You are on page 1of 11

MNRAS 478, 1392–1402 (2018) doi:10.

1093/mnras/sty1155
Advance Access publication 2018 May 4

Energy analysis in the elliptic restricted three-body problem

Yi Qi and Anton de Ruiter‹


Department of Aerospace Engineering, Ryerson University, 350 Victoria Street, Toronto, ON M5B 2K3, Canada

Downloaded from https://academic.oup.com/mnras/article/478/1/1392/4992760 by Ryerson University Library user on 14 November 2020


Accepted 2018 April 30. Received 2018 April 30; in original form 2018 April 6

ABSTRACT
The gravity assist or flyby is investigated by analysing the inertial energy of a test particle
in the elliptic restricted three-body problem (ERTBP), where two primary bodies are moving
in elliptic orbits. First, the expression of the derivation of energy is obtained and discussed.
Then, the approximate expressions of energy change in a circular neighbourhood of the
smaller primary are derived. Numerical computation indicates that the obtained expressions
can be applied to study the flyby problem of the nine planets and the Moon in the Solar
system. Parameters related to the flyby are discussed analytically and numerically. The optimal
conditions, including the position and time of the periapsis, for a flyby orbit are found to
make a maximum energy gain or loss. Finally, the mechanical process of a flyby orbit is
uncovered by an approximate expression in the ERTBP. Numerical computations testify that
our analytical results well approximate the mechanical process of flyby orbits obtained by
the numerical simulation in the ERTBP. Compared with the previous research established in
the patched-conic method and numerical calculation, our analytical investigations based on a
more elaborate derivation get more original results.
Key words: methods: analytical – methods: numerical – celestial mechanics.

and found the most convenient position and direction of the veloc-
1 I N T RO D U C T I O N
ity impulse. An analytical technique, directly related to Tisserand’s
Since the gravity assist (or flyby or swing-by) can significantly criterion, that permits the quick identification of all viable gravity-
change the path and energy of a spacecraft flying close to a celestial assist sequences to a given destination was proposed by Strange &
body without maneuvers based on propulsive systems, it is widely Longuski (2002). For lunar flyby orbits in the Earth–Moon planar
applied to many deep space missions, such as the Jupiter flybys of CRTBP, Qi & Xu (2015) derived the approximate expressions of
the Voyager spacecraft, the multiple flybys of the Galileo spacecraft energy in a small neighbourhood of the Moon and uncovered their
for the Jupiter exploration, and the multiple flybys of the Rosetta mechanical process. Furthermore, taking advantage of the lunar
spacecraft for the comet exploration. flyby, low-energy transfers to the Moon and libration point orbits
Broucke (1988) investigated the celestial mechanics of the grav- were constructed (Qi & Xu 2016, 2017; Qi, Xu & Qi 2017a,b). Nu-
ity assist and obtained some important preliminary results from the merical results indicated that a lunar flyby can efficiently reduce the
patched-conic approach:There is a loss of energy if the flyby is in fuel cost and transfer duration compared with traditional transfers
front of the smaller primary, and there is an increase in the energy and other low-energy transfers without lunar flybys.
if the flyby is behind the smaller primary. Longuski & Williams In the Solar system, none of the orbital eccentricities of celestial
(1991) presented an approach to planetary mission design that au- bodies are exactly equal to zero, i.e. their orbits are not circular,
tomates the search for gravity-assist trajectories. Sims, Longuski and some of them even are highly eccentric. Therefore, the elliptic
& Staugler (1997) proposed a concept of V∞ leveraging to anal- restricted three-body problem (ERTBP), a more precise model than
yse the Earth gravity-assist trajectory. After that, using the circular the CRTBP, was proposed to depict the motion of a test particle, such
restricted three-body problem (CRTBP), Campagnola et al. (2014) as a spacecraft, comet, or an asteroid, in the gravitational field of
developed Tisserand-leveraging transfers to extend the concept of two primary bodies moving in elliptic motions around their barycen-
V∞ leveraging beyond the patched-conics domain. The powered tre (Szebehely 1967). Different from the autonomous CRTBP, the
swing-by is a technique combining the classic swing-by with an ERTBP is time-varying system. Therefore, some conclusions estab-
impulse delivered to the spacecraft by a propulsion system during lished in the CRTBP, such as the Jacobi constant and energetically
the swing-by. Casalino, Colasurdo & Pastrone (1999) investigated forbidden regions, cannot be applicable in the ERTBP. For example,
the single-impulse powered swing-by using the parametric analysis a spacecraft restricted in the inner Earth–Moon ERTBP region can
finally escape due to this non-conservative feature of the ERTBP
(Peng et al. 2015). Broucke (1969) systematically studied the sta-
 E-mail: aderuiter@ryerson.ca bility of periodic orbits in the planar ERTBP and pointed out that

C 2018 The Author(s)

Published by Oxford University Press on behalf of the Royal Astronomical Society


Energy analysis in ERTBP 1393
there exist discrete periodic orbits in the ERTBP. Hou & Liu (2011) the masses of larger and smaller primaries, respectively. The mass
studied motions around the collinear libration points in the ERTBP ratio of the ERTBP μ is defined as m2 /(m1 + m2 ). Fig. 1 shows the
and obtained literal expansions of the quasi-periodic and periodic schematic of the ERTBP. XYZ coordinates are inertial coordinates,
orbits in the ERTBP using the Lindstedt–Poincaré method. Peng whose origin O is located at the barycentre and the X-axis is along
& Xu (2015b) and Peng, Bai & Xu (2017b) numerically analysed the vector from O to the periapsis of the elliptic orbit of m2 . xyz
linear stabilities of several families of spatial periodic orbits in the coordinates are pulsating rotating coordinates, whose origin O is
ERTBP. In practice, both low-energy and impulsive transfers from similar to that of the inertial coordinates, and the x-axis is along
the Earth to these periodic orbits can be constructed in the Earth– the vector from m1 to m2 . In the dimensionless pulsating rotating
Moon ERTBP (Peng & Xu 2015a,c), and these periodic orbits can frame, larger and smaller primaries are located in (−μ, 0, 0) and

