The Gravity Perturbed Hohmann Transfer

Orbital Mechanics with MATLAB
The Gravity Perturbed Hohmann Transfer
This document is the user’s manual for a MATLAB script named phohmann which can be used to solve
the gravity perturbed Hohmann transfer between coplanar and non-coplanar circular Earth orbits. The
algorithm starts with a two-body Hohmann transfer initial guess and then uses the SNOPT nonlinear
programming (NLP) method to determine the optimum two impulse orbit transfer subject to non-
spherical Earth gravity perturbations. Appendix A summarizes the governing equations for the two-
body Hohmann transfer. It also includes information about the MATLAB implementation used to solve
this classic astrodynamics problem. This script is valid for “exterior” Hohmann transfers from a lower
altitude circular orbit to a higher altitude circular orbit.
The phohmann script uses modified equinoctial orbital elements to solve the gravity perturbed orbit
transfer “targeting” problem. Additional information about these orbital elements can be found in
Appendix B. That appendix also explains how to use components and combinations of these non-
singular elements to calculate a variety of final orbital element targets or boundary conditions.
MATLAB versions of SNOPT for several computer platforms can be found at Professor Philip Gill’s
web site which is located at http://scicomp.ucsd.edu/~peg/. Professor Gill’s web site also includes a
PDF version of the SNOPT software user’s guide. A brief introduction to nonlinear programming can
be found in Appendix C.
User interaction with the script
The phohmann MATLAB script will interactively prompt the user for the name of the simulation
definition input data file. This prompt is similar to the following;
The file type defaults to names with a *.in filename extension. However, you can select any
phohmann compatible ASCII data file by selecting the Files of type: field or by typing the name of the
file directly in the File name: field.
page 1
Input file format and contents
The phohmann software is “data-driven” by a user-created text file. This text file should be simple
ASCII format with no special characters.
The following is a typical input file used by this MATLAB script. In the following discussion the actual
input file contents are in courier font and all explanations are in times italic font. This example is a
Hohmann transfer from a low Earth orbit (LEO) to a geosynchronous orbit (GSO). In this data file, user
provided inputs are in bold font.
Each data item within an input file is preceded by one or more lines of annotation text. Do not delete
any of these annotation lines or increase or decrease the number of lines reserved for each comment.
However, you may change them to reflect your own explanation. The annotation line also includes the
correct units and when appropriate, the valid range of the input. ASCII text input is not case sensitive
but must be spelled correctly.
The first five lines of any input file are reserved for user comments. These lines are ignored by the
software. However the input file must begin with five and only five initial text lines.
********************************************
* input data file for phohmann MATLAB script
* impulsive LEO-to-GSO orbital transfer
* filename ==> leo2gso.in
********************************************
The first inputs to the program define the initial UTC calendar date and time for the simulation. The
data for the calendar year should include all four digits. The calendar date and time are required in
order to correctly calculate the tesseral or longitude-dependent components of the Earth’s gravity.
initial calendar date

(1 <= month <= 12, 1 <= day <= 31, year = all digits!)
------------------------------------------------------
2, 25, 2013
initial UTC
(0 <= hours <= 24, 0 <= minutes <= 60, 0 <= seconds <= 60)
----------------------------------------------------------
20, 18, 33.0
The next three inputs are the altitude, orbital inclination and right ascension of the ascending node
(RAAN) of the initial circular orbit.
*************
initial orbit
*************
altitude (kilometers; altitude > 0)

-----------------------------------
185.32
orbital inclination (degrees; 0 <= inclination <= 180)

------------------------------------------------------
28.5
right ascension of the ascending node (degrees; 0 <= raan <= 360)
-----------------------------------------------------------------
100.0
page 2
The next two inputs are the altitude and orbital inclination of the final circular orbit.
***********
final orbit
***********
altitude (kilometers; altitude > 0)

-----------------------------------
35788.0955
orbital inclination (degrees; 0 <= inclination <= 180)

------------------------------------------------------
0.0
The Earth gravitational constant and radius are user-defined by the next two inputs.
****************************************
astrodynamic constants and gravity model
****************************************
central body gravitational constant (km^3/sec^2)

------------------------------------------------
398600.4415
central body radius (kilometers)

--------------------------------
6378.14
Finally, the name of the Earth gravity model to use in the simulation and the order and degree of this
model are set by the following three inputs.
name of Earth gravity model data file
-------------------------------------
egm96.dat
order of the gravity model (zonals)

-----------------------------------
4
degree of the gravity model (tesserals)

---------------------------------------
4
The following is the phohmann solution for this example. The first part of the display is the two-body
Hohmann transfer solution. The second section summarizes the SNOPT iterations and summary. The
final section is the perturbed Hohmann transfer solution found during the optimization.
two-body Hohmann transfer solution
initial orbit altitude 185.3200 kilometers
initial orbit inclination 28.5000 degrees
initial orbit velocity 7792.9603 meters/second
final orbit altitude 35788.0955 kilometers
final orbit inclination 0.0000 degrees
final orbit velocity 3074.5848 meters/second
first inclination change 2.1645 degrees
page 3
second inclination change 26.3355 degrees
total inclination change 28.5000 degrees
first delta-v 2481.9821 meters/second
second delta-v 1790.3423 meters/second
total delta-v 4272.3244 meters/second
transfer orbit semimajor axis 24364.8478 kilometers
transfer orbit eccentricity 0.73061765
transfer orbit perigee velocity 10251.8686 meters/second
transfer orbit apogee velocity 1595.7727 meters/second
transfer orbit coast time 18924.5926 seconds

315.4099 minutes
5.2568 hours
orbital elements and state vector of the initial orbit

------------------------------------------------------
sma (km) eccentricity inclination (deg) argper (deg)

+6.56346000000000e+03 +0.00000000000000e+00 +2.85000000000000e+01 +0.00000000000000e+00
raan (deg) true anomaly (deg) arglat (deg) period (min)

