JOURNAL OF PROPULSION AND POWER

Vol. 24, No. 6, November–December 2008

Opportunities for a Liquid Rocket Feed System

Based on Electric Pumps

N. Soldà∗ and D. Lentini†

University of Rome “La Sapienza,” 00184 Rome, Italy
DOI: 10.2514/1.35074
A feed system for liquid-propellant rocket engines based on electric pumps powered by batteries is proposed. It is
proven to stand as a viable alternative to the pressure-gas feed system. The dependence of the feed system mass on the
different operating parameters is obtained so as to identify the conditions favoring its adoption, that is, a relatively
long burning time and a fairly high chamber pressure. Under such conditions, the proposed system is shown to offer
significant mass savings with respect to the pressure-gas system when advanced batteries are used. This advantage is
further enhanced by the beneficial effect of chamber pressure on the engine effective exhaust velocity. A test case for a
low Earth orbit to geostationary equatorial orbit transfer is also presented to identify the optimum value of the
burning time, deriving from the competition between the feed system mass and the effect of gravitational losses.

Nomenclature t = stress admitted by tank wall material

CF = thrust coefficient
= tank wall thickness
C1 , C2 = constants defined by Eqs. (12) and (13)
c = effective exhaust velocity Subscripts
c
= characteristic velocity
b = batteries
D1 –D8 = constants defined by Eqs. (31–38)
c = combustion chamber
Ee = electric energy
ep = electric pumps
f, g = functions defined by Eqs. (14) and (39)
eps = electric pump system
Mg = pressurizing gas molar mass
f = fuel
m = mass
fs = feed system
O=F = oxidizer-to-fuel mass ratio
g = pressurizing gas
Pe = electric power
min = minimum
p = pressure
o = oxidizer
Qf , Qo = quantities defined by Eqs. (21) and (22)
p = propellant
R0 = universal gas constant
pgs = pressure-gas system
r = tank radius
t = tank
T = absolute temperature
0 = initial conditions in pressure-gas tank
tb = burning time
tba , tbp = apogee and perigee burning times
V = volume
v = velocity I. Introduction

,
f ,
o = quantities defined by Eq. (4)
g

va ,
vb ,
vtot
=
=
pressurizing gas specific heat ratio
apogee, perigee, and total velocity
C URRENT feed systems for liquid rocket engines are either
turbopump fed or pressure-gas fed [1,2]. It is well known that
the former ensures a lightweight design, yet with the drawbacks of
increment mechanical complexity and the ensuing limited reliability (e.g., a
E , P = battery energy and power densities large fraction of launch failures are due to malfunctioning turbopump
t = density of tank wall material systems), long development times, and high costs; further,
ep = overall efficiency of electric pumps restartability is limited. On the other hand, the latter system is
b = safety factor for batteries conceptually simple and ensures unlimited restartability, but is very
g = safety factor for pressurizing gas mass heavy owing to the tanks being under pressure; further, development
p1 , p2 , p3 = pressure ratios defined by Eqs. (2), (16), and may be lengthy, despite its apparent simplicity.
(17) A third option based on electric feed pumps powered by batteries
t = safety factor for tank wall thickness is proposed here. It is apparent that the bottleneck in this case is the
u = quantity defined by Eq. (3) mass of the batteries, which is dictated by the most stringent one of
 = payload mass ratio two requirements: 1) ensure the electrical power Pe required by the
ep = electric pump specific mass pumps, and 2) ensure the electrical energy Ee required to drive the
 = fluid density pumps throughout their operating time tb . Given the relatively
limited power and energy density of today’s batteries, it is felt that the
proposed system cannot compete with a turbopump-fed system (e.g.,
Received 10 October 2007; accepted for publication 5 July 2008. as applied in the booster stages). However, the comparison is found
Copyright © 2008 by the American Institute of Aeronautics and Astronautics, to be favorable with respect to the pressure-gas system, that is, in
Inc. All rights reserved. Copies of this paper may be made for personal or applications to upper stages and spacecraft propulsion systems.
internal use, on condition that the copier pay the $10.00 per-copy fee to the
Consequently, the comparison reported herein is restricted to just the
Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA
01923; include the code 0748-4658/08 $10.00 in correspondence with the
pressure-gas and electric pump systems. Derivations aimed at
CCC. identifying the mass of the feed system for both the standard
∗
Research Student. pressure-gas system and the proposed system are presented in Sec. II.
† Then, a comparison of the two systems is given in Sec. III, including
Associate Professor, Dipartimento di Meccanica e Aeronautica, Via
Eudossiana 18; diego.lentini@uniroma1.it. Senior Member AIAA. a test case of a low Earth orbit (LEO) to a geostationary equatorial
1340
 SOLDÀ AND LENTINI 1341

