Академический Документы
Профессиональный Документы
Культура Документы
www.elsevier.com/locate/actaastro
Abstract
To assess the fundamental issues of orbital transfers for Mars missions, in particular, the interplay between .ight time,
characteristic velocity, and mass ratio, we study optimal round-trip LEO-LMO-LEO trajectories for a variety of boundary
conditions, with LEO denoting a low Earth orbit and LMO denoting a low Mars orbit. In all cases, the criterion of optimization
is the minimization of the characteristic velocity, which is the sum of all the velocity impulses; the optimization problems are
solved via the sequential gradient-restoration algorithm in mathematical programming format. The assumed physical model
is the restricted four-body model, the four bodies being the Sun, Earth, Mars, spacecraft.
The round-trip trajectories considered di6er from one another in the boundary conditions, speci7cally: (T1) stay time in
LMO free, total time free, phase angle travel free, (T2) stay time in LMO of 30 days, total time free, phase angle travel
free, (T3) stay time in LMO of 30 days, total time free, phase angle travel of the spacecraft equal to that of Earth, (T4) stay
time in LMO of 30 days, total time of 440 days, phase angle travel of the spacecraft equal to that of Earth, (T5) zero stay
time in LMO, total time free, phase angle travel free.
Trajectories T1–T4 involve the circularization of the motion into LMO, Trajectory T5 is a free return trajectory with a
Mars .yby. In the group of trajectories circularizing the motion into LMO, Trajectory T1 is a minimum energy trajectory,
Trajectory T2 is a compromise trajectory, Trajectories T3 and T4 are fast transfer trajectories.
From computation and analysis, the main results are as follows.
(i) For a robotic spacecraft, the best trajectory is the minimum energy Trajectory T1, characterized by a .ight time of
970 days, a characteristic velocity of 11:30 km=s, and a mass ratio of 20.
(ii) For a manned spacecraft, a substantial shortening of the .ight time is needed, but this translates into sti6 penalties in
characteristic velocity and mass ratio. Indeed, for Trajectory T2, the .ight time is 842 days, while the characteristic velocity
is 15:61 km=s and the mass ratio is 69; for Trajectory T3, the .ight time is 546 days, while the characteristic velocity is
18:52 km=s and the mass ratio is 150; for Trajectory T4, the .ight time is 440 days, while the characteristic velocity is
20:79 km=s and the mass ratio is 304.
(iii) The free return Trajectory T5 is of no interest for Mars missions. While its characteristic velocity is nearly the same
as that of Trajectory T1, the .ight time has increased to 1214 days and the mass ratio has increased to 31.
It must be noted that the above mass ratios refer to a round-trip LEO- LMO-LEO and do not include the ascent from/descent
to the Earth surface (mass ratio of about 18) and the descent to/ascent from the Mars surface (mass ratio of about 3.6).
When the above mass ratios are included, one obtains overall mass ratios of order 1000 for Earth–Mars–Earth transfer via
a minimum energy trajectory and of order 10 000 for Earth–Mars–Earth transfer via a fast transfer trajectory.
Paper presented at the 51st IAF Congress, Rio de Janeiro, Brazil, 2–6 October 2000 (IAA-00-IAA.13.1.09).
∗ Corresponding author. Tel.: +1-713-348-4907; fax: +1-713-348-5407.
E-mail address: miele@rice.edu (A. Miele).
0094-5765/$ - see front matter c 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.actaastro.2003.12.018
80 A. Miele et al. / Acta Astronautica 55 (2004) 79 – 94
At the national level, the wisest strategy appears to be to continue the exploration of Mars via robotic spacecraft and
delay the exploration of Mars via manned spacecraft to such time when, thanks to advances yet to be achieved in the areas
of spacecraft structural factors and engine speci7c impulses, more reasonable values can be achieved for the characteristic
velocity components and the mass ratio components.
c 2004 Elsevier Ltd. All rights reserved.
developed by McDonnell Douglas Technical Service sumed at the departure from LEO and arrival to
Company [16]. LEO; no velocity impulse is applied at LMO.