+1.00000000000000e+02 +0.00000000000000e+00 +0.00000000000000e+00 +8.81980522880484e+01
rx (km) ry (km) rz (km) rmag (km)

-1.13973286818979e+03 +6.46374629458551e+03 +0.00000000000000e+00 +6.56346000000000e+03
vx (kps) vy (kps) vz (kps) vmag (kps)

-6.74454148515196e+00 -1.18924463633983e+00 +3.71847929670569e+00 +7.79296034444086e+00
orbital elements and state vector of the transfer orbit after the first impulse
-------------------------------------------------------------------------------

+2.43648477500000e+04 +7.30617647713395e-01 +2.63354979218110e+01 +0.00000000000000e+00

+1.00000000000000e+02 +0.00000000000000e+00 +0.00000000000000e+00 +6.30819751739617e+02

-1.13973286818979e+03 +6.46374629458550e+03 +0.00000000000000e+00 +6.56346000000000e+03

-9.04826108087521e+00 -1.59545255705264e+00 +4.54800087376082e+00 +1.02518685807620e+01
orbital elements and state vector of the transfer orbit prior to second impulse
-------------------------------------------------------------------------------

+2.43648477500000e+04 +7.30617647713395e-01 +2.63354979218110e+01 +0.00000000000000e+00

+1.00000000000000e+02 +1.80000000000000e+02 +1.80000000000000e+02 +6.30819751739617e+02

+7.32208995364962e+03 -4.15256356357386e+04 +2.29083173533909e-12 +4.21662355000000e+04

+1.40842308946174e+00 +2.48342990924876e-01 -7.07927123702901e-01 +1.59577274464277e+00
page 4
orbital elements and state vector of the final orbit
----------------------------------------------------

+4.21662355000000e+04 +0.00000000000000e+00 +0.00000000000000e+00 +0.00000000000000e+00

+1.00000000000000e+02 +1.80000000000000e+02 +1.80000000000000e+02 +1.43617371924092e+03

+7.32208995364962e+03 -4.15256356357386e+04 +0.00000000000000e+00 +4.21662355000000e+04

+3.02787492676107e+00 +5.33896043798643e-01 -0.00000000000000e+00 +3.07458477809479e+00
initial delta-v vector, magnitude and steering angles

-----------------------------------------------------
x-component of delta-v -2303.719596 meters/second

y-component of delta-v -406.207921 meters/second
z-component of delta-v 829.521577 meters/second
delta-v magnitude 2481.982050 meters/second
pitch angle -0.000000 degrees

yaw angle -8.975045 degrees
final delta-v vector, magnitude and steering angles

---------------------------------------------------
x-component of delta-v 1619.451837 meters/second

y-component of delta-v 285.553053 meters/second

yaw angle 49.627319 degrees
Nonlinear constraints 5 Linear constraints 1

Nonlinear variables 6 Linear variables 0
Jacobian variables 6 Objective variables 6
Total constraints 6 Total variables 6
The user has defined 0 out of 36 first derivatives
Major Minors Step nCon Feasible Optimal MeritFunction nS Penalty

0 5 1 5.7E+01 8.9E-06 4.2723244E+00 2 r c
Major Minors Step nCon Feasible Optimal MeritFunction nS Penalty
0 6 1 5.7E+01 8.9E-06 4.2723244E+00 2 r c
1 2 1.0E+00 2 2.9E-01 1.3E-06 4.2749551E+00 1.3E-07 n r c
2 2 1.1E-01 3 2.6E-01 1.0E-06 4.2749548E+00 2 1.3E-07 s c
3 1 1.0E+00 4 6.7E-04 (2.4E-07) 4.2749546E+00 2 1.3E-07 c
4 1 1.0E+00 5 2.6E-05 (1.9E-10) 4.2749546E+00 2 1.3E-07 c
5 1 1.0E+00 6 (2.5E-11)(9.7E-13) 4.2749546E+00 2 1.3E-07 c
SNOPTA EXIT 0 -- finished successfully

SNOPTA INFO 1 -- optimality conditions satisfied
Problem name
No. of iterations 13 Objective value 4.2749546090E+00
No. of major iterations 5 Linear objective 0.0000000000E+00
Penalty parameter 1.316E-07 Nonlinear objective 4.2749546090E+00
No. of calls to funobj 98 No. of calls to funcon 98
Calls with modes 1,2 (known g) 6 Calls with modes 1,2 (known g) 6
Calls for forward differencing 6 Calls for forward differencing 6
Calls for central differencing 72 Calls for central differencing 72
No. of superbasics 2 No. of basic nonlinears 4
No. of degenerate steps 0 Percentage .00
page 5
Max x 1 2.3E+00 Max pi 5 2.5E+00
Max Primal infeas 0 0.0E+00 Max Dual infeas 5 2.9E-10
Nonlinear constraint violn 8.7E-11
gravity-perturbed Hohmann transfer solution
orbital elements and state vector of the initial orbit

------------------------------------------------------

+6.56346000000000e+03 +0.00000000000000e+00 +2.85000000000000e+01 +0.00000000000000e+00

+1.00000000000000e+02 +0.00000000000000e+00 +0.00000000000000e+00 +8.81980522880484e+01

-1.13973286818979e+03 +6.46374629458551e+03 +0.00000000000000e+00 +6.56346000000000e+03

-6.74454148515196e+00 -1.18924463633983e+00 +3.71847929670569e+00 +7.79296034444086e+00
orbital elements and state vector of the transfer orbit after the initial delta-v
---------------------------------------------------------------------------------

+2.44593744898382e+04 +7.31658863355877e-01 +2.63335418348091e+01 +3.59907068451479e+02

+1.00000000000000e+02 +9.29315485206550e-02 +0.00000000000000e+00 +6.34494336505482e+02

-1.13973286818979e+03 +6.46374629458551e+03 +0.00000000000000e+00 +6.56346000000000e+03

