solution
Must control de-orbit latitude and longitude at given geocentric altitude
Final time free
Control parameters are
tswap = time until ballistic coefficient is changed
Cb1 = ballistic coefficient from t0 to tswap
Cb2 = ballistic coefficient from tswap to tterm
Spacecraft maintains some predetermined drag profile after tterm
Given enough time, variation of these parameters is sufficient to target any de-
orbit point with latitude below the orbit inclination
Analytical Solution Assumptions
Circular orbit around spherical Earth
Density is a function of semi major axis
If density is a function of altitude in a circular orbit around a spherical Earth, density is also a function of
semi major axis
De-Orbit point is before aerodynamic forces exceed gravitational forces (above 70 km altitude)
orbital elements still valid
Shrinking horizon strategy eliminates errors resulting from these assumptions
17
18. Analytical Mapping from Initial to Final State
Fundamental building block of analytical solution
If a satellite with Cb1 takes time t1 to achieve some change in semi major axis
and experiences a change in argument of latitude ∆𝑢1 during this time
𝑢 = 𝜔 + 𝜃 , then for a satellite with the same initial conditions and Cb2
𝑡2 =
𝐶 𝑏1 𝑡1
𝐶 𝑏2
∆𝑢2=
𝐶 𝑏1∆𝑢1
𝐶 𝑏2
The average rate of change of right ascension Ω for a given change in semi
major axis is independent of ballistic coefficient
ΔΩ = Ω𝑡
Ballistic coefficient is defined as 𝐶 𝑏 =(drag_coefficient*drag_area)/(2*mass)
18
19. Characterizing New Trajectory Based on Old Trajectory
Divide trajectories into four phases from same initial
to final semi major axes
Cb values are unchanging in each phase
Time and changes in orbital elements during each
phase in the new trajectory calculated based on
corresponding phase in old trajectory and analytical
relations
Final time and orbital elements based on initial
conditions and sum of changes during each phase
𝑡𝑓 = 𝑡𝑖𝑛𝑖𝑡 +
𝑖=1
4
𝐶 𝑏𝑖0Δ𝑡𝑖0
𝐶 𝑏𝑖
𝑢 𝑓 = 𝑢𝑖𝑛𝑖𝑡 +
𝑖=1
4
𝐶 𝑏𝑖0Δ𝑢𝑖0
𝐶 𝑏𝑖
Ω 𝑓 = Ω𝑖𝑛𝑖𝑡 +
𝑖=1
4
ΔΩ𝑖0
Δ𝑡𝑖0
Δ𝑡𝑖
𝜔 𝑓 = 𝑒𝑓 = 0
𝑖 𝑓 = 𝑖𝑖𝑛𝑖𝑡
Time and orbital elements of the new spacecraft at de-
orbit point can be calculated and used to calculate de-
orbit latitude and longitude. 19
20. Analytical Guidance Trajectory Solution
Given initial trajectory propagated with some Cb1, Cb2, and
tswap
Orbit life and argument of latitude increases (Δ𝑡 𝑑, Δ𝑢 𝑑)
required for proper targeting calculated
Argument of latitude at target latitude calculated
Required increase in argument of latitude (Δ𝑢 𝑑) calculated
Any Δ𝑢 𝑑 + 2𝜋𝑛 acceptable
Spacecraft lat-long calculated where spacecraft passes over target
latitude
Required life increase calculated (longitude 𝜆 positive east)
Δ𝜆 = 𝜆 𝑑 − 𝜆 𝑎
Δ𝑡 𝑑 = −
Δ𝜆
2𝜋
𝑇⨁
𝑇⨁ =86,164 s = sidereal day
Total required orbit life and argument of latitude change
Δ𝑡𝑡 = Δ𝑡𝑡0 + Δ𝑡 𝑑, Δ𝑢 𝑡 = Δ𝑢 𝑡0 + Δ𝑢 𝑑
z
Life
increase
required
Target
Actual de-
orbit
20
21. Calculating New Control Parameters
Control parameters to achieve desired
Δ𝑢 𝑡 and Δ𝑡𝑡
𝐶 𝑏2 =
𝐶 𝑏20 Δ𝑡20Δ𝑢10 − Δ𝑡10Δ𝑢20
Δ𝑡𝑡Δ𝑢10 − Δ𝑡10Δ𝑢 𝑡
𝐶 𝑏1 =
𝐶 𝑏10 Δ𝑡10Δ𝑢20 − Δ𝑡20Δ𝑢10
Δ𝑡𝑡Δ𝑢20 − Δ𝑡20Δ𝑢 𝑡
𝑡 𝑠𝑤𝑎𝑝 =
Δ𝑡10 𝐶 𝑏10
𝐶 𝑏1
Compute control solution for multiple
initial values of tswap to explore full
control space
Select solution with maximum remaining
orbit lifetime controllability
21
22. Ensuring Feasible Parameter Ranges
Can analytically calculate 𝑢10, 𝑢20, Δ𝑡10, Δ𝑡20 given a numerically propagated trajectory
and initial 𝐶 𝑏10, 𝐶 𝑏20 and 𝑡 𝑠𝑤𝑎𝑝