This paper is organized as follows. Section 2 de- Having adopted the restricted four-body model, 7ve
scribes the system equations. Section 3 presents the motions must be considered: the inertial motions of
boundary conditions. Section 4 describes the optimal Earth, Mars, and the spacecraft with respect to the Sun;
trajectory problems. Section 5 presents the planetary the relative motions of the spacecraft with respect to
and orbital data. Section 6 discusses the numerical Earth and Mars. To study these motions, we employ
results. Finally, Section 7 contains the conclusions. three coordinate systems: the Sun coordinate system
(SCS), Earth coordinate system (ECS), and Mars co-
ordinate system (MCS). All of these systems are con-
2. System equations tained in the idealized Earth/Mars orbital plane and
their axes point toward 7xed directions in space.
2.1. Assumptions
(SCS) This system is centered in the Sun center; the
x-axis points toward the Earth initial position;
Let LEO denote a low Earth orbit, and let LMO
the y-axis is perpendicular to the x-axis and
denote a low Mars orbit. The missions considered in
is directed as the Earth initial velocity vector.
this paper involve the spacecraft transfer from LEO
(ECS) This system is centered in the Earth center;
to LMO for the outgoing trip and from LMO to LEO
the x-axis and y-axis of ECS are parallel to
for the return trip. To study these missions, we em-
the corresponding axes of SCS.
ploy the restricted four-body model (Sun, Earth, Mars,
(MCS) This system is centered in the Mars center;
spacecraft) and the following assumptions:
the x-axis and y-axis of MCS are parallel to
(Al) the Sun is 7xed in space, the corresponding axes of SCS.
(A2) Earth and Mars are subject to the Sun gravity,
(A3) the eccentricity of the Earth and Mars orbits 2.3. Notations
around the Sun is neglected, implying circular
planetary motions, Cartesian coordinates or polar coordinates are em-
(A4) the inclination of the Mars orbital plane ployed, depending on the need. Single subscripts de-
vis-a-vis the Earth orbital plane is neglected, note inertial quantities, while double subscripts denote
implying planar spacecraft motion, relative-to-planet quantities. For coordinate transfor-
(A5) the spacecraft is subject to the gravity 7elds of mations, see Ref. [17].
the Sun, Earth, Mars along the entire trajectory, (SCS) In Cartesian coordinates, the inertial motions
(A6) LEO and LMO are circles centered in their re- of Earth E, Mars M, and the spacecraft P with respect
spective planets; circularization of the space- to the Sun S are described by the following quantities:
craft motion is assumed prior to departure and
after arrival, xE ; yE ; uE ; wE ; (1a)
(A7) for a round trip executed via a minimum energy
trajectory, four velocity impulses are assumed xM ; yM ; uM ; wM ; (1b)
at the departure from LEO, arrival to LMO, de-
parture from LMO, and arrival to LEO; for a xP ; yP ; uP ; wP ; (1c)
round trip executed via a fast transfer trajec-
with the pair (x; y) identifying the position vector and
tory, an additional midcourse impulse might be
the pair (u; w) identifying the velocity vector. In polar
needed in either the outgoing trip or the return
coordinates, the quantities corresponding to (1) are
trip; for a round trip executed via a free-return
trajectory, only two velocity impulses are as- rE ; E ; VE ; E or E ; (2a)
82 A. Miele et al. / Acta Astronautica 55 (2004) 79 – 94
3.4. Departure from LMO, return trip VPE (t5 ) = VLEO + NVLEO (t5 );
√
In the Mars-centered system and using polar coordi- VLEO = (E =rLEO ); (23b)
nates, the spacecraft conditions at the departure from
LMO (time t = t3 ) are given by PE (t5 ) = 0 or
VPM (t3 ) = VLMO + NVLMO (t3 ); with PE (t5 ) free. Formally, Eqs. (23) can be obtained
√ from Eqs. (18) by simply replacing the time t = t0
VLMO = (M =rLMO ); (21b) with the time t = t5 . However, there is a di6erence
A. Miele et al. / Acta Astronautica 55 (2004) 79 – 94 85
PM (t2 ) = PM (t2 ) − PM (t2 ) + =2; (25b) NV = NVLEO (t0 ) + NVMID (t1 ) + NVLMO (t2 )
with PM (t2 ) and VPM (t2 ) free. +NVLMO (t3 ) + NVMID (t4 ) + NVLEO (t5 ): (34)
With this understanding, we formulated the following
general problem: (P) Minimize the total characteris-
4. Optimal trajectory problems
tic velocity (34) subject to constraints (7) –(25) plus
the possible presence of additional constraints on the
Let OUT , STAY , RET , denote the .ight time of
variables appearing in Eqs. (27), (30), (33).