Downloaded from https://academic.oup.com/mnras/article/478/1/1392/4992760 by Ryerson University Library user on 14 November 2020


provide novel deep space satellite constellations (Peng & Bai 2018). (1 − μ, 0, 0), respectively. The length unit LU is set as the distance
Recently, Peng et al. (2017a) showed that the ERTBP model can between two primary bodies, i.e. r in Fig. 1, which is time varying.
be used to accelerate trajectory searching through GPU parallel The time unit TU is set as Tell /2π, where Tell is the period of elliptic
computing. The gravitational capture in the ERTBP was also in- orbit. We designate the true anomaly f of the elliptic orbit as the
vestigated by many researchers (Prado & Neto 2006; Hyeraci & independent variable of the system. For convenience, in this paper,
Topputo 2010; Qi, Xu & Qi 2014). we use the superscript () to denote differentiation with respect to f.
In addition, some scholars focused on the gravity-assist problem Then, for the ERTBP, the equations of motion of test particle in the
in the ERTBP. The swing-by trajectories were studied and classified dimensionless rotating frame is expressed as (Szebehely 1967)
in the ERTBP by Prado (1997). The effects of the eccentricity and
x  − 2y  = ωx ,
true anomaly of primaries in their elliptic orbits on the swing-by
were quantified. His results could be used to find optimal sets of y  + 2x  = ωy ,
parameters for several problems, such as optimal escape orbit and z + z = ωz , (1)
so forth. Ferreira et al. (2017) studied the powered swing-by ma-
neuver in the ERTBP. The impulse with different magnitudes and where ω = /(1 + e cos f) is effective potential in the ERTBP; ωx ,
directions was applied when the spacecraft passed by the periapsis ωy and ωz are the partial derivatives of ω with respect to the position
of its orbit around the body. They provided the best conditions in variables x, y and z. e is the eccentricity of the orbital motion of m2
terms of gains or losses of energy. Ferreira et al. (2018) derived a around barycentre O, and
set of analytical equations that describe a swing-by maneuver in the 1 2 1−μ μ 1
= (x + y 2 + z2 ) + + + μ(1 − μ),
ERTBP using the patched-conic approach. From the obtained ana- 2 r1 r2 2
lytical equations, they found that the variations of energy, velocity,
while
and angular momentum were influenced by the angle between the 
velocity vector of the secondary body and the line connecting the r1 = (x + μ)2 + y 2 + z2 ,
primaries. 
r2 = (x + μ − 1)2 + y 2 + z2 ,
However, we found that all existing literature about the gravity
assist in the ERTBP were based on the numerical computation or are dimensionless distances of the test particle from the larger and
patched-conic approach, the real analytical equations established smaller primaries, respectively.
in the ERTBP are not found in the literature (to the best of the
authors’ knowledge). In an attempt to get a thorough understanding
3 I N E RT I A L E N E R G Y
of a flyby in the ERTBP, in this paper, the inertial energy of a test
particle in the ERTBP will be analysed in detail. Several useful In the barycentric inertial coordinates (see Fig. 1), the position
and original expressions are derived and elaborated in the ERTBP. coordinates of the test particle can be obtained from
Then, the obtained expressions will be applied to investigate the ⎛ ⎞ ⎛ ⎞
X x
flyby problems in the Solar system. The best conditions for energy ⎝ Y ⎠ = r(f )T (f )⎝ y ⎠, (2)
change are analytically obtained from expressions. In addition, the
Z z
mechanical process of a flyby orbit in the ERTBP will be uncovered
by an approximate expression that is also a gap in the existing where
literature. ⎛ ⎞
cos f − sin f 0
According to the above discussion, this paper is divided into four T = ⎝ sin f cos f 0 ⎠, (3)
parts. The dynamical model of this paper, the ERTBP, is introduced 0 0 1
in Section 2. The inertial energy of the test particle in the ERTBP
is derived in Section 3. In Section 4, we analyse characteristics of
a(1 − e2 ) p
the inertial energy. In Section 5, the obtained approximate expres- r(f ) = = , (4)
sions are applied to investigate the ERTBP in the Solar system. In 1 + e cos f g
Section 6, we give our conclusions of this paper. while g = 1 + ecos f. a and p = a(1 − e2 ) are the semimajor axis and
semilatus rectum of the orbital motion of m2 around the barycentre,
2 E L L I P T I C R E S T R I C T E D T H R E E - B O DY respectively. Taking the derivative of equation (2) with respect to f,
P RO B L E M we can get
⎛ ⎞ ⎛ ⎞ ⎛ ⎞
In the ERTBP, the mass of a test particle, such as the asteroid, X x x
⎝ Y ⎠ = r (f )T (f )⎝ y ⎠ + r(f )T (f )⎝ y ⎠
  
comet, or spacecraft, is supposed to be negligible, and only the
effects of the two primary bodies are considered. The ERTBP is Z z z
⎛ ⎞ ⎛  ⎞
built on the hypothesis that two primary bodies move in elliptic x x − y + Ax
motions about their barycentre. In this paper, m1 and m2 represent + r(f )T (f )⎝ y  ⎠ = r(f )T (f )⎝ y  + x + Ay ⎠, (5)
z z + Az

MNRAS 478, 1392–1402 (2018)


1394 Y. Qi and A. de Ruiter

Downloaded from https://academic.oup.com/mnras/article/478/1/1392/4992760 by Ryerson University Library user on 14 November 2020


Figure 1. Schematic of the ERTBP.

where A = e sin f/(1 + e cos f) = −g /g = r /r is an important As we can see from equation (9), there exist two terms determined
parameter in the following derivation. by the position of the test particle, including
The true anomaly f and the time t have following relationship: 
1 1
 P1 = y − 3 ,
r13 r2
df G(m1 + m2 )
= (1 + e cos f )2 x+μ x+μ−1
dt a 3 (1 − e2 )3 P2 = − . (10)
r13 r23
 g2
= G(m1 + m2 )  , (6) In addition, E = −G(m1 + m2 )/2a is the specific mechanical
p3 energy of the elliptic orbit of celestial bodies (Chobotov 2002).
Let G(m1 + m2 )/a = −2E = 1, then using equations (4) and (10),
where G is the Newtonian gravitational constant.
equation (9) can be rewritten as
Let V denote the velocity magnitude of the test particle in the
barycentric inertial frame, then according to equations (5) and (6), dE G(m1 + m2 )μ(1 − μ)g
= (P1 + AP2 )
we can obtain df a(1 − e2 )
   μ(1 − μ)g
dX 2
dY 2
dZ 2 = (P1 + AP2 )
V2 = + + 1 − e2
dt dt dt μ(1 − μ)
 = [P1 (1 + e cos f ) + P2 e sin f ] . (11)
df 2  2 1 − e2
(X + Y  + Z  )
2 2
=
dt If e = 0, the ERTBP is degraded into the CRTBP, and
G(m1 + m2 )g 
= [(x − y + Ax)2 + (y  + x + Ay)2 dE dE df
r =
dt df dt
+ (z + Az)2 ]. (7)
g 2 μ(1 − μ)
=  [P1 (1 + e cos f ) + P2 e sin f ]
The potential energy U of the test particle with respect to the p 3 1 − e2