For a given required total change in argument of latitude (Δ𝑢 𝑡), the feasible Δ𝑡𝑡 values
can be calculated based on min and max Cb values
For Δ𝑡𝑡 𝑚𝑎𝑥
for given Δ𝑢 𝑡 need largest Cb2 and smallest Cb1
Want to spend as long as possible in higher orbit with longer period
Δ𝑡𝑡 𝑚𝑎𝑥
= min
𝐶 𝑏20 Δ𝑡20Δ𝑢10 − Δ𝑡10Δ𝑢20 + 𝐶 𝑏 𝑚𝑎𝑥
Δ𝑡10Δ𝑢 𝑡
𝐶 𝑏 𝑚𝑎𝑥
Δ𝑢10
,
𝐶 𝑏10 Δ𝑡10Δ𝑢20 − Δ𝑡20Δ𝑢10 + 𝐶 𝑏 𝑚𝑖𝑛
Δ𝑡20Δ𝑢 𝑡
𝐶 𝑏 𝑚𝑖𝑛
Δ𝑢20
Δ𝑡𝑡 𝑚𝑖𝑛
= max
𝐶 𝑏20 Δ𝑡20Δ𝑢10 − Δ𝑡10Δ𝑢20 + 𝐶 𝑏 𝑚𝑖𝑛
Δ𝑡10Δ𝑢 𝑡
𝐶 𝑏 𝑚𝑖𝑛
Δ𝑢10
,
𝐶 𝑏10 Δ𝑡10Δ𝑢20 − Δ𝑡20Δ𝑢10 + 𝐶 𝑏 𝑚𝑎𝑥
Δ𝑡20Δ𝑢 𝑡
𝐶 𝑏 𝑚𝑎𝑥
Δ𝑢20
If desired Δ𝑡𝑡 not is feasible range, pick closest achievable Δ𝑡𝑡
22
23. Drag-Work Enforcement Method
During orbit propagation, total work done by aerodynamic drag
(per unit mass) is a state variable in addition to ECI position
and velocity
𝑊𝑑 = 𝑎 𝑑 𝑣
Desired work done by drag at each point in new trajectory can
be calculated prior to trajectory propagation
During propagation of new trajectory sections that will be used
in guidance trajectory (first 𝑡 𝑔 seconds), satellite 𝐶 𝑏 iteratively
varied until actual 𝑊𝑑 equals desired 𝑊𝑑 𝑑𝑒𝑠
from analytical
solution
𝐶 𝑏 updated and trajectory section re-run until 𝑊𝑑 = 𝑊𝑑 𝑑𝑒𝑠
achieved
Update law: 𝐶 𝑏 𝑢𝑝
= 𝐶 𝑏 𝑜𝑙𝑑
𝑊 𝑑 𝑑𝑒𝑠
𝑊 𝑑
First 𝑡 𝑔 seconds of
guidance propagated
with 𝐶 𝑏
Drag work 𝑊𝑑 at 𝑡 𝑔
calculated
𝑊𝑑 compared to
desired drag work
𝑊𝑑 𝑑𝑒𝑠
𝐶 𝑏 updated based on
difference between 𝑊𝑑 and
𝑊𝑑 𝑑𝑒𝑠
Init 𝐶 𝑏 from
analytical
solution
𝑊𝑑 − 𝑊𝑑 𝑑𝑒𝑠
above threshold
Remainder of
trajectory
propagated𝑊𝑑 − 𝑊𝑑 𝑑𝑒𝑠
below
threshold
23
24. Full State Feedback Guidance Tracking using Schweighart Sedgwick Relative Motion
Equations
Tracker ensures spacecraft follows guidance trajectory despite drag
uncertainties
SS Equations provide linear approximation of 𝒓 𝑟𝑒𝑙 over time in LVLH
frame
Like Clohessy-Wiltshire equations but include J2 effects
Approximation based on circular orbit assumption
𝒓 𝑟𝑒𝑙 = 𝒓 𝑠𝑐 − 𝒓 𝑔𝑢𝑖𝑑 =
𝛿𝑥
𝛿𝑦
𝛿𝑧
Linear Quadratic Regulator (LQR) full state feedback law based on SS
dynamics used to regulate 𝒓 𝑟𝑒𝑙 to zero
24
26. Full State Feedback Control
𝐶 𝑏 𝑠𝑐
= 𝐶 𝑏 𝑔𝑢𝑖𝑑
− 𝐾𝒙
𝐾 = 𝑙𝑞𝑟(𝐴, 𝐵, 𝑄, 𝑅, 0)
System of form 𝒙 = 𝐴𝒙 + 𝐵𝑢
𝒙 is relative position and velocity, 𝑢 = Δ𝐶 𝑏 = 𝐶 𝑏 𝑠𝑐
− 𝐶 𝑏 𝑔𝑢𝑖𝑑
R is a 1x1 control weighting matrix
Q is a 4x4 error penalty matrix
Gain value K minimizes cost functional 𝐽 = 0
∞
(𝒙 𝑡 𝑄𝒙 + 𝑢 𝑡 𝑅𝑢) 𝑑𝑡
26
27. Tracker with Constant Drag Bias Error of 1.5
Actuator running .12% of the time or 1.2 seconds for every 1000
seconds of orbit
Simulations assume 4 minutes needed to go from Cbmin to Cbmax
27
28. Tracking Simulation Results with Drag
Errors
To simulate drag error, density multiplied
by error factor consisting of
Bias error
Period errors at frequencies of
26 days (synodic period)
1 day (Earth’s solar day)
5400 seconds (roughly one orbital period)
Noise free tracking example
28
29. Discrete Time Extended Kalman Filter for LQR Guidance Trajectory
Tracking
State will be relative position and velocity
Measurement 𝑧 = relative position and velocity derived from GPS measurement and guidance state
𝑧𝑖 = 𝐺𝑥𝑖, 𝑥𝑖 ≈ Φ𝑖 𝑥𝑖−1
𝐺 = 𝐼 4𝑥4, Φ𝑖 = 𝑒(𝐴−𝐵𝐾)𝑡
𝑊 = measurement noise covariance
𝑄 = Process noise covariance
Λ = Fading term
f represents numerical propagation from ti to ti-1
𝑥1
−
and 𝑃𝑖
−
are a-priori state and state error covariance estimates
𝑥𝑖
−
= 𝑓(𝑡𝑖, 𝑡𝑖−1, 𝑥𝑖−1
+
)
𝑃𝑖
−
= Φ𝑖 𝑃𝑖−1
+
Φ𝑖
𝑇
+ 𝑄
𝑆 = 𝐺𝑃𝑖
−
𝐺 𝑇 + 𝑊
𝐾𝑖 = 𝑃𝑖 𝐺 𝑇
𝑆 −1
𝑥𝑖
+
= 𝑥𝑖
−
+ 𝐾𝑖 𝑧𝑖 − 𝐺𝑥𝑖
−
𝑃𝑖
+
= 𝐼 − 𝐾𝑖 𝐺 𝑃𝑖Λ
29
30. Kalman Filter with Measurement Noise, Bias, and Density Error