-9.05235339099006e+00 -1.58903790996246e+00 +4.54905377104437e+00 +1.02549516628076e+01
orbital elements and state vector of the transfer orbit prior to the final delta-v
----------------------------------------------------------------------------------

+2.43663822497766e+04 +7.30510103763810e-01 +2.63224626979287e+01 +4.49503355132814e-02

+9.99180887891227e+01 +1.79955049664487e+02 +1.80000000000000e+02 +6.30879346284528e+02

+7.26271659870086e+03 -4.15360610054494e+04 -4.54747350886464e-13 +4.21662355000000e+04

+1.40980173813377e+00 +2.43062653381369e-01 -7.07743262330976e-01 +1.59609552335964e+00
orbital elements and state vector of the final orbit

----------------------------------------------------

+4.21662354999999e+04 +2.56314888882674e-15 +1.01379591428343e-13 +0.00000000000000e+00

+0.00000000000000e+00 +2.79918088789123e+02 +2.79918088789123e+02 +1.43617371924091e+03

+7.26271659870086e+03 -4.15360610054494e+04 -4.54747350886465e-13 +4.21662355000000e+04

+3.02863510092978e+00 +5.29566788146922e-01 +5.44009282066327e-15 +3.07458477809479e+00
page 6
initial delta-v vector, magnitude and steering angles
-----------------------------------------------------
x-component of delta-v -2307.811906 meters/second

y-component of delta-v -399.793274 meters/second
pitch angle 0.162032 degrees

yaw angle -8.974637 degrees
final delta-v vector, magnitude and steering angles

---------------------------------------------------
x-component of delta-v 1618.833363 meters/second

y-component of delta-v 286.504135 meters/second

yaw angle 49.614537 degrees
total delta-v 4274.954609 meters/second
transfer time 18904.205381 seconds

315.070090 minutes
5.251168 hours
degree of gravity model 4

order of gravity model 4
The following is a brief summary of the information provided by this MATLAB script.
sma (km) = semimajor axis in kilometers
eccentricity = orbital eccentricity (non-dimensional)
inclination (deg) = orbital inclination in degrees
argper (deg) = argument of perigee in degrees
raan (deg) = right ascension of the ascending node in degrees
true anomaly (deg) = true anomaly in degrees
arglat (deg) = argument of latitude in degrees. The argument of latitude is the sum of
true anomaly and argument of perigee.
period (min) = orbital period in minutes
rx (km) = x-component of the position vector in kilometers
ry (km) = y-component of the position vector in kilometers
rz (km) = z-component of the position vector in kilometers
rmag (km) = scalar magnitude of the position vector in kilometers
vx (km/sec) = x-component of the velocity vector in kilometers per second
vy (km/sec) = y-component of the velocity vector in kilometers per second
vz (km/sec) = z-component of the velocity vector in kilometers per second
page 7
vmag (km/sec) = scalar magnitude of the velocity vector in kilometers per second
x-component of delta-v = ECI x-component of the impulsive delta-v maneuver in meters per second
y-component of delta-v = ECI y-component of the impulsive delta-v maneuver in meters per second
z-component of delta-v = ECI z-component of the impulsive delta-v maneuver in meters per second
delta-v magnitude = scalar magnitude of the impulsive delta-v maneuver in meters per second
transfer time = time interval between the two impulsive maneuvers in seconds, minutes and hours
Note that ECI implies an Earth-centered-inertial coordinate system.
Additional information about the data displayed by the optimization algorithm can be found in the
SNOPT user’s manual which is available at Professor Gill’s website which is located at
http://scicomp.ucsd.edu/~peg/.
The pitch and yaw angles for each impulsive maneuver are computed and displayed in a local-vertical-
local horizontal coordinate system. The following diagram illustrates the geometry of the pitch and yaw
angles in this system. In this figure, the radial direction is along the geocentric radius vector directed
away from the Earth, the tangential direction is tangent to the orbit in the direction of the orbital motion,
and the normal direction is along the angular momentum vector of the orbit. The pitch angle is positive
above the local horizontal plane formed by the tangential and normal directions, and the yaw angle is
positive in the direction of the angular momentum vector which is perpendicular to the orbit plane.
uR
  pitch
 = yaw

uT

uN
The phohmann script will also create a graphics display of the initial, transfer and final orbits. The
following is the graphics display for this example. The initial orbit trace is red, the transfer orbit is blue
and the final mission orbit is green. The dimensions are Earth radii (ER) and the plot is labeled with an
ECI coordinate system where green is the x-axis, red is the y-axis and blue is the z-axis. The location of
each impulse is marked with a small blue circle.
page 8
Trajectory image files are saved to disk in both encapsulated, color Postscript format and MATLAB fig
format. The disk file names are phohmann1.eps and phohmann1.fig. The interactive features of
MATLAB graphics allow the user to re-load and manipulate the fig version of the trajectory display.
These capabilities allow the user to interactively find the best viewpoint as well as verify basic three-
dimensional geometry of the orbital maneuver.
Technical discussion
In this MATLAB script, the orbital motion is modeled with respect to a true-of-date Earth-centered-
inertial (ECI) coordinate system. The origin of this system is the center of the Earth and the
fundamental plane is the Earth’s equator. The x-axis is aligned with the true-of-date Vernal Equinox,
the z-axis is aligned with the Earth’s spin axis, and the y-axis completes this orthogonal, right-handed
coordinate system.
Acceleration due to Earth gravity
This MATLAB script uses a spherical harmonic representation of the Earth’s geopotential function
given by
   0R 0   n R m
n n
  r, ,      Cn   Pn  u      Pn  u   Snm sin m  Cnm cos m 