barycentre can be expressed as 1 1
= μ(1 − μ)P1 = μ(1 − μ)y − 3 , (12)
 r13 r2
G(m1 + m2 ) 1−μ μ
U =− + . (8) that accords with the result in Qi & Xu (2015).
r r1 r2
Fig. 2 displays distributions of dE/df for μ = 0.1, e = 0.4, z = 0,
The total energy of the test particle with respect to the barycentre and different f in the dimensionless rotating frame. Dashed lines
in the inertial frame E = K + U, where K = V2 /2 is the kinetic in these figures are lines of x = 0.5 − μ and y = 0. As we can
energy. Using equations (1), (7) and (8), and the definitions of g and see from the figures, in most region of the orbital plane, values of
A, we can derive dE/df are very close to zero, and only in the neighborhoods of m1
and m2 values of dE/df suddenly surge, which is the mechanical
dE dK dU reason of the flyby or gravity assist. Based on equation (11), dE/df
= +
df df df is determined by many parameters, such as μ, e, f, and the position
G(m1 + m2 ) (x, y, z). As we can see from Fig. 2, f can influence the values
= (y x − x y) + A(x x + y y + z z)
r and distributions of dE/df. For example, magnitudes of |dE/df| of
− A (x 2 + y 2 + z2 ) + rU f = 0o in the neighborhoods of m1 and m2 are larger than those of
f = 180o . Compared with results of f = 0o and 180o , distributions
= (Ux y − Uy x) − A(Ux x + Uy y + Uz z) − AU
 1 of f = 90o and 270o seem to be twisted along different directions in
G(m1 + m2 ) 1 the neighborhoods of m1 and m2 .
= μ(1 − μ) y 3
− 3
r r1 r2 In an attempt to locate E-increasing regions with dE/df > 0 and E-

x+μ x+μ−1  decreasing regions with dE/df < 0 in the orbital plane, distributions
+A − . (9) of ln |dE/df| in the dimensionless rotating frame are displayed in
r13 r23
Fig. 3. Therefore, dark red lines denote the curves of dE/df = 0,
where x , y , and z are the partial derivatives of  with respect and divide E-increasing and E-decreasing regions. By looking at the
to the position variables; Ux , Uy , and Uz are the partial derivatives figures, distributions of ln |dE/df| are symmetric around the point
of U with respect to the position variables. (0.5 − μ, 0, 0), and E-increasing regions are distributed at the

MNRAS 478, 1392–1402 (2018)


Energy analysis in ERTBP 1395

Downloaded from https://academic.oup.com/mnras/article/478/1/1392/4992760 by Ryerson University Library user on 14 November 2020


Figure 4. Distribution of dE/df of spatial ERTBP in the dimensionless
rotating frame for μ = 0.1, e = 0.4, and f = 90o .

Figure 2. Distribution of dE/df in the dimensionless rotating frame.

Figure 5. Distribution of ln |dE/df| of spatial ERTBP in the dimensionless


rotating frame for μ = 0.1, e = 0.4, and f = 90o .

slices of images, we find that distributions of dE/df are symmet-


ric about the orbital plane (or xy plane). Distributions of dE/df in
horizontal planes with different z have similar patterns but different
magnitudes. In general, for a point with the given x and y, the value
of |dE/df| on the orbital plane (or xy plane) is maximum. From
Fig. 5, surfaces of dE/df = 0 pass through m1 and m2 transversely.
When f = 90o , there are transverses of surfaces of dE/df = 0 on the
yz plane.

4 DISCUSSION

Figure 3. Distributions of ln |dE/df| of planar ERTBP for μ = 0.1, e= 0.4, 4.1 Energy analysis
and different f.
In order to solve the extremum of dE/df with respect to f, we take
upper left and lower right parts of the xy plane, but E-decreasing the derivative of equation (11) with respect to f,
regions are distributed at the upper right and lower left parts of d2 E eμ(1 − μ)
the xy plane. According to equation (11), when f = 0o and 180o , = [P2 cos f − P1 sin f ] . (13)
df 2 1 − e2
points of dE/df = 0 are distributed in lines of x = 0.5 − μ and  
y = 0, which can be seen in Fig. 3, and distributions of ln |dE/df| Let cos α = P1 / P12 + P22 and sin α = P2 / P12 + P22 , then the
are symmetric around lines of dE/df = 0. When f = 90o , as we can above equation can be rewritten as
see from Fig. 3, curves of dE/df = 0 are twisted, so the E-increasing 
d2 E eμ(1 − μ)
region with dE/df > 0 is connected, but two E-decreasing regions = − P12 + P22 sin (f − α). (14)
with dE/df < 0 are totally separated by the E-increasing region. df 2 1 − e2
However, results of f = 270o are just the opposed results of f = 90o . Therefore, dE/df has an extreme value at f∗ = α + kπ, k ∈ N. f∗
From Fig. 3, curves of dE/df = 0 pass through points (− μ, 0, 0) is determined by the position and μ, but e has no influence on f∗ .
and (1 − μ, 0, 0) no matter what f is. Fig. 6 shows distribution of f∗ in the dimensionless rotating frame
Spatial distributions of dE/df and ln |dE/df| for μ = 0.1, e = 0.4, for μ = 0.1, e = 0.3, and z = 0. f∗ in the left figure corresponds to
and f = 90o are shown in Figs. 4 and 5, respectively. By observing the maximum of dE/df, i.e. (dE/df)max , while f∗ in the right figure

MNRAS 478, 1392–1402 (2018)


1396 Y. Qi and A. de Ruiter

Downloaded from https://academic.oup.com/mnras/article/478/1/1392/4992760 by Ryerson University Library user on 14 November 2020


Figure 6. Distributions of f∗ in the dimensionless rotating frame for μ = 0.1 and e = 0.3.