Motor runs 3.5% of the time assuming 240 seconds for full
deployment
5% actuator deadband
Truly a “worst case”
Tracking to 90 km altitude
30
31. Monte Carlo Simulations
Variable Range Probability Distribution
Semi Major Axis [6698, 6718] km Uniform
True Anomaly [0, 360] degrees Uniform
Eccentricity [0, .004] Uniform
Right Ascension [0, 360] degrees Uniform
Argument of the Periapsis [0, 360] degrees Uniform
Inclination [1, 97] degrees Uniform
Impact Latitude [0, max(inclination, 180-
inclination)-. 1] degrees
Uniform
Impact Longitude [-180, 180] degrees Uniform
Cbmax [.033, .067] Uniform
Cbmin [.0053, .027] Uniform
epoch [11/1/2003, 11/1/2014] Uniform
31
1,000 guidance trajectory generation and tracking simulations were conducted for the randomly varying
simulation parameters in the table above.
32. Monte Carlo Simulations
1000 guidance and tracking simulations with
randomized initial conditions run to 120 km geodetic
altitude
997 guidance cases below 25 km error threshold, all
guidance errors below 106 km
All tracking errors below 6 km
32
33. Trajectory simulated to ground with terminal ballistic coefficient
and initial mean orbital elements: 𝑎 = 6418, 𝑒 = 0, 𝑖 = 𝑖0, 𝜔 =
0, 𝑢 = 0, Ω = 0
Argument of latitude increase Δ𝑢 needed to achieve target
latitude calculated
RAAN increase ΔΩ equal to longitude difference between target
location and point beneath the orbital plane at desired 𝑢
Initial conditions updated with Δ𝑢 and ΔΩ and trajectory
propagated again
May update with smaller percentage of Δ𝑢 to aid
convergence
When initial conditions found that lead to desired impact
location, DEO point is the trajectory state at 100 km geocentric
altitude
Target DEO point with guidance trajectory generation algorithm
and adjust final 𝐶 𝑏 to ensure landing on ground target
Ω decrease,
𝑢 increase
Target
Actual
impact
Initial conditions
Propagate to
ground
Compute Δ𝑢 and ΔΩ
needed for desired
ground impact Extract DEO point if
errors below threshold
Update initial
conditions with
Δ𝑢 and ΔΩ
33
De-Orbit Point Point Selection Algorithm
35. Spacecraft Design CAD Model
ISIS Turnstile antenna system
SkyFox piPATCH GPS antenna
SkyFox piNav-NG GPS
Clyde Space CPUT UTRX
half duplex radio
D3 control board
Clyde Space 20 Whr battery
D3 system
D3 adapter stage
Clyde Space standard 1U structure
DHV Technologies solar cell
Clyde Space 3rd generation 1U
EPS
35
Magnetorquer Brackets
38. Ongoing/Future Work
Targeting a ground impact point using
aerodynamic drag
Desired re-entry point at 100 km altitude
selected through iterative simulations to ground
D3 flexible body attitude dynamics analysis
Implementation of ground targeting
algorithm on MISTRAL and IPERDRONE
satellites using IRENE heat shield as drag
device
Completion and launch of D3 CubeSat
Flight of D3 unit on Ames TechEdSat
CubeSat
38
39. Conclusions
Targeting a re-entry point using solely aerodynamic drag is possible
Guidance trajectory generation
Guidance trajectory tracking
Navigation
Algorithm reliability demonstrated via Monte Carlo simulations and case
studies
Have developed a retractable Drag De-orbit Device (D3) to provide
attitude stability and facilitate aerodynamic re-entry point targeting
D3 CubeSat in development at UF to test drag device and algorithms
Collaborations with NASAAmes and the Italian Aerospace Research
Centre (CIRA) to use hardware and algorithms for re-entry point and
ground targeting
39
43. Drag Devices
IRENE [1]
In development