the outgoing trip, stay time in LMO, .ight time of
The solution of Problem (P) is called Trajectory T.
the return trip, and total .ight time. By de7nition, the
While Problem (P) looks like a formidable problem
following relations hold:
of optimal control, it can be reduced to a mathemat-
OUT = t2 − t0 ; STAY = t3 − t2 ; ical programming problem involving relatively few
constraints and parameters if the computation is or-
RET = t5 − t3 ; = t 5 − t0 ; (26) ganized in such a way that most of constraints (7) –
so that (25) are satis7ed in a trivial way Refs. [9,10]. With
this understanding, 7ve particular aspects of Problem
= OUT + STAY + RET : (27)
(P) are considered below.
Let
4.1. Minimum energy trajectory T1
N (t) = M (t) − E (t) (28)
denote the Mars/Earth inertial phase angle di6erence In this problem, we assume that there is no mid-
at any time. Use of Eqs. (7b) and (9b) between the course impulse in either the outgoing trip or the return
times t2 and t3 yields the relation trip,
N (t3 ) = N (t2 ) + (!M − !E )(t3 − t2 ); (29) NVMID (t1 ) = 0; NVMID (t4 ) = 0: (35)
86 A. Miele et al. / Acta Astronautica 55 (2004) 79 – 94
Hence, the characteristic velocity (34) reduces to With the above understanding, we formulate the fol-
lowing problem: (P1) Minimize the performance in-
NV = NVLEO (t0 ) + NVLMO (t2 ) + NVLMO (t3 ) dex (36) with respect to the parameters (38) consistent
+NVLEO (t5 ): (36) with the essential constraints (20), (23), (27), (30).
To sum up, we are in the presence of a mathemat-
Also, we assume that the total .ight time and the stay ical programming problem involving n = 12 parame-
time in LMO are unconstrained, more precisely, ters, q = 8 constraints, and hence n − q = 4 degrees
of freedom. These degrees of freedom must be satu-
= free; STAY = free: (37) rated in such a way that the performance index (36)
is minimized. The solution of Problem P1 is called
The reduction to a mathematical programming
Trajectory T1.
problem is done by recognizing that the essential
constraints are these:
4.2. Compromise trajectory T2
(i) for the outgoing trip, the LMO arrival conditions
(20); In this problem, we assume that there is no mid-
(ii) for the return trip, the LEO arrival conditions course impulse in either the outgoing trip or return
(23); trip, so that (35) and (36) still hold. Also, we assume
(iii) for the entire trip, the time relation (27) connect- that the total .ight time is unconstrained, while the
ing the total .ight time and the partial times; stay time in LMO is 7xed, more precisely,
(iv) for the stay on Mars, the relation (30) connect-
= free; (39a)
ing the stay time in LMO with the Mars/Earth
phase angle di6erences at the arrival to LMO and
departure from LMO. STAY = 30 days: (39b)
The reduction to a mathematical programming
The parameters appearing in the above constraints are
problem is done by recognizing that the essential
these:
constraints are still (20), (23), (27), (30), while the
(A) the velocity impulses variables are the same as (38) minus STAY , which is
7xed (see Eq. (39b)].