corresponds to the minimum of dE/df, i.e. (dE/df)min . Dashed lines


in figures are lines of x = 0.5 − μ and y = 0. As we can see
from the figures, distributions of f∗ are symmetric around the point
(0.5 − μ, 0, 0). In the neighborhoods of m1 and m2 , contour lines
of f∗ converge at points (−μ, 0, 0) and (1 − μ, 0, 0).
Next, we investigate the distribution of dE/df in the neigh-
 of m1 and m2 . For theplanar ERTBP, z = ż = 0, so
bourhood
r1 = (x + μ)2 + y 2 and r2 = (x + μ − 1)2 + y 2 . Based on
equation (11), curves of dE/df = 0 can be obtained from
   
dE μ(1 − μ)g 1 1 x+μ x+μ−1
= y − +A −
df 1 − e2 r13 r23 r13 r23
= 0. (15)
The above equation can be simplified as
r23 [A(x + μ) + y] = r13 [A(x + μ − 1) + y] . (16)
As mentioned in Fig. 3, curves of dE/df = 0 pass through m1 and
m2 . In addition, we find that curves of dE/df = 0 in the neighbour-
hood of m1 and m2 can be approximated by tangent lines at (−μ, 0)
and (1 − μ, 0). Taking the derivative of equation (16) with respect
to x, then we can obtain Figure 7. Distribution of ln |dE/df| of planar ERTBP for μ = 0.1, e = 0.4,
and f = 90o .
3r22 r2x [A(x + μ) + y] + r23 (A + yx )
= 3r12 r1x [A(x + μ − 1) + y] + r13 (A + yx ), (17)
where the subscript ()x denotes differentiation with respect to x, (1 − μ, 0). Obviously, based on equation (18), γ is only determined
such as yx = dy/dx, and r1x = dr1 /dx. by e and f. When f = 0o and 180o , γ = 0o that accord with the
When r1 → 0, then r2 → 1, A(x + μ) + y → 0, and A(x + μ − 1) + y results in Fig.3. Taking the derivative of equation (18) with respect
→ −A. From equation (17), we can get the slope of the tangent line to f, we can easily find that γ reaches its extreme values when
at (− μ, 0), limr1 →0 yx = −A. Similarly, when r2 → 0, we can also f = f̃ = arccos(−e) + kπ, k ∈ √ N. Then, the maximum √ and mini-
get the slope of the tangent line at (1 − μ, 0), limr2 →0 yx = −A mum of γ are equal to arctan(e/ 1 − e2 ) and -arctan(e/ 1 − e2 ),
from equation (17). For example, Fig. 7 shows the distribution of respectively. Hence, we conclude that the twist angles of curves of
ln |dE/df| for μ = 0.1, e = 0.4, and f = 90o . Dashed lines are tangent dE/df = 0 increase with the increase of e.
lines at (1 − μ, 0) and (−μ, 0), and dotted lines are x-axis and y-
axis. We use the angle γ to represent the angle between tangent
lines and x-axis. According to the character of the elliptic orbit in 4.2 Expressions of energy change
the two-body model (Chobotov 2002), we can obtain the following
relationship: When we focus on the flyby problem near the smaller or secondary
primary m2 , a polar coordinate system centred at m2 is presented
e sin f vr
tan γ = −yx = A = = , (18) to depict flyby orbits (see Fig. 8), where ψ is the polar angle with
1 + e cos f vu respect to the polar axis, i.e. the x-axis, and the distance to m2 is the
where vr is the radial velocity component of the elliptic orbit of radius, i.e.r2 . Then, we can get x = 1 − μ + r2 cos ψ, y = r2 sin ψ,
m2 around barycentre O, and vu is the normal velocity component and r1 = 1 + r22 + 2r2 cos ψ.
(Chobotov 2002). Therefore, the inertial velocity of m2 in the el- In the neighbourhood of m2 , we postulate that r1 ≈ 1, x + μ ≈
liptic orbit v in is perpendicular to the tangent line of dE/df = 0 at 1, and x + μ − 1 = r2 cos ψ. Besides, according to equation (18),

MNRAS 478, 1392–1402 (2018)


Energy analysis in ERTBP 1397

Downloaded from https://academic.oup.com/mnras/article/478/1/1392/4992760 by Ryerson University Library user on 14 November 2020


Figure 8. Schematic of a flyby orbit in the polar coordinates centred at m2 .

equation (11) can be approximated by


    Figure 9. Schematic of a prograde flyby orbit in the neighbourhood of m2 .
dE μ(1 − μ)g 1 r2 cos ψ
≈ r 2 sin ψ 1 − + tan γ 1 −
df 1 − e2 r23 r23 If e in the ERTBP is small, we consider that flyby orbits in the