Planned flights on MISTRAL and
IPERDRONE satellites
Can modulate area
Serves as re-entry heat shield
Cylde Space AEOLDOS module [2]
No sale option seen
Single deploy
ExoBrake [3]
Multiple flights on CubeSats
Goal is to perform guided re-entry
iDod [4]
Inflatable device, not retractable
CubeSail [5]
Single-deploy sail used for drag and solar
sailing
Freedom Drag Sail [6]
Built by JAXA
Successful ISS launch and de-orbit
43
44. Targeted Re-Entry Algorithms
Dutta’s Algorithm [3]
Orbit initial conditions and satellite’s ballistic coefficient range provided to Post2
numerical optimizer
Post2 calculates 7 discrete Cb changes to get satellite close to desired de-orbit
location
Undesirable for onboard use
Computationally intensive and no convergence guarantee
Virgili’s Algorithm [7]
Analytical solution used to compute guidance directly without any orbit
propagation
Assumes circular orbit, exponential density, J2 as only perturbation
Guidance trajectories are not very accurate
Patera’s Algorithm [8]
Change ballistic coefficient during final de-orbit phase to reduce casualty risk
44
45. Contributions
Aerodynamic re-entry point targeting algorithms
Guidance trajectory generation algorithm capable of running onboard a small
satellite
Uses high fidelity force model
Robust tracking algorithm to follow guidance trajectory
Robust and reliable drag De-Orbit Device (D3) hardware implementation
Capable of changing Cb while providing attitude stability
Will be used to perform re-entry point targeting algorithms on-orbit
45
46. Old PADDLES Drag Device
Origami drag sail
Rotating shaft unfurls sail
Design issues
Does not provide passive
aerodynamic attitude stability
Difficult to determine exact sail
geometry
Deployment mechanism
sometimes snags
46
48. Attitude Simulation Framework
Numerically integrate equations of motion in state space
form
Simultaneous attitude and orbit propagation in ECI (Earth
Centered Inertial) frame
𝒙 = 𝒓 𝑻
𝒗 𝑻
𝒒 𝑻
𝝎 𝑻 𝑇
𝒙 = 𝒗 𝑻 𝒗 𝑻 𝒒 𝑻 𝝎 𝑻 𝑇
𝒒 is quaternion that describes rotation from ECI frame to
body frame. Rotation of ECI frame about 𝒆 by 𝜃 yields
body frame.
𝒒 =
𝒆 sin 𝜃/2
cos 𝜃/2
= 𝑞1 𝑞2 𝑞3 𝑞4
𝑇
𝒒 =
1
2
𝑞4 −𝑞3 𝑞2 𝑞1
𝑞3 𝑞4 −𝑞1 𝑞2
−𝑞2 𝑞1 𝑞4 𝑞3
−𝑞1 −𝑞2 −𝑞3 𝑞4
𝜔 𝑥
𝜔 𝑦
𝜔 𝑧
0
𝝎 = 𝑰−𝟏 𝑻 𝒏𝒆𝒕 − 𝝎 × 𝑰𝝎
𝒗 =
𝑭 𝒏𝒆𝒕
𝑚 48
Variable Definition
𝒙 State vector
𝒓 ECI position
𝒗 ECI velocity
𝒒 Attitude
quaternion
𝝎 Angular velocity
𝑰 Sat moment of
inertia tensor
𝑻 𝒏𝒆𝒕 Net torque
𝑭 𝒏𝒆𝒕 Net force
𝑚 Satellite mass
49. Magnetic Torques
Running a current through a wire generates a magnetic
field
A loop of wire, also known as a solenoid or
magnetorquer, impersonates a bar magnetic with
magnetic moment 𝝁
𝝁 = 𝐼𝐴𝑛 𝒏 (𝐴𝑚𝑝 ∗ 𝑚𝑒𝑡𝑒𝑟2)
𝐼 is current in wire, 𝐴 is loop area, 𝑛 is number of loops in
magnetorquer, 𝒏 is magnetorquer normal unit vector
Including a ferromagnetic material inside the loop increases
the magnetic moment
A dipole magnet wants to align with the ambient
magnetic field B (Tesla)
Alignment torque is
𝑻 = 𝝁 × 𝑩 (𝑁𝑒𝑤𝑡𝑜𝑛 ∗ 𝑚𝑒𝑡𝑒𝑟)
49
50. Gravity Gradient Torques
Caused by differing values of gravitational attraction on different parts of
orbiting object
𝑀 𝑥 =
3𝜇 𝑒 𝑅 𝑦 𝑅 𝑧
𝑅5
𝐼𝑧𝑧 − 𝐼 𝑦𝑦
𝑀 𝑦 =
3𝜇 𝑒 𝑅 𝑥 𝑅 𝑧
𝑅5
𝐼 𝑥𝑥 − 𝐼𝑧𝑧
𝑀𝑧 =
3𝜇 𝑒 𝑅 𝑥 𝑅 𝑦
𝑅5
𝐼 𝑦𝑦 − 𝐼 𝑥𝑥
𝑴 =
3𝜇 𝑒
𝑅3
𝑹 × 𝑰 𝑹
𝑴 = 𝑀 𝑥 𝑀 𝑦 𝑀𝑧
𝑇 is the gravity gradient torque in the satellite body
frame (Newton*meter)
𝑹 = 𝑅 𝑥 𝑅 𝑦 𝑅 𝑧
𝑇 is the vector from the center of the Earth to the
satellite center of mass expressed in the satellite body frame (km). 𝑅 = 𝑹 .