NVLEO (t0 ); NVLMO (t2 ); With the above understanding, we formulate the fol-
lowing problem: (P2) Minimize the performance in-
NVLMO (t3 ); NVLEO (t5 ); (38a) dex (36) with respect to the parameters (38) consis-
(B) the transfer/stay times tent with the constraints (20), (23), (27), (30), with
STAY excluded from the parameter count.
OUT ; STAY ; RET ; ; (38b) To sum up, we are in the presence of a mathemat-
ical programming problem involving n = 11 parame-
(C) the Mars/Earth inertial phase angle di6erences ters, q = 8 constraints, and hence n − q = 3 degrees
at the departure from LEO and departure from of freedom. These degrees of freedom must be satu-
LMO, rated in such a way that the performance index (36)
is minimized. The solution of Problem P2 is called
N (t0 ) = M (t0 ) − E (t0 );
Trajectory T2.
N (t3 ) = M (t3 ) − E (t3 ); (38a)
4.3. Fast transfer trajectory T3
(D) the spacecraft/Earth relative phase angles at the
departure from LEO and departure from LMO, The solutions T1 and T2 of the minimum energy
PE (t0 ) = PE (t0 ) − E (t0 ); Problems P1 and P2 satisfy Eq. (33b); namely, they
are such that the phase angle travel of Earth is larger
PM (t3 ) = PM (t3 ) − M (t3 ): (38d) than that of the spacecraft by 360◦ . Note that, for
A. Miele et al. / Acta Astronautica 55 (2004) 79 – 94 87
Problems P1 and P2, Eq. (33b) is not a constraint, but 4.4. Fast transfer trajectory T4
a result of the optimization process.
To generate a round-trip fast transfer trajectory, one This problem is complementary to Problem P3, in
must force the phase angle travels of Earth and the the sense that we assume the presence of an acceler-
spacecraft to be equal or nearly equal; hence, one must ating velocity impulse at midcourse of the outgoing
force the satisfaction of Eq. (33a) and for that matter trip; hence,
the violation of Eq. (33b). Physically, this requires the
NVMID (t1 ) = 0; NVMID (t4 ) = 0: (44)
presence of a midcourse velocity impulse in either the
outgoing trip or the return trip. Clearly, the characteristic velocity (34) reduces to
In this section, we assume the presence of a deceler-
NV = NVLEO (t0 ) + NVMID (t1 ) + NVLMO (t2 )
ating velocity impulse at midcourse of the return trip,
hence +NVLMO (t3 ) + NVLEO (t5 ): (45)
NVMID (t1 ) = 0; NVMID (t4 ) = 0: (40) Further, we assume that both the total .ight time and
Clearly, the characteristic velocity (34) reduces to the stay time in LMO are given, more precisely,
NV = NVLEO (t0 ) + NVLMO (t2 ) + NVLMO (t3 ) = 440 days; (46a)
there are no velocity impulses at LMO. In addition, (D) the spacecraft/Earth relative phase angle at de-
there is no need for a midcourse impulse. Since parture from LEO,
NVLMO (t2 ) = NVLMO (t3 ) = 0; PE (t0 ) = PE (t0 ) − E (t0 ): (52d)
NVMID (t1 ) = NVMID (t4 ) = 0; (48)
With the above understanding, we formulate the
the characteristic velocity (34) reduces to following problem: (P5) Minimize the performance
NV = NVLEO (t0 ) + NVLEO (t5 ): (49) index (49) with respect to the parameters (52), subject
to the constraints (23), (25), (51).
The Mars .yby occurs tangentially to LMO in the To sum up, we are in the presence of a mathemat-
relative motion of the spacecraft with respect to Mars. ical programming problem including n = 7 parame-
The .yby time is t2 = t3 . ters, q = 6 constraints, and hence n − q = 1 degree of
The total .ight time is unconstrained and the stay freedom. This single degree of freedom must be sat-
time in LMO is vanishingly small; hence, urated in such a way that the performance index (49)
= free; (50a) is minimized. The solution of Problem P5 is called
Trajectory T5.