μ(1 − μ)g sin ψ + tan γ cos ψ neighbourhood of m2 are similar to those in the CRTBP, so we also
= r2 sin ψ + tan γ − assume that the trajectory orbit during a flyby in the rotating frame
1 − e2 r22
 can be approximated by a hyperbolic orbit of the inertial frame, that
μ(1 − μ)g sin (ψ + γ )
= r2 sin ψ + tan γ − . (19) is to say, equation (24) is still available in the ERTBP when e is
1 − e2 r22 cos γ small. For the given state of periapsis withrespect to m2 , (x, y, x , y ),
Compared with ‘1/r22 ’, the terms of ‘r2 sin ψ’ and ‘tan γ ’ are both in the dimensionless rotating frame, v0 =  x 2 + y 2 is the velocity
small quantities, hence, the above equation can be further simplified of periapsis with respect to m2 , and D = (x + μ − 1)2 + y 2 is
into the distance of periapsis from m2 , so equation (23) can be expressed
dE μ(1 − μ)g sin (ψ + γ ) as
=− . (20)  ψ2
df (1 − e2 ) cos γ r22 E2 − E1 = ∓
μ(1−μ)g
sin (ψ +γ )dψ
(1−e2 ) cos γ v0 D ψ1
In Fig. 8, v = (x  , y  ) is the velocity in the rotating frame. vu2 and
vr2 are normal and radial velocity components with respect to m2 , μ(1−μ)g
=± cos (ψ2 + γ )−cos (ψ1 +γ ) , (25)
then we can get (1−e2 ) cos γ v0 D
r2 dψ where signs + and − correspond to the prograde and retrograde
df = ± , (21) motions with respect to m2 , respectively.
vu2
When a test particle passes though a circular neighbourhood of m2
where the sign + corresponds to the prograde motion (anticlockwise with a radius R, ψ 1 = ψ 0 ± θ 1, and ψ 2 = ψ 0 ± θ 2 , where ψ 0 is the
with respect to m2 ), and the sign − corresponds to the retrograde polar angle of periapsis, θ 1 and θ 2 are true anomalies of the approx-
motion (clockwise with respect to m2 ). imate hyperbolic orbit on the boundary of circular neighbourhood.
Therefore, equation (21) can be rewritten as In addition, signs + and − correspond to prograde and retrograde or-
μ(1 − μ)g sin (ψ + γ ) bits. For example, Fig.9 shows a trajectory passing by m2 along the
dE = ∓ dψ, (22) prograde motion. Since the trajectory is approximated by a hyper-
(1 − e2 ) cos γ vu2 r2
bolic orbit, we can get θ2 = −θ1 = arccos [(D(1 + eh )/R − 1)/eh ],
where g, γ , r2 , and vu2 are functions of f, and based on equation (21), where eh is the eccentricity of the approximate hyperbolic orbit.
those variables can be expressed as functions of ψ. For a trajectory Therefore, the change of E after the flyby is
passing by m2 , the terms ‘g’ and ‘γ ’ are slow variables with respect
2μ(1 − μ)g
to ψ, but the terms ‘vu2 ’ and ‘r2 ’ are fast variables with respect to ψ. E = E2 − E1 = − sin θ2 sin (ψ0 + γ ).
Hence, for a flyby orbit in the neighbourhood of m2 , we postulate (1 − e2 ) cos γ v0 D
that g and γ can be regarded as constants. Then, we can calculate (26)
the expression of E by √
Based on (Qi & Xu 2015), v0 = V0 ∓ D = μ(1 + eh )/D ∓ D,
 E2  ψ2
μ(1 − μ)g sin (ψ + γ ) where V0 is the inertial velocity of periapsis with respect to m2 ,
dE = ∓ dψ. (23) and signs + and − correspond to retrograde and prograde motions,
E1 (1 − e 2 ) cos γ
ψ1 vu2 r2
respectively. We use f0 to denote the value of f when the test particle
According to Qi & Xu (2015), in the Earth–Moon CRTBP, if is located at the periapsis with respect to m2 that can be regarded as
the lunar flyby occurs in a short period of time, the trajectory near the time of periapsis. Then, g and γ can be obtained from e and f0 .
the Moon in the rotating coordinates can be approximated by a In addition, as mentioned before, θ 2 is determined by D, eh , and R.
hyperbolic orbit of the inertial coordinates. Based on the conserva- Therefore, based on equation (26), E actually is a function E(μ,
tion of angular momentum in the two-body model, they derived the e, D, ψ 0 , f0 , R, eh ). If μ, e, D, f0 , R, and eh are fixed, from equation
following relationship: (26), E reaches its maximum at ψ 0 = 270o − γ ; E reaches its
minimum at ψ 0 = 90o − γ ; E = 0 at ψ 0 = −γ or 180o − γ . From
vu2 r2 ≈ v0 D, (24)
equation (26), the smaller D the larger | E| is. Therefore, if we
where v0 is the velocity of perilune in the Earth–Moon rotating desire to give a full play to a flyby or gravity assist, the altitude of
frame, and D is the distance of perilune from the Moon’s centre. the periapsis should be as small as possible. Similarly, the smaller

MNRAS 478, 1392–1402 (2018)


1398 Y. Qi and A. de Ruiter
Table 1. ERTBP system in the Solar system.

System μ e Rsoi /a

Earth–Moon 0.012151 0.0549 0.172172


Sun–Saturn 2.893267 × 10−4 0.0565 0.038424
Sun–Mars 3.253253 × 10−7 0.0935 0.002541
Sun–Mercury 1.681354 × 10−7 0.2056 0.001951
Sun–Plutoa 6.68893 × 10−9 0.2488 5.372083 × 10−4
Note: a It should be noted that the barycentre of the Pluto-system is outside

Downloaded from https://academic.oup.com/mnras/article/478/1/1392/4992760 by Ryerson University Library user on 14 November 2020


Pluto because of the mass ratio between the planet and its moon Charon
which is unique in the Solar System. However, in the following study of the
Sun-Pluto ERTBP, Charon will not be taken into consideration, because the
complicated application for the actual Pluto-system is beyond the scope of
this paper.

Figure 10. Distribution of η for μ = 0.012151, D = 0.0172172,


R = 0.172172, and eh = 2.5.

v0 or eh the larger | E| is, but we require that eh should be larger


than 1, otherwise, the approximate orbit is not hyperbolic and the
above expressions will be invalid.
To find the application condition of equation (26), we present
an index η to evaluate the relative error between the real energy
change in the circular neighbourhood Ereal and the result of the
above approximate expression E. Assuming that values of μ, D,
R, and eh are given,
max | Ereal − E|
ψ0 ∈[0,2π ]
η(e, f0 ) = , (27)
max | Ereal |
ψ0 ∈[0,2π ]