orbit (GEO) transfer. Finally, conclusions about the possible Here, t;g is the maximum stress admitted by the wall material; in
convenience of the proposed feed system are drawn in Sec. IV. particular, it refers to the yield stress for plastic materials and to the
ultimate stress for fragile materials. Accordingly, the mass of the
pressurizing gas tank turns out to be
II. Evaluation of Feed System Mass
A. Pressure-Gas System 3 t;g mp pc
mt;g 
p1 u g t;g g (8)
The main components of the total mass mpgs of a feed system using 2 t;g 1  p1 pc =p0
pressurized gas are the mass mg of the pressurizing gas and the
masses of the tanks for the gas itself, the fuel and the oxidizer, where t;g is the density of the wall material. Similarly, for the
denoted as mt;g , mt;f , and mt;o , respectively: propellant tanks, still assumed to be spherical, it results in

mpgs  mg  mt;g  mt;f  mt;o (1) 3 t;p

mt;f 
f p1 u t;p m p (9)
2 t;p p c
The mass of the plumbing is not included in Eq. (1) for mpgs ;
however, it will certainly be greater for the system presently under
3 t;p
consideration as compared with the system proposed herein in view mt;o 
o p1 u t;p m p (10)
of the higher pressures involved. Therefore, neglecting it stands as a 2 t;p p c
conservative assumption in the comparison of the two feed concepts
in Sec. III.
We consider a regulated pressure system and, for the sake of Incidentally, nonspherical tank shapes would entail larger wall
conciseness, we further assume that the pressure inside the tanks is thicknesses due to stress concentrations [1]. Notice that in pressure-
the same for the two propellants, pt;f  pt;o (this assumption, as well fed systems the tanks are thick walled. After Eq. (1), the total mass of
as similar ones that will be introduced next, can be easily relieved if such a system results in
necessary). We accordingly introduce a ratio between the propellant 
 
Mg 3 t;g 1
tank pressure and the chamber pressure: mpgs  g 0  t;g 
R T0 2 t;g g 1  p1 pc =p0
pt;f pt;o 
p1   (2) 3 t;p
pc pc  t;p
p1 u mp pc (11)
2 t;p
We then introduce, to account for ullage, a ratio between the tank
Such an expression can be put in a more synthetic form by defining
volume and the (initial) propellant volume, which we also assume to
the constants:
be the same for the two propellants:


Vt;f Vt;o Mg 3 t;g
C1  g 0  t;g
g p1 u (12)
u   (3) R T0 2 t;g
Vf Vo

Further coefficients to be defined to account for design margins are a 3 t;p

factor g multiplying the pressure-gas mass strictly required, a safety C2 
t;p p1 u (13)
2 t;p
factor t;g relating to the thickness of the wall of the pressure-gas
tank, and a similar factor t;p for the propellant tanks (again, assumed which allow, in particular, the recasting of the expression of the ratio
to be the same for both propellants). of mpgs to the propellant mass in the following form:
The different items at the right-hand side of Eq. (1) will now be