r r n 1  r  r n 1 m1  r 
where  is the geocentric latitude,  is the geocentric east longitude and r  r  x 2  y 2  z 2 is the
geocentric distance. In this expression the S’s and C’s are unnormalized harmonic coefficients of the
geopotential, and the P’s are associated Legendre polynomials of degree n and order m with argument
u  sin  .
The software calculates the acceleration due to the Earth’s gravity field with a vector equation derived
from the gradient of the potential function expressed as
a g  r, t    r, t 
page 9
This acceleration vector is a combination of pure two-body or point mass gravity acceleration and the
gravitational acceleration due to higher order non-spherical terms in the Earth’s geopotential. In terms
of the Earth’s geopotential  , the inertial rectangular cartesian components of the spacecraft’s
acceleration vector are as follows
1 z   1 

x  x 2 y
 r  r r 2 x 2  y 2    x  y 2
 
 
1 z   1 

y   y 2 x
 r  r r 2 x 2  y 2    x  y 2
 
 
1  x2  y2   
z  z   
r r   r 2
 
 
The three partial derivatives of the geopotential with respect to r, ,  are given by
 1   N  R
n
        n  1   Cnm cos m  Snm sin m  Pnm  sin  

n
r r  r  n 2  r  m 0
  N R
n n
 
  r  n2  r 
 C
m 0
m
n cos m  Snm sin m   Pnm1  sin    m tan  Pnm  sin   
  N R
n n
 
  r  n2  r 
mS
m 0
m
n cos m  Cnm sin m  Pnm  sin  
where
R  radius of the Earth
r  geocentric distance
Snm , Cnm  harmonic coefficients
  geocentric latitude  sin 1  z r 
  longitude     g
  right ascension  tan 1  ry rx 
 g  right ascension of Greenwich
Right ascension is measured positive east of the vernal equinox, longitude is measured positive east of
Greenwich, and latitude is positive above the Earth’s equator and negative below.
For m  0 , the coefficients are called zonal terms, when m  n the coefficients are sectorial terms, and
for n  m  0 the coefficients are called tesseral terms.
The Legendre polynomials with argument sin  are computed using recursion relationships given by:
page 10
1
Pn0  sin     2n  1 sin  Pn01  sin     n  1 Pn02  sin   
n
Pnn  sin     2n  1 cos  Pnn11  sin   , m  0, m  n
Pnm  sin    Pnm2  sin     2n  1 cos  Pnm11  sin   , m  0, m  n
where the first few associated Legendre functions are given by
P00 sin    1, P10 sin    sin , P11 sin    cos 
and Pi j  0 for j  i .
The trigonometric arguments are determined from expansions given by
sin m  2 cos  sin  m  1   sin  m  2  
cos m  2 cos  cos  m  1   cos  m  2  
m tan    m  1 tan   tan 
The following are the first 14 lines of the 18 by 18 egm96.dat gravity model file included with this
script. Column 1 is the degree l, column 2 is the order m, column 3 is the C coefficients and the last
column contains the S gravity model coefficients.
2 0 -1.08262668355E-003 0.00000000000E+000
3 0 2.53265648533E-006 0.00000000000E+000
4 0 1.61962159137E-006 0.00000000000E+000
5 0 2.27296082869E-007 0.00000000000E+000
6 0 -5.40681239107E-007 0.00000000000E+000
7 0 3.52359908418E-007 0.00000000000E+000
8 0 2.04799466985E-007 0.00000000000E+000
9 0 1.20616967365E-007 0.00000000000E+000
10 0 2.41145438626E-007 0.00000000000E+000
11 0 -2.44402148325E-007 0.00000000000E+000
12 0 1.88626318279E-007 0.00000000000E+000
13 0 2.19788001661E-007 0.00000000000E+000
14 0 -1.30744533118E-007 0.00000000000E+000
Gravity model coefficients are often published in normalized form. The relationship between
normalized Cl ,m , Sl ,m and un-normalized gravity coefficients Cl ,m , Sl ,m is given by the following
expression:
 l  m ! Cl ,m 
12
Cl ,m   1
    
 Sl ,m    2   m 0  2l  1  l  m  !   Sl ,m 
where  m 0 is equal to 1 if m is zero and equal to zero if m is greater than zero.
page 11
The following is the MATLAB source code for the function that opens and reads a gravity model file
(fname) and creates matrices of the un-normalized coefficients.
function [ccoef, scoef] = readgm(fname)
% read gravity model data file
% input
% fname = name of gravity data file
% output
% ccoef, scoef = gravity model coefficients
% data file format (space delimited ascii)
% column 1 is the degree, column 2 is the order, column 3 are the C coefficients
% and the last column contains the S gravity model coefficients. For example,
% 2 0 -1.08262668355E-003 0.00000000000E+000
% 3 0 2.53265648533E-006 0.00000000000E+000
% 4 0 1.61962159137E-006 0.00000000000E+000
% Orbital Mechanics with MATLAB
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% read the data file
gdata = dlmread(fname);
nrows = size(gdata, 1);
% initialize coefficients
idim = gdata(nrows, 1) + 1;
ccoef = zeros(idim, idim);
scoef = zeros(idim, idim);
% create gravity model coefficients
for n = 1:nrows
i = gdata(n, 1);
j = gdata(n, 2);
ccoef(i + 1, j + 1) = gdata(n, 3);
scoef(i + 1, j + 1) = gdata(n, 4);
end
Solving the trajectory optimization problem
As mentioned earlier, the trajectory optimization uses the two-body Hohmann transfer solution for the
initial guess. For this problem, the components of the initial and final impulsive delta-v vector are
control variables. The transfer time from the initial to the final impulse is also an “indirect” or implicit
control variable as will be explained later. The objective or cost function for this problem is the sum of
the scalar magnitude of the two impulses given by
page 12
f  vi  v f
This is the scalar value we want to minimize.
The SNOPT algorithm requires initial guesses (xg) for the six components of the delta-v vectors as well
as lower (xlwr) and upper (xupr) bounds on each component. It also requires lower (flow) and upper
(fupp) bounds on the objective function and any linear or nonlinear constraints.
The following is the MATLAB source code that sets up this information.
% initial guess for components of initial delta-v
xg(1) = vti(1) - vi(1);