where Ereal = E2 − E1 is obtained by numerical integration in the


ERTBP, and E is calculated from equation (26).
Fig. 10 shows the distribution of η for prograde flyby orbits with
D = 0.0172172, R = 0.172172, and eh = 2.5. The value of μ is set as Figure 11. Distribution of E for retrograde flyby orbits with different
the mass ratio in the Earth–Moon system, 0.012151. In this paper, position of periapsis.
we require that η cannot exceed 5 per cent, otherwise, we consider
that equation (26) will be invalid. As we can see from the figure, the mass ratio μ of the Earth–Moon system is the largest. Hence,
when e < 0.3, ηs for all f0 are smaller than 5 per cent. More results based on the discussion in the previous section, we conclude that
of η with different μ can be obtained using equation (27), but they the obtained approximate expressions in the previous section are
are not displayed herein for the sake of space. Further, numerical applicable to the flyby problem in the Solar system. Table 1 lists
results illustrate that distributions of η with different μ have similar data of ERTBP systems with e > 0.05, where Rsoi is the radius of
patterns, but the dark red regions with η > 0.1 gradually enlarge the sphere of influence, and it can be obtained from Bate, Mueller
and go down with the increase of μ. Therefore, the upper boundary & White (1971)
of the feasible e for equation (26) decreases with an increase of μ.  
m2 2/5 μ 2/5
Rsoi = a =a . (28)
m1 1−μ
5 A P P L I C AT I O N S I N T H E S O L A R S Y S T E M
Taking the Sun–Pluto ERTBP as an example, Fig. 11 displays
In this section, the approximate expressions of energy obtained in the distribution of E in the neighbourhood of Pluto for retrograde
the previous section will be applied to investigate the flyby problem flyby orbits with f0 = 90o , eh = 2.0, R = Rsoi /3a, and different
in the Solar system. Parameters, especially the position of periapsis position of periapsis. E is calculated from equation (26). As we
(D, ψ), and the time of periapsis f0 , will be analysed in detail. can see from Fig. 11, E reaches its maximum and minimum at
In this paper, for the flyby problem or gravity assists in the Solar ψ 0 = 270o − γ and 90o − γ , respectively, and the line of E = 0
system, we focus on celestial bodies with large masses and exclude passes through Pluto, with angle relative to the x-axis is equal to γ .
small bodies, such as comets, asteroids, and meteoroids, because In addition, the smaller D the larger | E| is. Those results accord
the effects of those small bodies on the path and energy of the with our previous analysis about equation (26). Based on Fig. 7, the
test particle are rather slight, even though most of small bodies are inertial velocity of m2 in the elliptic orbit v in is perpendicular to the
running in elliptic orbits. According to the Planetary Fact Sheet line of E = 0 (see Fig. 11). Consequently, we conclude that a flyby
of NASA Space Science Data Coordinated Archive (Update: 2017 with energy loss occurs when 0o < ψ 0 + γ < 180o , i.e. the periapsis
June 7; Williams 2017), for the nine planets and the Moon in the is in front of the motion of m2 ; a flyby with energy gain occurs
Solar system, their orbital eccentricities are all smaller than 0.3, and when 180o < ψ 0 + γ < 360o , i.e. the periapsis is behind the motion

MNRAS 478, 1392–1402 (2018)


Energy analysis in ERTBP 1399

Downloaded from https://academic.oup.com/mnras/article/478/1/1392/4992760 by Ryerson University Library user on 14 November 2020


Figure 12. Curves of E versus ψ 0 for prograde flyby orbits with different
Figure 13. Comparison of the results between equation (26) and equation
f0 s and ERTBP systems.
(29).

of m2 . Those conclusions are consistent with the results of Ferreira


et al. (2018), but the latter was derived from the patched-conic
model rather than the ERTBP. According to Fig.11, the distribution
of E is obviously rotated by γ . Based on equation (18), γ is
determined by e and f0 , and reaches its maximum or minimum
at f0 = f̃ = arccos(−e) + kπ, k ∈ N. In particular, γ = 0 in the
CRTBP, so regions of energy gain or loss are divided by the x-axis
that is consistent with the results of other studies (Broucke 1988;
Qi & Xu 2015).
Fig. 12 shows curves of E versus ψ 0 for prograde flyby orbits
with different f0 s and ERTBP systems, where dashed lines corre-
spond to results of numerical integration in the ERTBP, and solid
lines correspond to the calculation results from equation (26). In
the numerical computation, eh = 2.5, D = Rsoi /10a, and R = Rsoi /a.
Data of the ERTBP systems are directly obtained from Table 1, and Figure 14. Curves of E versus f0 for prograde flyby orbits with different
since the eccentricities of the Earth–Moon system and Sun–Saturn ψ 0 s in the Sun–Pluto ERTBP.
system are very close, we do not display the result of the Sun–
Saturn system for the sake of space. In Fig. 12, four values of – where
f0 , 0o , 180o , arccos(−e), and arccos(−e) + 180o – are chosen and 
μ(eh − 1)
corresponding results are denoted by different colours. Based on Vinf = ,
D
the analysis in Section 4.1, we can get the maximum twist angle γ
1 1
or maximum horizontal shift of curves when f0 = arccos(−e) and sin δ = 2 = ,
arccos(−e) + 180o . As we can see from figures, curves of E ob- 1+
DVinf e h
μ
tained from the expression (26) are very close to the corresponding π
results of numerical integration in the ERTBP that testify the appli- β = + γ.
2
cability of our approximate expression (26). From the Earth–Moon
Taking the Sun–Pluto ERTBP as an example, the comparison of
system to the Sun–Pluto system, with the increase of e, variations
the results between equation (26) and equation (29) is illustrated
of amplitudes and phases (i.e. horizontal shifts) of curves of E
in Fig. 13, where red dashed lines, blue, and green solid lines cor-
among different f0 s both gradually enlarge.
respond to the results of numerical simulation, equations (26), and
Ferreira et al. (2018) also derived an approximate expression of
(29), respectively. As we can see from the figure, except f0 = 180o ,
E of a flyby orbit in the ERTBP using the patched-conic method:
our results are closer to the numerical simulation than those from
  equation (29).
2g Fig. 14 shows curves of E versus f0 for prograde flyby orbits
E = 2Vinf (1 − μ) − 1 sin δ cos(ψ0 + β), (29)
1 − e2 with different ψ 0 s in the Sun–Pluto ERTBP, where dashed lines

MNRAS 478, 1392–1402 (2018)


1400 Y. Qi and A. de Ruiter
correspond to results of numerical integration in the ERTBP, and
solid lines correspond to the calculation results from equation (26).
Results of different ψ 0 s are distinguished by colours. In the numer-
ical computation, eh = 2.0, D = Rsoi /8a, and R = Rsoi /3a. In Fig. 14,
the positions of extreme points are shown by vertical dashed lines.
Similarly, we observe that curves of E obtained from the expres-
sion (26) are very close to the corresponding results of numerical
integration in the ERTBP, especially when ψ 0 = 0o and 180o . As we
can see from the figure, f0 has a significant influence on E for the