STAY = 0: (50b)
Since we dispense with the fast transfer condition 5. Mission data
(33a), one must expect the round-trip free return tra-
jectory to be consistent with (33b), although this equa- 5.1. Planetary data
tion is not imposed as a constraint of the problem.
The reduction to a mathematical programming The gravitational constants for the Sun, Earth, and
problem is done by recognizing that, since Eq. (30) Mars are given by
reduces to an identity, the essential constraints are
these: S = 1:327E11; E = 3:986E05;
(A) the velocity impulses In particular, the angular velocity di6erence between
Earth and Mars is
NVLEO (t0 ); NVLEO (t5 ); (52a)
N! = !E − !M = 0:4614 (deg=day): (55)
(B) the transfer times
OUT ; RET ; ; (52b) 5.2. Orbital data
(C) the Mars/Earth inertial phase angle di6erence at
For the outgoing trip, the spacecraft is to be trans-
departure from LEO,
ferred from a low Earth orbit to a low Mars orbit; for
N (t0 ) = M (t0 ) − E (t0 ); (52c) the return trip, the spacecraft is to be transferred from
A. Miele et al. / Acta Astronautica 55 (2004) 79 – 94 89
a low Mars orbit to a low Earth orbit. The radii of the Table 2
terminal orbits are Characteristic velocities (km/s) for a round-trip LEO-LMO-LEO
rLEO = 6841; rLMO = 4097 (km); (56a) NV T1 T2 T3 T4 T5
corresponding to the altitudes Outgoing trip NVLEO 3.55 3.79 3.62 5.07 7.57
NVMID 0.00 0.00 0.00 2.68 0.00
hLEO = 463; hLMO = 200 (km); (56b) NVLMO 2.10 4.15 2.72 5.06 0.00
since the Earth and Mars surface radii are given by Return trip NVLMO 2.10 3.92 4.62 3.80 0.00
NVMID 0.00 0.00 3.48 0.00 0.00
RE = 6378; RM = 3397 (km): (56c)
NVLEO 3.55 3.76 4.08 4.18 3.76
The circular velocities (subscript c) at LEO and LMO
Round trip NVOUT 5.65 7.93 6.34 12.81 7.57
are given by
NVRET 5.65 7.68 12.17 7.97 3.76
(Vc )LEO = 7:633; (Vc )LMO = 3:451 (km=s) (56d) NV 11.30 15.61 18.52 20.79 11.33
Table 3
Major parameters for a round trip LEO-LMO-LEO
Quantity T1 T2 T3 T4 T5
Notation: = t5 − t0 , Flight time (days); NV , Characteristic velocity (km/s); P (t5 ) − P (t0 ), Spacecraft phase angle travel (deg);
E (t5 ) − E (t0 ), Earth phase angle travel (deg); M (t5 ) − M (t0 ), Mars phase angle travel (deg); m(t0 )=m(t5 ), Ratio of mass departing
from LEO to mass returning to LEO, computed for a multistage rocket with uniform structural factor = 0:1 and engine speci7c impulse
ISP = 450 s.
Table 4 Table 7
Phase angle histories of Earth, Mars, spacecraft for Trajectory T1 Phase angle histories of Earth, Mars, spacecraft for Trajectory T4
Table 5 Table 8
Phase angle histories of Earth, Mars, spacecraft for Trajectory T2 Phase angle histories of Earth, Mars, spacecraft for Trajectory T5
t (days) E (deg) M (deg) P (deg) For Trajectory T3, the .ight time of 546 days in-
t0 = 0:0 0.0 46.8 0.0 cludes 208 days for the outgoing trip, 308 days for
t2 = 207:7 204.7 155.7 155.7 the return trip, and 30 days staying in LMO. The
t3 = 237:7 234.3 171.4 171.4 characteristic velocity is 18:52 km=s and the corre-
t5 = 545:9 538.0 333.0 538.0 sponding mass ratio is 150. Table 6 shows that, upon
completion of the round trip, the spacecraft phase an-
gle travel is the same as the Earth phase angle travel.