xg(2) = vti(2) - vi(2);
xg(3) = vti(3) - vi(3);
% initial guess for components of final delta-v
xg(4) = vf(1) - vtf(1);

xg(5) = vf(2) - vtf(2);
xg(6) = vf(3) - vtf(3);
xg = xg';
% define lower and upper bounds for components of delta-v vectors (kilometers/second)
dvm = norm(xg(1:3));
for i = 1:1:3
xlwr(i) = xg(i) - 0.01;
xupr(i) = xg(i) + 0.01;
end
dvm = norm(xg(4:6));
for i = 4:1:6
xlwr(i) = xg(i) - 0.01;
xupr(i) = xg(i) + 0.01;
end
xlwr = xlwr';
xupr = xupr';
% bounds on objective function
flow(1) = 0.0d0;
fupp(1) = +Inf;
% enforce final modified equinoctial equality constraints
flow(2) = 0.0d0;
fupp(2) = 0.0d0;
page 13
flow(3) = 0.0d0;
fupp(3) = 0.0d0;
flow(4) = 0.0d0;
fupp(4) = 0.0d0;
flow(5) = 0.0d0;
fupp(5) = 0.0d0;
if (itarget <= 1.0d-8)
% equatorial orbit constraint
flow(6) = 0.0d0;
fupp(6) = 0.0d0;
end
flow = flow';
fupp = fupp';
% read SNOPT specs file
snspec('snopt_specs.txt');
% solve the orbital TPBVP using SNOPT
snscreen on;
[x, f, inform, xmul, fmul] = snopt(xg, xlwr, xupr, flow, fupp, 'tpbvp');
The tpbvp MATLAB function defines the current value of the objective function and the mission
constraints. The function that evaluates the Earth gravity model and the first order equations of motion
is called ceqm1. Here’s the source code for this function.
function ydot = ceqm1 (t, y)
% first order form of Cowell's equations of orbital motion
% version for ode45
% input
% t = current simulation time

% y = current eci state vector
% output
% ydot = eci acceleration vector
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% compute non-spherical gravity perturbations
agrav = gravity(t, y);
% total acceleration vector
ydot = [ y(4)
y(5)
page 14
y(6)
agrav(1)
agrav(2)
agrav(3)];
Notice that the main script will read a SNOPT “specs” file named snopt_specs.txt which has the
following contents.
Begin SNOPT options

major iterations limit 50
minor iterations limit 100
derivative option 0
major optimality tolerance 1.0d-6
solution Yes
End SNOPT options
Additional information about this special file can be found in the SNOPT documentation.
The tpbvp objective function starts with the two-body solution for the position and velocity vectors of
the maneuver on the initial orbit and the current value for the delta-v vector and numerically integrates
the first-order form of the orbital equations to the descending node. At the descending node the software
adds the current value of the second delta-v vector to the velocity vector of the transfer orbit to
determine the velocity vector on the final mission orbit.
The position vector at the descending node and the resultant velocity vector are used to compute the
current modified equinoctial orbital elements of the final mission orbit.
The following is the MATLAB source code within the tpbvp function that uses the “event finding”
feature of the built-in ode45 algorithm to predict the descending node conditions. The bound for the
search time (tend) is 102% of the two-body Hohmann transfer time.
% set up options for ode45
options = odeset('RelTol', 1.0e-12, 'AbsTol', 1.0e-12, 'Events', @nc_event);
% solve for nodal crossing condition
rwrk = xi(1:3);
vwrk = xi(4:6);
% maximum search duration = 102% of two-body transfer time (seconds)
tend = 1.02 * tof;
[t, ysol, tevent, yevent, ie] = ode45(@ceqm1, [0 tend], [rwrk vwrk], options);
The following is the MATLAB source code for descending node objective function nc_event. The
computed value is simply the current z-component of the unit position vector.
function [value, isterminal, direction] = nc_event(t, y)
% nodal crossing event function
% required by phohmann.m

page 15
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% z-component of the unit position vector
value = y(3) / norm(y(1:3));
isterminal = 1;
direction = -1;
Note that setting direction = -1 ensures a descending node crossing since we want to search for
values of the z-component of position that are decreasing. This implementation essentially treats the
coast time between the maneuvers as an “indirect” or implicit control variable.
Targeting to the final mission orbit
This section summarizes the technique used to compute and enforce the nonlinear constraints that define
the final mission orbit. Since the final orbit of the Hohmann transfer is circular, we enforce the
following three modified equinoctial element nonlinear constraints
pt  p p  0 ft  f p  0 gt  g p  0
In these three equations, the t subscript implies “target” or desired values and the p subscript implies
values “predicted” by the optimization process. The targeted values are computed using the position and
velocity vectors of the mission orbit determined during the two-body Hohmann transfer solution.
Here’s the MATLAB source code that evaluates these constraints.
% enforce semiparameter and circular orbit constraints (p, f = g = 0)
f(2) = mee_final(1) - mee_target(1);
If the final mission orbit is equatorial (inclination = 0), we want to enforce the following two constraints.
ht  hp  0 kt  k p  0
If the orbital inclination of the final mission orbit is non-zero, we enforce the following single nonlinear
mission constraint.
 ht2  kt2    hp2  k p2   0
Here’s the MATLAB source code that evaluates this constraint.
% enforce mission orbit inclination constraint(s)
if (itarget <= 1.0d-8)
% equatorial orbit (h = k = 0)
f(5) = mee_final(4);
page 16
f(6) = mee_final(5);
else
% non-equatorial mission orbit (h^2 + k^2 = 0)
f(5) = (mee_final(4)^2 + mee_final(5)^2) - (mee_target(4)^2 + ...