Downloaded from https://academic.oup.com/mnras/article/478/1/1392/4992760 by Ryerson University Library user on 14 November 2020


given D and ψ 0 , i.e. the given position of the periapsis. In addition,
the position of f0 corresponding to the maximum or minimum E
depends on ψ 0 .
For a given position of periapsis with respect to m2 , we can locate
the position of f0 corresponding to the maximum or minimum E
by
 
d E 2μ(1 − μ) sin θ d g sin (ψ0 + γ )
=−
df0 (1 − e2 )v0 D df0 cos γ Figure 15. Curves of Emax , Emin , and E(f0 , ψ 0 ) with different f0 s in
2μ(1 − μ)e sin θ the Sun–Pluto ERTBP.
=− cos (f0 + ψ0 ). (30)
(1 − e2 )v0 D
Hence, when f0 = kπ/2 − ψ 0 , k ∈ N, E reaches its maximum
or minimum. On the other hand, in the neighbourhood of m2 , based
on equation (10) and definitions of cos α and sin α in Section 4.1,
we can derive
P1
lim cos α = lim  = − sin ψ0 ,
r2 →0 r2 →0 P12 + P22
P2
lim sin α = lim  = − cos ψ0 , (31)
r2 →0 r2 →0 P12 + P22
therefore α tends to −π/2 − ψ 0 . Then, according to the anal-
ysis in Section 4.1, the extreme points of dE/df are located at
f∗ = α + kπ = kπ/2 − ψ 0 , k ∈ N, which can be verified by Fig. 6.
Therefore, we conclude that at a given position of periapsis, when
dE/df reaches its maximum or minimum at f0 = f∗ , the correspond-
ing E also reaches its maximum or minimum. This conclusion
can also be testified by comparing Figs 6 and 14. Furthermore, for
given values of e, D, R, and eh , we can obtain


Emax = max E(f0 , ψ0 ) = E − ψ0 , ψ0 , Figure 16. Curves of Emax and Emin for prograde flyby orbits in differ-
f0 ∈[0,2π ] 2
π  ent ERTBP systems.
Emin = min E(f0 , ψ0 ) = E − ψ0 , ψ0 . (32)
f0 ∈[0,2π ] 2
the figures, it can be seen that with the increase of e, the differ-
Fig. 15 shows curves of Emax , Emin , and E(f0 , ψ 0 ) for pro-
ence between Emax and Emin is more obvious. In addition, the
grade flyby orbits with different f0 s in the Sun–Pluto ERTBP, where
dashed lines correspond to results of numerical integration in the maximum and minimum of both Emax and Emin are located at
ψ 0 = 270o and ψ 0 = 90o , respectively. The maximum of Emax
ERTBP, and solid lines correspond to the calculation results from
and the minimum of Emin can, respectively, be regarded as the
equation (26). Curves of Emax and Emin are denoted by red and
blue lines, respectively. Green lines represent curves of E(f0 , ψ 0 ) global maximum and minimum of E for different f0 s and ψ 0 s (see
Fig. 15). Therefore, we conclude that for a flyby orbit passing by
with different f0 . In the numerical computation, eh = 2.5, and D
m2 , the global maximum energy gain occurs when ψ 0 = 270o and
and R are set as Rsoi /10a and Rsoi /a, respectively. As we can see
from the figure, curves of Emax and Emin are the upper and lower f0 = 270o − ψ 0 = 0o ; the global maximum energy loss occurs when
ψ 0 = 90o and f0 = 90o − ψ 0 = 0o .
envelopes of E(f0 , ψ 0 ), respectively. In addition, we observe that
Let E0 denote the inertial energy at the periapsis. Based on equa-
curves of Emax and Emin obtained from the expression (26) well
approximate corresponding results of numerical integration in the tion (25), E in the neighbourhood of m2 can be expressed as,
ERTBP. μ(1 − μ)g
E = E0 ± cos (ψ0 ± θ + γ )
Fig. 16 displays curves of Emax versus ψ 0 and Emin versus (1 − e2 ) cos γ v0 D
ψ 0 for prograde flyby orbits in different ERTBP systems, where
− cos (ψ0 + γ )
dashed and solid lines correspond to results of numerical integra-
μ(1 − μ)g
tion in the ERTBP and computational results from equation (26), = E0 − sin (ψ0 + γ ) sin θ
respectively. In the numerical computation, eh = 2.5, and D and (1 − e2 ) cos γ v0 D
R are also set as Rsoi /10a and Rsoi /a, respectively. According to ± cos (ψ0 + γ )(1 − cos θ) , (33)

MNRAS 478, 1392–1402 (2018)


Energy analysis in ERTBP 1401

Downloaded from https://academic.oup.com/mnras/article/478/1/1392/4992760 by Ryerson University Library user on 14 November 2020


Figure 17. Change curves of E for prograde flyby orbits with different ψ 0 s, f0 s, and ERTBP systems.

where prograde and retrograde motions correspond to signs + and were affected by the eccentricity e. f corresponding to the extremum
−, respectively. θ is the true anomaly of the approximate hyper- of dE/df and the tangent lines of curves dE/df = 0 passing through
bolic orbit. The functions sin θ (f) and cos θ (f) can be obtained by m1 and m2 were solved analytically. In addition, an approximate
the classical formulas of the two-body model (Chobotov 2002). expression of energy change E in a circular neighbourhood of m2
Readers can find the detail in Appendix A. Therefore, the approx- was derived. Numerical computation indicated that for the ERTBP
imate expression of the energy E as a function of f can be ob- with e < 0.3 and μ ≤ 0.012151, the relative errors of our equations
tained by equation (33). Since θ (f0 ) = 0, E(f0 ) = E0 . It should with respect to the numerical simulation in the ERTBP were smaller
be noted that, for a flyby orbit in the neighbourhood of m2 , we than 5 per cent.
still postulate that g and γ are constants and can be obtained from Furthermore, the obtained above expressions were applied to
e and f0 . study the flyby problem in the ERTBP of the Solar system. Parame-
Fig. 17 shows change curves of E for prograde flyby orbits in a ters related to the flyby were discussed analytically and numerically.
circular neighbourhood of m2 with different ψ 0 s, f0 s, and ERTBP The optimal position and time of the periapsis during a flyby were
systems, where dashed and solid lines correspond to results of nu- found to make a maximum energy gain or loss. The analytical re-
merical integration in the ERTBP and computational results from sult showed that for a flyby orbit passing by m2 with a given phase
equation (33), respectively. In the numerical computation, eh = 2.0, angle of periapsis ψ 0 , when dE/df of the periapsis reaches its max-
D = Rsoi /10a, and the radius of the circular neighbourhood is Rsoi /3a. imum at the time of periapsis f0 = 270o − ψ 0 , the corresponding
Results with different f0 s are distinguished by different colours. E also reaches its maximum; when dE/df reaches its minimum
From figures, we find that the change curves of equation (33) well at f0 = 90o − ψ 0 , the corresponding E also reaches its mini-
approximate numerical integration results of the ERTBP that tes- mum. In addition, the global maximum energy gain occurs when
tifies the applicability of our derivation. As we can see from the ψ 0 = 270o and f0 = 0o ; the global maximum energy loss occurs
figures, for the same ψ 0 but different f0 s, change curves of E differ when ψ 0 = 90o and f0 = 0o . Finally, the mechanical process of
in their positions and shapes, and a larger e results in more obvi- a flyby orbit in the ERTBP was uncovered by an approximate ex-
ous differences. Using expression (33), we unveil the mechanical pression that successfully fills this research gap in previous studies.
process of a flyby orbit of the ERTBP in detail. These results fill a Numerical computations testified that our analytical results well
research gap of the ERTBP for a flyby orbit in a small region near approximate the mechanical process of flyby orbits obtained by
m2 . numerical simulation in the ERTBP. The results obtained in this
paper could be used in the future deep space missions, especially
interplanetary missions.
6 CONCLUSIONS
In this paper, the gravity assist or flyby problem in the ERTBP was
AC K N OW L E D G E M E N T S
investigated by analysing the inertial energy E. The expression of
the derivation of E with respect to the true anomaly f was obtained This work is supported by the Natural Sciences and Engineering
in the ERTBP. The detailed analysis uncovered that, different from Research Council of Canada through a Discovery Accelerator Sup-
results in the CRTBP, distributions of dE/df were time varying and plement under Grant RGPAS-493042-2016.