Trajectory T3. This is a fast transfer trajectory com- Comparing Trajectories T3 and T1, we see that a sti6
puted for a 7xed stay time in LMO of 30 days, free price must be paid for shortening the total time: a
total time, and phase angle travel of the spacecraft 64% increase in characteristic velocity, which trans-
equal to that of Earth; a midcourse velocity impulse lates into a mass ratio increase according to a multi-
is assumed for the return trip. plicative factor of 7.5.
A. Miele et al. / Acta Astronautica 55 (2004) 79 – 94 91
Arrival to LMO
Arrival to LMO
Sun
Earth
Earth
Sun
Mars
Mars
Departure from LEO
Arrival to LEO
Departure from LMO
Mars
Mars
Sun
Earth
Arrival to LEO
Sun
Fig. 2. Compromise trajectory in interplanetary space, Sun coor-
dinates, total time = 841.7 days, stay time in LMO = 30.0 days,
Departure from LMO total velocity impulse = 15:61 km=s, no midcourse impulse.
Earth
the total time: an 84% increase in characteristic ve-
Fig. 1. Minimum energy trajectory in interplanetary space, Sun locity, which translates into a mass ratio increase ac-
coordinates, total time = 970.0 days, stay time in LMO = 454.3 cording to a multiplicative factor of 15.1.
days, total velocity impulse = 11:30 km=s, no midcourse impulse. Trajectory T5. This is a free-return trajectory com-
puted for zero stay time in LMO, free total time, and
free phase angle travel; there is no circularization of
Trajectory T4. This is a fast transfer trajectory com- the motion in LMO; there is no midcourse velocity
puted for a 7xed stay time in LMO of 30 days, a 7xed impulse in either the outgoing trip or the return trip.
total time of 440 days, and phase angle travel of the The spacecraft departs from LEO, executes a Mars
spacecraft equal to that of Earth; a midcourse velocity .yby at the prescribed altitude of 200 km above the
impulse is assumed for the outgoing trip. Mars surface, and returns to LEO.
For Trajectory T4, the .ight time of 440 days in- For Trajectory T5, the .ight time of 1214 days in-
cludes 261 days for the outgoing trip, 149 days for cludes 246 days for the outgoing trip and 968 days for
the return trip, and 30 days staying in LMO. The the return trip; there is no stay time in LMO. The char-
characteristic velocity is 20:79 km=s and the corre- acteristic velocity is 11:33 km=s and the correspond-
sponding mass ratio is 304. Table 7 shows that, upon ing mass ratio is 31. Table 8 shows that, upon com-
completion of the round trip, the spacecraft phase an- pletion of the round trip, the spacecraft phase angle
gle travel is the same as the Earth phase angle travel. travel is 360◦ less than the Earth phase angle travel.
Comparing Trajectories T4 and T1, we see that an Comparing Trajectories T5 and T1, we see that,
even sti6er price must be paid for further shortening even though the characteristic velocity has remained
92 A. Miele et al. / Acta Astronautica 55 (2004) 79 – 94
Earth
Earth
Sun
Sun
Sun
Earth
Earth
Sun Mars
References
[14] A. Miele, R.E. Pritchard, I.N. Damoulakis, Sequential [16] B.H. Rishikof, B.R. McCormick, R.E. Pritchard, S.J.
gradient-restoration algorithm for optimal control problems, Sponaugle, SEGRAM: a practical and versatile tool for
Journal of Optimization Theory and Applications 5 (4) (1970) spacecraft trajectory optimization, Acta Astronautica 26
235–282. (8–10) (1992) 599–609.
[15] A. Miele, T. Wang, V.K. Basapur, Primal and dual [17] A. Miele, Flight Mechanics, Vol. 1: Theory of Flight Paths,
formulations of sequential gradient-restoration algorithms for Addison-Wesley Publishing Company, Reading, MA, 1962.
trajectory optimization problems, Acta Astronautica 13 (8)
(1986) 491–505.