determined. Let mp denote the total (oxidizer plus fuel) propellant mpgs C1
mass and O=F the oxidizer-to-fuel mass ratio. Then let g and Mg   C2 pc  fpc ; p0  (14)
mp 1  p1 pc =p0
indicate the specific heat ratio and the molar mass of the pressurizing
gas, T0 and p0 its initial temperature and pressure, and R0 the with emphasis on the arguments.
universal gas constant. Upon introducing the positions
B. Electric Pump System
1 1 O=F 1

f  ;
o  ;

f 
o The main components of the total mass meps of the proposed feed
f 1  O=F o 1  O=F
system based on electric pumps are, again, a (small) pressurizing gas
(4) mass required to avoid cavitation, the masses of the tanks for the gas
itself, the fuel and the oxidizer, plus the masses mep and mb of the
the required mass of the pressurizing gas can be determined after [2], electric pumps (here taken inclusive of the power conditioning/
when the margins defined above are taken into account, as
controller unit mass) and the batteries, respectively:
g Mg mp pc meps  mg  mt;g  mt;f  mt;o  mep  mb (15)
mg 
p1 u g (5)
R0 T0 1  p1 pc =p0
We again assume that the ratio, here denoted as p2 (obviously,
As far as the tank masses are concerned, they can be expressed as p2  p1 ), between the pressure in the tanks and in the chamber is
the wall material density times the tank wall surface times the wall the same for both propellants:
thickness. By assuming spherical tanks, the radius of the pressurizing
gas tank can be related to its volume Vt;g and then to the gas mass as pt;f pt;o
p2   (16)

1=3
  1=3 pc pc
3 3 mg 1=3 3 mg
rt;g  Vt;g   (6) and further that the pressure rise provided by the pumps is also the
4 4 0 4 p0 Mg =R0 T0 
same for both propellants, so that a ratio
where 0 denotes the initial gas density. The wall thickness is given
pf
po
by Laplace’s law, multiplied by the relevant safety factor t;g : p3   (17)
pc pc
 1=3
rt;g p0 t;g p0 3 mg

g  t;g  (7) can be defined. Actually, if one of the propellants is used for
2 t;g 2 t;g 4 p0 Mg =R0 T0  regenerative cooling (usually fuel), it must undergo a higher pressure
1342 SOLDÀ AND LENTINI

rise to compensate for losses in the cooling channels; this can be As far as the batteries are concerned, their mass depends on whether
easily accommodated in the formulation if necessary, but at this stage the constraint on the power or the energy is more stringent:
it is not considered an essential point. The masses of the various

entries of Eq. (15) will now be derived. For mg and mt;g , expressions P E
mb  b max e ; e (30)
analogous to Eqs. (5) and (8) hold, but with p2 in place of p1 . As far P E
as the propellant tanks are concerned, the wall thickness required is,
in principle, again given by Laplace’s law, for example, for fuel and where a factor b is introduced to account for the design margins. We
oxidizer, respectively, further introduce the following constants:

1=3

p2 t;p 3u
f Mg 3 t;g

f  pc mp (18) D1  g 0  t;g
g p1 u (31)
2 t;p 4 R T0 2 t;g

1=3 3 t;p
p2 t;p 3u
o D2 
f t;p p2 u (32)

o  pc mp (19) 2 t;p
2 t;p 4

However, in the case of pump feeding, the tanks are thin walled, D3  41=3 3u
f 2=3
min t;p (33)
and the thickness as defined by Eqs. (18) and (19), may turn out to be
too small to withstand acceleration loads. To account for this
circumstance, a minimum thickness
min is introduced; it can be
easily seen that the condition that causes the wall thickness to stick at 3 t;p
D4 
o t;p p2 u (34)
this limit is 2 t;p

pc mp1=3 < Qf for fuel; pc mp1=3 < Qo for oxidizer (20)

where D5  41=3 3u
o 2=3

min t;p (35)