MNRAS 478, 1392–1402 (2018)


1402 Y. Qi and A. de Ruiter
REFERENCES APPENDIX A: SOME EXPRESSIONS FOR A
HYPERBOLIC ORBIT
Bate R. R., Mueller D. D., White J. E., 1971, Fundamentals of Astrodynam-
ics. Dover Press, New York For a hyperbolic orbit in the two-body model, if its eccentricity eh
Broucke R., 1969, AIAA J, 7, 1003 and distance of periapsis D are given, the semimajor axis ah can be
Broucke R., 1988, in Astrodynamics Conference. p. 4220 expressed by
Campagnola S., Boutonnet A., Schoenmaekers J., Grebow D. J., Petropoulos
A. E., Russell R. P., 2014, J. Guid. Control Dyn., 37, 1202 Dr
ah = , (A1)
Casalino L., Colasurdo G., Pastrone D., 1999, J. Guid. Control Dyn., 22, eh − 1
156

Downloaded from https://academic.oup.com/mnras/article/478/1/1392/4992760 by Ryerson University Library user on 14 November 2020


Chobotov V., 2002, Orbital Mechanics, American Institue of Aeronautics where r = p/g is the distance between m1 and m2 . For an approxi-
and Astronautics, 1801 Alexander Bell Drive, Reston, Virginia 20191- mate hyperbolic orbit in the neighbourhood of m2 , we consider that
4344 g = 1 + e cos f0 , so r is a constant.
Ferreira A. F., Prado A. F., Winter O. C., Santos D. P., 2017, Adv. Space Based on equation (6), the time t and the variable f of the ERTBP
Res., 59, 2071 has an approximate equation as follows,
Ferreira A. F., Prado A. F., Winter O. C., Santos D. P., 2018, Ap&SS, 363, 
24 p3
Hou X., Liu L., 2011, MNRAS, 415, 3552 t − t0 = (f − f0 ). (A2)
g2
Hyeraci N., Topputo F., 2010, J. Guid. Control Dyn., 33, 1814
Then, the mean anomaly M of the hyperbolic orbit is related to f
Longuski J. M., Williams S. N., 1991, Celest. Mech. Dyn. Astron., 52, 207
by
Peng H., , Bai X., , 2018, Acta Astronaut.
Peng H., Xu S., 2015a, Adv. Space Res., 55, 1015    3
μ μ p
Peng H., Xu S., 2015b, Celest. Mech. Dyn. Astron., 123, 279 M = (t − t 0 ) = (f − f0 )
Peng H., Xu S., 2015c, Ap&SS, 357, 87
ah3 ah3 g 2

Peng H., Qi Y., Xu S., Li Y., 2015, in AAS/AIAA Spaceflight Mechanics μ(eh − 1)3
Conference. Williamsburg, VA, p. 1 = (f − f0 ). (A3)
Peng H., Bai X., Masdemont J. J., Gómez G., Xu S., 2017a, J. Guid. Control gD 3
Dyn., 40, 3155
Since the flyby orbit is hyperbolic with respect to m2 , we can get
Peng H., Bai X., Xu S., 2017b, Commun. Nonlinear Sci. Numer. Simul., 47,
1
Kepler’s equation in the form of
Prado A. F., 1997, J. Guid. Control Dyn., 20, 797
M = eh sinh F − F , (A4)
Prado A. F., Neto E. V., 2006, J. Astronaut. Sci., 54, 567
Qi Y., Xu S., 2015, Ap&SS, 360, 55 where F is the eccentric anomaly. Using Newton–Raphson method,
Qi Y., Xu S., 2016, Aerosp. Sci. Technol., 55, 282
F can be solved from equation (A4) for the given eh and M.
Qi Y., Xu S., 2017, Acta Astronaut., 134, 106
Qi Y., Xu S., Qi R., 2014, in 24th International Symposium on Space Flight
Finally, using the eccentric anomaly F obtained from the above
Dynamics, Baltimore, Maryland Kepler’s equation, we can derive
Qi Y., Xu S., Qi R., 2017a, Acta Astronaut., 133, 145 
Qi Y., Xu S., Qi R., 2017b, MNRAS, 468, 2265 eh2 − 1 sinh F
sin θ = ,
Sims J. A., Longuski J. M., Staugler A. J., 1997, J. Guid. Control Dyn., 20, eh cosh F − 1
409 eh − cosh F
Strange N. J., Longuski J. M., 2002, J. Spacecr. Rockets, 39, 9 cos θ = . (A5)
eh cosh F − 1
Szebehely V., 1967, Theory of Orbit: The Restricted Problem of Three
Bodies. Academic Press, New York
Williams D. R., 2017, NASA Space Science Data Coordinated Archive,
Available at: https://nssdc.gsfc.nasa.gov/planetary/factsheet/ This paper has been typeset from a TEX/LATEX file prepared by the author.

MNRAS 478, 1392–1402 (2018)

You might also like