𝑰 is satellite moment of inertia tensor
R
𝜃
50
51. Aerodynamic Torques
Mean free path between particles is large, so fluid assumption not
valid
Panel methods can be used to compute the aerodynamic force and
torque acting on each panel and sum to get total effect on satellite
Can assume specular reflection for simplicity where each particle collides
elastically with panel (kinetic energy conserved) and bounces of at the same
angle
Drag force on each panel is 𝑭 𝒅 = −2𝜌𝐴𝑣⊥ 𝒗⊥ (𝑁𝑒𝑤𝑡𝑜𝑛𝑠)
Force acts at geometric center of panel, normal to panel
Torque generated by this force is 𝑻 𝒅 = 𝒓 × 𝑭 𝒅 (𝑁𝑒𝑤𝑡𝑜𝑛 ∗ 𝑚𝑒𝑡𝑒𝑟)
𝒓 from satellite center of mass to panel center of pressure
𝒗∞
𝑭 𝒅
𝒗⊥
51
52. Attitude Stabilization Procedure
Desired attitude: z-axis ram aligned, x-axis zenith
pointing
Run BDot de-tumble controller until satellite angular
velocity below specified threshold
Partially deploy all booms to 1 m and continue BDot
with fixed magnetic moment vector superimposed
along satellite x-axis
Causes x-axis to pull toward Earth magnetic field vector
When magnetic field most zenith pointing, fully
deploy +x and –x booms, deploy +y and –y booms
half way, and remove fixed magnetic moment vector
Orbit propagated in advance to determine when magnetic
field is most zenith pointing
Continue BDot until attitude stabilizes with
oscillations damped
x
y
z
52
53. Simulation Setup
2U, 2 kg, D3 equipped CubeSat simulated in
ISS orbit
52 degree inclination, 400 km circular orbit
To compute appropriate time for final boom
deployment, primary simulation paused and
forecast made of magnetic field over the next
orbit
Boom deployment when magnetic field most
zenith pointing for proper gravity gradient
stabilization
Final attitude displayed with respect to Local
Vertical Local Horizontal (LVLH) frame
x-axis aligned with zenith vector, z-axis aligned
with orbit angular momentum
53
54. Simulation Results
BDot running until 𝑡 = 10,000 𝑠 with
gain of -5
All booms deployed to 1 𝑚 and Bdot
continues running with the addition of
fixed magnetic moment vector
.015 0 0 𝑇 𝐴𝑚2 until 𝑡 = 20,000 𝑠
(5.56 hrs)
Magnetic field forecast for one orbit.