mee_target(5)^2);
end
Note that we are not enforcing the right ascension of the ascending node (RAAN) or the true anomaly of
the final mission orbit.
Additional information about targeting with the modified equinoctial orbital elements can be found in
Appendix B.
Algorithm resources
(1) Walter Hohmann, Die Erreichbarkeit der Himmelskorper, Oldenbourgh, Munich, 1925. Also, The
Attainability of Heavenly Bodies, NASA Technical Translation F-44, 1960.
(2) Jean-Pierre Marec, Optimal Space Trajectories, Elsevier, 1979.
(3) R. P. Brent, Algorithms for Minimization Without Derivatives, Prentice-Hall, 1972.
(4) R. H. Battin, An Introduction to the Mathematics and Methods of Astrodynamics, AIAA, 1987.
(5) D. F. Lawden, Optimal Trajectories for Space Navigation, Butterworths, London, 1963.
(6) John E. Prussing, “Simple Proof of the Global Optimality of the Hohmann Transfer”, AIAA Journal
of Guidance, Control and Dynamics, Vol. 15, No. 4.
(7) A. Miele, M. Ciarcia, and J. Mathwig, “Reflections on the Hohmann Transfer”, Journal of
Optimization Theory and Applications, Vol. 123, No. 2, pp. 233-253, November 2004.
page 17
Appendix A
Two-body Hohmann Transfer
The coplanar circular orbit-to-circular orbit transfer was discovered by the German engineer Walter
Hohmann in 1925 and described in his classic report, The Attainability of Celestial Bodies. The transfer
consists of a velocity impulse on an initial circular orbit, in the direction of motion and collinear with the
velocity vector, which propels the space vehicle into an elliptical transfer orbit. At a transfer angle of
180 degrees from the first impulse, a second velocity impulse or V , also collinear and in the direction
of motion, places the vehicle into a final circular orbit at the desired final altitude. The impulsive V
assumption means that the velocity, but not the position, of the vehicle is changed instantaneously. This
is equivalent to a rocket engine with infinite thrust magnitude. Therefore, the Hohmann formulation is
the ideal and minimum energy solution to this type of orbit transfer problem.
Coplanar Equations
For the coplanar Hohmann transfer both velocity impulses are confined to the orbital planes of the initial
and final orbits. For a Hohmann transfer from a lower altitude orbit to a higher altitude circular orbit,
the first impulse creates an elliptical transfer orbit with a perigee altitude equal to the altitude of the
initial circular orbit and an apogee altitude equal to the altitude of the final orbit. The second impulse
circularizes the transfer orbit at apogee. Both impulses are posigrade which means that they are in the
direction of orbital motion.
We begin by defining three normalized radii as follows:
rf ri ri
R1  2 R2  R3  2
ri  rf rf ri  rf
where ri is the geocentric radius of the initial circular park orbit and rf is the radius of the final circular
mission orbit. The relationship between radius and initial orbit altitude hi and the final orbit altitude h f
is as follows:
ri  re  hi
rf  re  h f
where re is the radius of the Earth.
The magnitude of the first impulse is

V1  Vlc 1  R12  2 R1
and is simply the difference between the speed on the initial orbit and the perigee speed of the transfer
orbit. The scalar magnitude of the second impulse is
V2  Vlc R22  R22 R32  2 R22 R3
which is the difference between the speed on the final orbit and the apogee speed of the transfer ellipse.
page 18
In each of these V equations Vlc is called the local circular velocity. It can be determined from
Vlc   ri .
and represents the scalar speed in the initial orbit. In these equations  is the gravitational constant of
the central body. The transfer time from the first impulse to the second is equal to one half the orbital
period of the transfer ellipse
a3
 

where a is the semimajor axis of the transfer orbit and is equal to  ri  rf  / 2 . The orbital eccentricity of
the transfer ellipse is
max  ri , rf   min  ri , rf 
e
rf  ri
The following diagram illustrates the geometry of the coplanar Hohmann transfer.
transfer orbit
V1
ri
rf
initial orbit
V2
final orbit
Non-coplanar Equations
The non-coplanar Hohmann transfer involves orbital transfer between two circular orbits which have
different orbital inclinations. For this problem the propulsive energy is minimized if we optimally
partition the total orbital inclination change between the first and second impulses.
The scalar magnitude of the first impulse is
V1  Vlc 1  R12  2 R1 cos 1
where  1 is the plane change associated with the first impulse. The magnitude of the second impulse is
page 19
V2  Vlc R22  R22 R32  2R22 R3 cos 2
where  2 is the plane change associated with the second impulse. These two equations are different
forms of the law of cosines.
The total V required for the maneuver is the sum of the two impulses as follows
V  V1  V2
The relationship between the plane change angles is
t  1   2
where  t is the total plane change angle between the initial and final orbits.
Optimizing the non-coplanar Hohmann transfer involves allocating the total plane change angle between
the two maneuvers such that the total V required for the mission is minimized. We can determine this
answer by solving for the root of a derivative.
The partial derivative of the total V with respect to the first plane change angle is given by:
V R1 sin 1 R 2 R  sin  t cos1  cost sin 1 