2 t;p
min 4 1=3
Qf  (21)
p2 t;p 3u
f
p3 ep
D6  (36)
ep


2 t;p
min 4 1=3
Qo  (22)
p2 t;p 3u
o

b p3
D7  (37)
ep P
Accordingly, if pc mp1=3 > Qf , then the fuel tank mass is
3 t;p
mt;f 
f p2 u t;p m p (23)
b p3
2 t;p p c D8  (38)
ep E
and if pc mp1=3 > Qo , then the oxidizer tank mass is After Eq. (15), the ratio of the total mass of the electric pump system
3 t;p to the propellant mass can be put in the following form:
mt;o 
o p2 u t;p m p (24)

2 t;p p c meps D1 D3
 pc  max D2 pc ; 1=3
mp 1  p2 pc =p0 mp
However, in the event that pc mp1=3 < Qf and/or pc mp1=3 < Qo , the


relevant tank masses are determined as D5 pc Dp
 max D4 pc ; 1=3  D6  max 7 c ; D8 pc
mp tb tb
mt;f  41=3 3u
f 2=3
min t;p mp2=3 (25)
 gpc ; p0 ; mp ; tb  (39)

mt;o  41=3 3u
o 2=3

min t;p mp2=3 (26)
III. Comparison of the Two Systems
As already indicated, the mass of the batteries in the proposed
The definition of the masses of the electric pumps and the batteries system depends on either the energy or the power requirements,
is now considered. The electric energy required over the whole according to which is the most stringent. For a given battery, the
engine operating time tb is values of the power density P and the energy density E depend on

p3 the discharge time, which in the present context is the engine burning
Ee  m p (27) time tb ; such a relationship is given by Ragone’s plot [3,4]. Figure 1
ep p c
shows such a plot for commercially available, premium Li-polymer
where ep is the overall efficiency of the conditioning/controller and batteries,‡ which are representative of the best technology currently
electric pump unit, and the power is obviously available. Notice that the line interpolating the data is limited to a
minimum discharge time of about 12.5 min; accordingly, this is the
Pe  Ee =tb  (28) minimum burning time considered in the comparison. It is also
important to note that the choice of energy storage device is not
Accordingly, the mass of the electric pumps can be determined in limited to batteries, but might also include supercapacitors and fuel
terms of their specific mass ep (mass per unit electric power cells. However, the former suffer from limited energy density
absorbed):
‡
Data available online at www.symmetryresourcesinc.com [retrieved on
mep  ep Pe  ep Ee =tb  (29) 26 August 2008].
 SOLDÀ AND LENTINI 1343

POWER AND ENERGY DENSITIES RATIO FEED SYSTEM / PROPELLANT MASS

1000
900 0.15

in
m
800

0
60
700

=
600

tb
500
0.12
400
300
in
m

in
m
δP (W/kg)

0
18

0.09

18
=

200

mfs / mp
tb

tb
0.06
100
in
m
60

in
m
=

6
tb

=
tb
0.03

200 400 600 8001000 0

0 1 2 3 4 5
δE (W.h/kg) pc (MPa)
Fig. 1 Ragone plot of premium Li-polymer batteries. Fig. 2 Ratio of feed system mass over total propellant mass as a
function of chamber pressure for p0
20 MPa and tb
1000 s. Solid
line: pressure-gas system; dashed line: proposed system mp
1000 kg;
whereas the latter, despite giving a performance even higher than long-dashed line: mp
3000 kg; and extra-long-dashed line:
batteries on a mass basis, result in a much lower volumetric mp
10; 000 kg.
performance, unless the reactant storage pressure is assumed to be
greater than about 60 MPa [5],§ which is about 20 times as much as
current technology cells. Further, fuel cells require the handling of
cryogenic reactants (oxygen and hydrogen), which can be RATIO FEED SYSTEM / PROPELLANT MASS
inconvenient for small upper stages and spacecraft.
Quite obviously, the preference for one system or the other 0.15
(turbopumps or electric pumps) depends to a critical extent on the
values of several performance parameters absorbed in the constants
C1 , C2 , and D1 –D8 . To work out a comparison of the two systems, we 0.12
consider a nitrogen tetroxide/monomethylhydrazine (NTO/MMH)
propellant combination in a ratio that will ensure equal volume tanks.
Further, we assume helium as the pressurant gas, stored at ambient 0.09
mfs / mp