Magnetic field will be most zenith
pointing at 𝑡 = 20,800 𝑠
At 𝑡 = 20,800 𝑠, +𝑦 and −𝑦 booms
deployed to 1.85 m and +𝑥 and −𝑥 booms
deployed to 3.7 m. Fixed magnetic
moment removed
BDot continues for remainder of
simulation and boom deployment does not
change
54
57. Aerodynamic Model
Aerodynamic drag acceleration given by
𝒂 𝒅 = −
1
2𝑚
𝐶 𝑑 𝐴𝜌𝑣∞ 𝒗∞ = −𝐶 𝑏 𝜌𝑣∞ 𝒗∞
𝐶 𝑏 =
𝐶 𝑑 𝐴
2𝑚
𝒗∞ = 𝒗 − 𝝎 𝒆 × 𝒓
Density and drag coefficient most difficult to
predict
Factor of 100 difference between min and max
density
High fidelity density model critical
Drag coefficient will vary between two for a
sphere and four for a flat plate assuming
specular reflection
57
𝒗∞
𝑭 𝒅
𝒗⊥
61. Proving the Analytical Relation Between Cb changes and Orbit Behavior
Gauss variation of parameters for semi major axis is
𝑑𝑎
𝑑𝑡
=
2
𝑛 1 − 𝑒2
𝑒 sin 𝜃 𝐹𝑅 +
𝑝
𝑟
𝐹𝑠
For a circular orbit around a spherical Earth, 𝑒 = 0, 𝑝 = 𝑎 1 − 𝑒2
= 𝑟, and 𝐹𝑠 = 𝑎 𝑑
𝑑𝑎
𝑑𝑡
=
2𝑎 𝑑
𝑛
𝑛 =
𝜇
𝑎3
In a non-rotating atmosphere
𝑎 𝑑 = −𝐶 𝑏 𝜌𝑣2
𝐶 𝑏 =
𝐶 𝑑 𝐴
2𝑚
Making some substitutions
−
𝜇
𝑎3
1
2𝐶 𝑏 𝜌𝑣2
𝑑𝑎 = 𝑑𝑡
61
62. Proof Continued
𝑣 =
𝜇
𝑎
Substituting into last equation on previous page yields
−
𝑑𝑎
2 𝜇𝑎𝐶 𝑏 𝜌
= 𝑑𝑡
Integrate both sides of equation, assume density is a function of semi major
axis, and multiply by 𝐶 𝑏
∆𝑡𝐶 𝑏 =
𝑎0
𝑎 𝑓
−
𝑑𝑎
2 𝜇𝑎𝜌
∆𝑡𝐶 𝑏 = 𝐺 𝑎 𝑓 − 𝐺(𝑎0)
∆𝑡𝐶 𝑏 is constant for fixed initial and final 𝑎 so
∆𝑡2 =
𝐶 𝑏1∆𝑡1
𝐶 𝑏2
62
63. Proof Continued
In a circular orbit
𝑛 =
𝑑𝜃
𝑑𝑡
=
𝑑𝑢
𝑑𝑡
𝑑𝑎
𝑑𝑡
= −2 𝜇𝑎𝐶 𝑏 𝜌
Multiply 𝑛 =
𝑑𝑢
𝑑𝑡
by inverse of
𝑑𝑎
𝑑𝑡
and substitute formula for 𝑛
𝑑𝑢
𝑑𝑎
= −
𝑛
2 𝜇𝑎𝐶 𝑏 𝜌
= −
1
2𝑎2 𝐶 𝑏 𝜌
Multiply both sides by 𝐶 𝑏 𝑑𝑎 and integrate
∆𝑢𝐶 𝑏 =
𝑎0
𝑎 𝑓 𝑑𝑎
2𝑎2 𝜌
= 𝑃 𝑎 𝑓 − 𝑃 𝑎0
∆𝑢𝐶 𝑏 is constant for fixed initial and final 𝑎 so
∆𝑢2=
𝐶 𝑏1∆𝑢1
𝐶 𝑏2
63
64. Control Parameters Formula Proof
Subscript 1 indicates parameter between 𝑡0 and 𝑡 𝑠𝑤𝑎𝑝. Subscript 2 indicates parameter between 𝑡 𝑠𝑤𝑎𝑝
and 𝑡𝑡𝑒𝑟𝑚. Subscript 0 indicates parameter from old numerically propagated trajectory
If swap point occurs at same semi major axis in old and new trajectories, the following relations hold
∆𝑢1 + ∆𝑢2 = ∆𝑢 𝑡
∆𝑡1 + ∆𝑡2 = ∆𝑡𝑡
∆𝑢1 =
∆𝑢10 𝐶 𝑏10
𝐶 𝑏1
∆𝑢2 =
∆𝑢20 𝐶 𝑏20
𝐶 𝑏2
∆𝑡1 =
∆𝑡10 𝐶 𝑏10
𝐶 𝑏1
∆𝑡2 =
∆𝑡20 𝐶 𝑏20
𝐶 𝑏2
64
66. Single Stage Targeting Algorithm Used on TES6 ExoBrake
Increase in argument of latitude Δu needed to de-orbit at target latitude
calculated
Any Δ𝑢 + 2𝑛𝜋 also yields proper latitude if 𝑛 is an integer
Change 𝐶 𝑏1 to achieve desired total 𝑢 = 𝑢0 + Δ𝑢
𝐶 𝑏1 𝑛𝑒𝑤
= 𝐶 𝑏1
𝑢0
𝑢0+Δ𝑢
Ex: 𝑢0 = 100 𝑟𝑎𝑑, Δ𝑢 = 1 𝑟𝑎𝑑, 𝐶 𝑏1 = 10, 𝐶 𝑏1 𝑛𝑒𝑤
= 10 ∗
100
101
= 9.9
Can analytically estimate new de-orbit time and change in orbital elements at
de-orbit point using this 𝐶 𝑏 value and the analytical relations
Test all candidate Δ𝑢 values and select the one that yields lowest total error
Propagate trajectory with corresponding 𝐶 𝑏1 value
Repeat estimation process until trajectory found that converges to desired
location
66
67. Analytical Drag-Work Calculation
Work done by drag identical between old and new
trajectories at end of each phase
Same final orbital energy
Phase start and end times from analytical theory
For new trajectory at time Δ𝑡 beyond phase endpoint
with some 𝐶 𝑏, the time after the phase start in the initial
trajectory (Δ𝑡0) with the same energy state is
Δ𝑡0 =
𝐶 𝑏
𝐶 𝑏0
Δ𝑡
Desired work done by drag 𝑊𝑑 𝑑𝑒𝑠
is equal to work
done by drag in old trajectory at Δ𝑡0 beyond phase start
𝐶 𝑏0 = old trajectory ballistic coefficient during phase
Note that old trajectory has been numerically propagated
and 𝑊𝑑 is known at all points
67
68. Sensitivity Analysis
Investigates effects of changes in drag profile on de-orbit location