  2 3
1 1  R12  2 R1 cos1 R22  R22 R32  2 R22 R3 cos  t  1 
If we determine the value of  1 which makes this derivative zero, we have found the optimum plane
change required at the first impulse. Once  1 is calculated we can determine  2 from the total plane
change angle relationship and the velocity impulses from the previous equations.
Numerical Solution
This numerical algorithm has been implemented in an interactive MATLAB script called hohmann.m.
This script prompts the user for the initial and final altitudes in kilometers and the initial and final orbital
inclinations in degrees. The software then calls the Brent root-finding algorithm to solve the partial
derivative equation described above.
The call to the Brent root-finding algorithm is as follows:
[xroot, froot] = brent('hohmfunc', 0, dinc, rtol);
where hohmfunc is the objective function for this problem. Since we know that the optimum first plane
change angle is somewhere between 0 and the total plane change angle dinc, we pass these as the
bounds of the root. In the parameter list rtol is the user-defined root-finding convergence tolerance.
The following discussion pertains to a typical orbit transfer from a low altitude Earth orbit (LEO) at an
altitude of 185.2 kilometers and an orbital inclination of 28.5 degrees to a geosynchronous Earth orbit
(GSO) at an altitude of 35786.36 kilometers and 0 degrees inclination.
page 20
The following is a V diagram for the first maneuver of this orbit transfer example. In this view we are
looking along the line of nodes which is the mutual intersection of the park and transfer orbit planes with
the equatorial plane.
V
Vi
Vp
26.3o
28.5o
equator
In this diagram Vi is the speed on the initial park orbit, V p is the perigee speed of the elliptic transfer
orbit, and V is the impulse required for the first maneuver. The inclinations of the park and transfer
orbit are also labeled. From this geometry and the law of cosines, the required V is given by
V  Vi 2  Vp2  2Vi Vp cos i
where i is the inclination difference or plane change between the park and transfer orbits.
Algorithm resources
(1) Walter Hohmann, Die Erreichbarkeit der Himmelskorper, Oldenbourgh, Munich, 1925. Also, The
Attainability of Heavenly Bodies, NASA Technical Translation F-44, 1960.
(2) John E. Prussing, “Simple Proof of the Global Optimality of the Hohmann Transfer”, AIAA Journal
of Guidance, Control and Dynamics, Vol. 15, No. 4.
(3) A. Miele, M. Ciarcia, and J. Mathwig, “Reflections on the Hohmann Transfer”, Journal of
Optimization Theory and Applications, Vol. 123, No. 2, pp. 233-253, November 2004.
page 21
Appendix B
Targeting with Modified Equinoctial Orbital Elements
The modified equinoctial orbital elements are a set of orbital elements that are useful for trajectory
analysis and optimization. They are valid for circular, elliptic, and hyperbolic orbits. These direct
modified equinoctial equations exhibit no singularity for zero eccentricity and orbital inclinations equal
to 0 and 90 degrees. However, please note that two of the components are singular for an orbital
inclination of 180 degrees.
The classic reference for these elements is “A Set of Modified Equinoctial Orbital Elements”, M. J. H.
Walker, B. Ireland and J. Owens, Celestial Mechanics, Vol. 36, pp. 409-419, 1985.
The modified equinoctial elements are defined in terms of the classical orbital elements as follows:
p  a 1  e 2 
f  e cos    
g  e sin    
h  tan  i 2  cos 
k  tan  i 2  sin 
L    
where
p  semiparameter
a  semimajor axis
e  orbital eccentricity
i  orbital inclination
  argument of perigee
  right ascension of the ascending node
  true anomaly
L  true longitude
The classical orbital elements can be recovered from the modified equinoctial orbital elements with
semimajor axis
p
a
1  f 2  g2
orbital eccentricity
e f 2  g2
page 22
orbital inclination
i  2 tan 1  h2  k 2 
argument of periapsis
  tan 1  g, f   tan 1  k , h 
gh f k
sin  
e tan  i 2 
f h gk
cos  
e tan  i 2 
right ascension of the ascending node
  tan 1  k , h 
k
sin  
tan  i 2 
h
cos  
tan  i 2 
true anomaly
  L      L  tan 1  g , f 
1
sin    f sin L  g cos L 
e
1
cos   f cos L  g sin L 
e
In these expressions, an inverse tangent expression of the form   tan 1  a, b  denotes a four quadrant
evaluation where a  sin  and b  cos .
Constraint formulations that enforce both the sine and cosine of a desired orbital element should be used
whenever possible. This approach involves a combination of equality and inequality constraints and
ensures that the “targeted” orbital element is in the correct quadrant.
To illustrate this technique, here are several examples for different values of argument of perigee and the
corresponding mission constraints:
page 23
 sin   0  gh  f k  0
0    90  
 f h  g k  e tan i 2  cos 
 sin   0  gh  f k  0
  270  
cos   0  f h  gk  0
 g h  f k  e tan  i 2  sin 
  178  
 cos   0  f h  gk  0
The following is a sign table of the sine and cosine for each quadrant.
quadrant sine cosine

1  
2  
3  
4  
orbital eccentricity constraint
e f 2  g2
For a circular orbit, f  g  0 .
orbital inclination constraint
i
tan    h 2  k 2
2
For an equatorial orbit, h  k  0 .
argument of perigee constraints
gh f k
g h  f k  e sin  tan  i 2   sin  
e tan  i 2 
f h gk
f h  g k  e cos  tan i 2   cos  
e tan  i 2 
right ascension of the ascending node constraints
k
k  tan  i 2  sin   sin  
tan  i 2 
page 24
h
h  tan  i 2  cos   cos  
tan  i 2 
true anomaly constraints
  L      L  tan 1  g , f 
In general,
1
sin    f sin L  g cos L 
e
1
cos   f cos L  g sin L 
e
For a circular orbit,
sin   sin L cos   cos L sin 
cos  cos L cos   sin L sin 
For a circular, equatorial orbit,
  L , sin  sin L and cos  cos L .
Also, note that