temperature. As far as the tank material is concerned, Kevlar® is

chosen for the pressurant gas tank, whereas aluminum alloy is
assumed for the propellant tanks. Accordingly, the following values 0.06
are assumed in the present calculations:

p1  1:8 p2  0:3 p3  1:5

0.03
g  1:3 t;g  2:4 t;p  1:25
u  1:05 b  1:2
min  1 mm
0
0 10 20 30 40
g  1:667 Mg  4:0026 kg=kmol T0  288:15 K p0 (MPa)
o  1431 kg=m3 f  874 kg=m3 O=F  o =f Fig. 3 Ratio of feed system mass over total propellant mass as a
function of initial helium pressure for pc
3 MPa and tb
1000 s. Solid
t;g  1700 kg=m3 t;p  2800 kg=m3 line: pressure-gas system; dashed line: proposed system mp
1000 kg;
long-dashed line: mp
3000 kg; and extra-long-dashed line:
t;g  3300 MPa t;p  455 MPa
mp
10; 000 kg.
ep  0:68 ep  6 kg=kWe

Notice that the values assumed for the parameters are quite two feed systems under consideration. The curves for the electric
conservative, aimed at avoiding a bias toward one particular feed pump system, referring to the aforementioned values of mp , exhibit a
system. separate, nearly parallel, trend for relatively low values of pc ,
As seen in Eq. (14), the mass of a pressure-gas feed system, indicating that the factor limiting the wall thickness is the constraint
divided by the overall propellant mass, depends on the chamber

min . However, for chamber pressures slightly above 3.5 MPa,
pressure pc and the initial gas pressure p0 . The mass of the proposed the curves for mp  3000 and 10,000 kg coalesce into a single curve,
system based on electric pumps (39) also depends on pc and p0 and, as under such circumstances the wall thickness is greater than
min , so
in addition, on the burning time and the propellant mass itself in the that the walls are actually sized according to Laplace’s law. The
case that the tank thickness is dictated by the constraint not to be figure further suggests that, for pc greater than about 5 MPa, the
smaller than a given value
min . We will therefore first examine the curve for mp  1000 kg will also coalesce with the others. It is worth
trend of the ratio between the feed system mass mfs and the propellant noting that the separate trends for the three curves follow from the
mass mp as a function of the chamber pressure for a value of p0 set to assumption of the same minimum thickness for the three values of
20 MPa (see also the comment on Fig. 3 in the next paragraph), a mp under consideration; if the minimum thickness is instead
burning time set to 2000 s, and for three values of the propellant mass, assumed to be proportional to the cubic root of mp , the variable mp is
1000, 3000, and 10,000 kg. Such trends are shown in Fig. 2 for the dropped from the arguments of function g in Eq. (39). In summation,
Fig. 2 indicates that the proposed feed system is preferable when the
§
Data available online at gltrs.grc.nasa.gov [retrieved on 26 August 2008]. chamber pressure is above roughly 1 MPa.
1344 SOLDÀ AND LENTINI