An error in estimated drag force of .25% puts satellite on the other side of Earth
Closed loop control necessary
Changes in density profile between numerical propagations can cause analytical and numerical solutions to
diverge
System less sensitive when orbit life is shorter
300 km circular orbit with 1976 standard atmosphere,
Cb10=.025, Cb20=.01, ts0=150,000 s
270 km circular orbit with 1976 standard atmosphere,
Cb10=.025, Cb20=.01, ts0=150,000 s
68
70. Q and R Selection Strategy
𝑄 =
0 0 0 0
0 1 0 0
0
0
0
0
0
0
0
0
With this Q matrix, system only cares about minimizing error in the along-track direction
Minimizing along track error requires also minimizing radial error as well as error rates in the long run
Temporary increases in radial error and error rates allowed if they lead to faster minimization of along-track error
1 by 4 gain matrix K0 first calculated with R0 = 10000
𝛿𝑦𝑠𝑎𝑡 specified such that controller guaranteed to saturate at 𝛿𝑦 = 𝛿𝑦𝑠𝑎𝑡
Calculate second element in K required to ensure saturation at 𝛿𝑦𝑠𝑎𝑡
𝐾2 𝑟𝑒𝑞 =
𝐶 𝑏𝑚𝑎𝑥 − 𝐶 𝑏𝑚𝑖𝑛
𝛿𝑦𝑠𝑎𝑡
R required to achieve this gain given based on 𝑅0 and second element of K0 𝐾0(2)
𝑅 =
𝐾0 2
𝐾2 𝑟𝑒𝑞
2
𝑅0
LQR gain K re-evaluated with new R
Gain periodically updated as density changes
70
72. Noise Model
Noise based on PiNav-NG GPS Unit
Periodic bias error and Gaussian random noise on
position and velocity measurements
Position error = [.001; -.005; .002]*sin(2*pi*time/5400) +
randn(3,1)*.005 km
Velocity error = [.00005; -.00005;
.00005/2]*sin(2*pi*time/5400) +randn(3,1)*.00005 km/s
72
Error estimate from
manufacturer
73. Simulation with Navigation Errors
Navigation errors due to measurement noise have significant effects if
not filtered
Actuator running 84% of the time
73
74. Kalman Filtering for Measurement Noise Cancellation
Kalman filter implemented to remove measurement noise
Motor running .43% of the time
74
75. Re-Entry Safety Analysis
Effects of GNC errors and system uncertainties on ground impact point
characterized via Monte Carlo simulations and Inputs’ Statistics method
Safety box cross-track dimension only affected by cross-track GNC
errors
Remaining variables affect the safety box along-track dimension
75
78. D3 Board Components and External Connections
External Wiring
USB Power and Data Line
Ribbon cable containing
16 wires for 4 motors and 4 encoders
2 wires for ground and 5V for all encoders
6 wires for 3 magnetorquer circuits
Components
2 quad half h-bridge motor driver chips (SN754410)
2 quad half h-bridge chips to drive 3 magnetorquers
Can use 3 low voltage DRV8837DSGR chips if using 3.3 V bus for
magnetorquers
Possibly 1 IMU chip (TDK ICM-20948)
Beagle Bone Black Indsutrial high performance processor
1 Watchdog timer
1 30-minute countdown timer
RBF Pin
Antenna deployment controller
78
79. Desired Processor
Beaglebone Black Industrial Version
Significant flight heritage
1 Watt power use in low power
mode
May be possible to use less power if
HDMI and ethernet ports removed
Fits on D3 control board
79
80. CubeSat Mission Requirements
Success Level and Description Demonstration Verification Criteria
Required: D3 CubeSat ejects from
deployer
Prerequisite. Track CubeSat with radar.
Confirmation of launch from vehicle.
Required: Ground systems make
contact with CubeSat
Prerequisite. Make radio contact.
D3 booms are used to change the cross-
wind area of the CubeSat
Boom can operate in LEO. Commanded motor position
telemetry.
Track CubeSat with radar and look
for drag changes.
D3 stabilizes attitude of CubeSat Booms and magnetorquers can be used
to stabilize attitude in LEO.
Commanded motor position
telemetry.