p
rp   periapsis radius
1 f 2  g2
p
ra   apoapsis radius
1 f 2  g2
Algorithm resources
“On the Equinoctial Orbital Elements”, R. A. Brouke and P. J. Cefola, Celestial Mechanics, Vol. 5, pp.
303-310, 1972.
“A Set of Modified Equinoctial Orbital Elements”, M. J. H. Walker, B. Ireland and J. Owens, Celestial
Mechanics, Vol. 36, pp. 409-419, 1985.
“Equinoctial Orbit Elements: Application to Optimal Transfer Problems”, Jean A. Kechichian, AIAA
90-2976, AIAA/AAS Astrodynamics Conference, Portland, OR, August 20-22, 1990.
page 25
Appendix C
Nonlinear Programming Problem
A trajectory optimization problem can be described by a system of dynamic variables
 y  t 
z 
u  t 
consisting of the state variables y and the control variables u for any time t. In this discussion vectors
are denoted in bold font.
The system dynamics are defined by a vector system of ordinary differential equations called the state
equations that can be represented as follows
dy
y  f  y  t  , u  t  , p, t 
dt
where p is a vector of problem parameters that is not time dependent.
The initial dynamic variables at time t0 are defined by ψ0  ψ  y  t0  , u  t0  , t0  and the terminal
conditions at the final time t f are defined by ψ f  ψ  y  t f  , u  t f  , t f  . These conditions are called the
boundary values of the trajectory problem.
The problem may also be subject to path constraints of the form g  y  t  , u  t  , t   0 .
For any mission time t there are also simple bounds on the state variables
yl  y  t   yu
the control variables
ul  u  t   u u
and the problem parameters
pl  p  t   pu
The basic nonlinear programming problem (NLP) involves the determination of the control vector
history and problem parameters that minimize the scalar performance index or objective function given
by
J    y  t0  , t0 , y  t f  , t f , p
while satisfying all the user-defined mission constraints.
page 26
Algorithm resources
(1) “Direct Trajectory Optimization Using Nonlinear Programming and Collocation”, C. R. Hargraves
and S. W. Paris, AIAA Journal of Guidance, Control and Dynamics, Vol. 10, No. 4, July-August, 1987,
pp. 338-342.
(2) “Optimal Finite-Thrust Spacecraft Trajectories Using Direct Transcription and Nonlinear
Programming”, Paul J. Enright, Ph.D. Thesis, University of Illinois at Urbana-Champaign, 1991.
(3) “Using Sparse Nonlinear Programming to Compute Low Thrust Orbit Transfers”, John T. Betts, The
Journal of the Astronautical Sciences, Vol. 41, No. 3, July-September 1993, pp. 349-371.
(4) “Improved Collocation Methods with Application to Direct Trajectory Optimization”, Albert L.
Herman, Ph.D. Thesis, University of Illinois at Urbana-Champaign, 1995.
(5) “Survey of Numerical Methods for Trajectory Optimization”, John T. Betts, AIAA Journal of
Guidance, Control and Dynamics, Vol. 21, No. 2, March-April 1998, pp. 193-207.
(6) Practical Optimization, Philip E. Gill, Walter Murray and Margaret H. Wright, Emerald Group
Publishing Limited, 1982.
page 27

The Gravity Perturbed Hohmann Transfer

Uploaded by

Copyright:

The Gravity Perturbed Hohmann Transfer

Uploaded by

Document Information

Copyright

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

The Gravity Perturbed Hohmann Transfer

Uploaded by

Copyright:

Orbital Mechanics with MATLAB

The Gravity Perturbed Hohmann Transfer

User interaction with the script

Input file format and contents

initial calendar date

altitude (kilometers; altitude > 0)

orbital inclination (degrees; 0 <= inclination <= 180)

altitude (kilometers; altitude > 0)

orbital inclination (degrees; 0 <= inclination <= 180)

central body gravitational constant (km^3/sec^2)

central body radius (kilometers)

order of the gravity model (zonals)

degree of the gravity model (tesserals)

initial orbit altitude 185.3200 kilometers

initial orbit inclination 28.5000 degrees

initial orbit velocity 7792.9603 meters/second

final orbit altitude 35788.0955 kilometers

final orbit inclination 0.0000 degrees

final orbit velocity 3074.5848 meters/second

first inclination change 2.1645 degrees

total inclination change 28.5000 degrees

first delta-v 2481.9821 meters/second

second delta-v 1790.3423 meters/second

total delta-v 4272.3244 meters/second

transfer orbit semimajor axis 24364.8478 kilometers

transfer orbit eccentricity 0.73061765

transfer orbit perigee velocity 10251.8686 meters/second

transfer orbit apogee velocity 1595.7727 meters/second

transfer orbit coast time 18924.5926 seconds

orbital elements and state vector of the initial orbit

sma (km) eccentricity inclination (deg) argper (deg)

raan (deg) true anomaly (deg) arglat (deg) period (min)

rx (km) ry (km) rz (km) rmag (km)

vx (kps) vy (kps) vz (kps) vmag (kps)

sma (km) eccentricity inclination (deg) argper (deg)

raan (deg) true anomaly (deg) arglat (deg) period (min)

rx (km) ry (km) rz (km) rmag (km)

vx (kps) vy (kps) vz (kps) vmag (kps)

sma (km) eccentricity inclination (deg) argper (deg)

raan (deg) true anomaly (deg) arglat (deg) period (min)

rx (km) ry (km) rz (km) rmag (km)

vx (kps) vy (kps) vz (kps) vmag (kps)

sma (km) eccentricity inclination (deg) argper (deg)

raan (deg) true anomaly (deg) arglat (deg) period (min)

rx (km) ry (km) rz (km) rmag (km)

vx (kps) vy (kps) vz (kps) vmag (kps)

initial delta-v vector, magnitude and steering angles

x-component of delta-v -2303.719596 meters/second

delta-v magnitude 2481.982050 meters/second

pitch angle -0.000000 degrees

final delta-v vector, magnitude and steering angles

x-component of delta-v 1619.451837 meters/second

delta-v magnitude 1790.342317 meters/second

pitch angle -0.000000 degrees

Nonlinear constraints 5 Linear constraints 1

The user has defined 0 out of 36 first derivatives

Major Minors Step nCon Feasible Optimal MeritFunction nS Penalty

SNOPTA EXIT 0 -- finished successfully

gravity-perturbed Hohmann transfer solution

orbital elements and state vector of the initial orbit

sma (km) eccentricity inclination (deg) argper (deg)

raan (deg) true anomaly (deg) arglat (deg) period (min)

rx (km) ry (km) rz (km) rmag (km)

vx (kps) vy (kps) vz (kps) vmag (kps)