Magnetometer telemetry.
D3 device is used to actuate a desired
maneuver
D3 can be used to actuate a desired
maneuver.
JSpOC radar data and CubeSat GPS data.
D3 device is used to deorbit within a
desired interval
Ability of D3 to deorbit a CubeSat as
desired.
JSpOC radar data and CubeSat GPS data.
Maximum: d3 deorbits to within
1300km of a desired target interface
point at 90km altitude.
Ability of D3 to deorbit the CubeSat to a
safe location.
Track CubeSat with radar.
80
81. Hardware Configuration
Component Mass (g) Avg Power User (mW) Cost (USD)
Clyde Space 3rd Generation EPS 86 160 4900
Clyde Space 20 WHr Battery 256 0 2700
Clyde Space CPUT UTRX Half Duplex Radio 90 250 RX, 4000TX, 333 avg with 30 min daily TX 8850
ISISpace Turnstile Antenna System 30 0 6891
D3 Deployers 1100 200 avg (20% duty cycle), 16400 peak 2000
D3 Magnetorquers 101 Variable, max 1000 during de-tumble 100
Beaglebone Black Master CubeSat and D3
Micro-controller
24 1000 100
DHV Technologies Custom Solar Panels
(four 2U side panels, one 1U top panel)
400 total 4240 max gen. for 2U panels and 2120 max. gen.
for 1U panel
26000
1U Clyde Space Structure 200 0 3550
D3 Adapter Stage 200 0 200
SkyFox piNav-NG GPS Unit 100 139 9624
SkyFox piPATCH GPS Antenna 25 100 2238
Totals 2612 1932 average continuous use 67153
81
82. Robot Operating System (ROS) Implementation
• ROS allows multiple
software “Nodes” to run
simultaneously on
BeagleBone Black
• Nodes communicate
with each other via data
publications to “topics”
• Nodes written in c++
with speed or hardware
interfacing needed
• Python otherwise
82
83. Telemetry and Commands List
Ground Station Commands Spacecraft Telemetry
Reset microcontroller Battery voltage
Update software Solar panel voltages
Update F10.7 and Ap solar and geomagnetic indices
for density forecasting
Boom deployment levels
Target desired de-orbit location GPS position and velocity estimates
Change operation mode (normal, debug, bare-bones) Magnetometer readings
Manual boom deployment profile Motor and magnetorquer usage history
Request telemetry Any relevant error codes
Guidance update Current guidance trajectory
LQR and BDOT gain updates Temperatures
83
85. Operations Flowchart-from ConOps
Provides a plan for the sequence of spacecraft operations necessary to
meet the mission requirements
85
Uplink desired
de-orbit
location and
updated F10.7
and Ap indices
86. References
[1] R. Fortezza, R. Savino, and G. Russo, “MISTRAL (AIR-LAUNCHEABLE MICRO-SA℡LITE WITH
REENTRY CAPABILITY) A small spacecraft to carry out several missions in LEO,” 2013.
[2] P. Harkness, M. McRobb, P. Lützkendorf, R. Milligan, A. Feeney, and C. Clark, “Development status of
AEOLDOS – A deorbit module for small satellites,” Advances in Space Research, vol. 54, no. 1, pp. 82–91, Jul. 2014.
[3] S. Dutta, A. Bowes, A. M. Dwyer Cianciolo, C. Glass, and R. W. Powell, “Guidance Scheme for Modulation
of Drag Devices to Enable Return from Low Earth Orbit,” in AIAA Atmospheric Flight Mechanics Conference, 2017,
p. 0467.
[4] D. C. Maessen, E. D. van Breukelen, B. T. C. Zandbergen, and O. K. Bergsma, “Development of a Generic
Inflatable De-Orbit Device for CubeSats.” .
[5] N. Adeli and V. Lappas, “Deployment System for CubeSail nano-Solar Sail Mission,” presented at the
SmallSat 2010, Logan, UT.
[6] “Re-Entry: FREEDOM Drag Sail CubeSat – Spaceflight101.” .
[7] J. Virgili and P. Roberts, “Atmospheric Interface Reentry Point Targeting Using Aerodynamic Drag
Control,” Journal of Guidance, Control, and Dynamics, vol. 38, no. 3, pp. 1–11, 2015.
[8] R. Patera, “Drag Modulation as a Means of Mitigating Casualty Risk for Random Reentry,” in AIAA
Atmospheric Flight Mechanics Conference and Exhibit, American Institute of Aeronautics and Astronautics.
[9] F. Marcos, B. Bowman, and R. Sheehan, “Accuracy of Earth’s thermospheric neutral density models,” in
Proceedings of the AIAA/AAS Astrodynamics Specialist Conference, Keystone, CO, 2006.
[10] W. L. Hankey, Re-Entry Aerodynamics. American Institute of Aeronautics and Astronautics, 1988.
[11] A. W. Koenig, T. Guffanti, and S. D’Amico, “New State Transition Matrices for Spacecraft Relative Motion
in Perturbed Orbits,” Journal of Guidance, Control, and Dynamics, 2017.
86