RATIO FEED SYSTEM / PROPELLANT MASS EFFECTIVE EXHAUST VELOCITY

0.15 3250

0.12
3200

0.09

cvac (m/s)
mfs / mp

3150
0.06

3100
0.03

0 3050
0 1000 2000 3000 0 2 4 6 8 10
t b (s) pc (MPa)
Fig. 4 Ratio of feed system mass over total propellant mass as a
Fig. 5 Theoretical effective exhaust velocity of an isovolumetric NTO/
function of burning time for pc
3 MPa and p0
20 MPa. Solid line:
MMH propellant combination as a function of chamber pressure.
pressure-gas system; dashed line: proposed system mp
1000 kg; long-
dashed line: mp
3000 kg; and extra-long-dashed line: mp

10; 000 kg.
To further quantify the advantage associated with the proposed
system, an example is presented of the case of a propulsion system
Figure 3 shows the effect of the initial gas pressure p0 on the ratio tailored for a Hohmann transfer, that is, one aimed at injecting a
mfs =mp for the two feed systems and the three values of mp being payload initially in an equatorial LEO with a 300-km-altitude into a
considered for a chamber pressure set at 3 MPa and the same burning geostationary transfer orbit (GTO), and then circularizing the latter
time as already mentioned. It is seen that, for the pressure-gas system, into a GEO. The ideal (i.e., referring to an impulse maneuver)
a high initial pressure is desirable; however, the advantage becomes velocity increment requirements are 2426 m=s for the perigee kick
marginal over 20–25 MPa. For the proposed system, p0 turns out to and 1466 m=s for the apogee burn. For a finite time perigee burn, the
be rather inconsequential, as the pressurization is limited to the small
v requirements increase due to gravitational losses. This effect is
amount required to avoid pump cavitation. estimated here via a computational code based on assumptions
The effect of the burning time (for pc  3 MPa and adequate for the present level of analysis (tangential thrusting,
p0  20 MPa) on the ratio mfs =mp is depicted in Fig. 4. The range spherical Earth, neglect of aerodynamic drag effects, neglect of the
of burning times is bound by the minimum discharge time for the influence of third bodies, etc.), and the results are shown in Fig. 6,
batteries at hand, which can be estimated from Fig. 1 to be about based on an assumed value of the (real) effective exhaust velocity of
750 s, whereas the upper value is set to 3000 s because relatively little 3000 m=s, as appropriate to the NTO/MMH propulsion system
advantage is obtained by further prolonging the duration of the burn. outlined earlier (gravitational losses weakly depend on the effective
Notice that, in the case of multiple engine ignitions, here tb has the exhaust velocity). It is seen that the perigee velocity increment
vp
meaning of the cumulative burning time. It is clearly visible that, increases with the duration of the perigee burn tbp ; however, the value
although tb does not affect mfs =mp for the pressure-gas system, it of the required apogee velocity increment
va decreases, because
does have the effect of reducing the relative feed system mass for the longer perigee burning times result in GTOs of smaller eccentricity.
proposed system. In fact, once the propellant mass is assigned, a Notice that, in Fig. 6, the
va are computed by neglecting apogee
longer burning time implies a reduced propellant mass flow rate, that gravitational losses, an assumption justified in view of the very high
is, a reduced power (and then mass) of the electric pumps, and
possibly of the batteries, if their mass is power constrained (see also
∆v FOR LEO-GEO TRANSFER
Fig. 9).
As mentioned with reference to Fig. 2, a quantity critical to 4200
assessing the advantage of the proposed system over the pressure-gas ∆v tot
one is the chamber pressure, with high values of pc favoring the
proposed system, which is an obvious consequence of the tanks not 3600
being under pressure. Such a system would then be particularly
valuable for high-chamber-pressure engines, which exhibit the
additional advantage of a larger effective exhaust velocity (i.e., 3000
∆v (m/s)

specific impulse). Indeed, both the thrust coefficient CF and the

∆vp
characteristic velocity c
increase with increasing pressure. Figure 5
illustrates the trend of the ideal effective exhaust velocity (in a
2400
vacuum, for a nozzle expansion ratio of 40) cvac  CF c
as a function
of chamber pressure for a similar test case using an isovolumetric
NTO/MMH propellant combination, as computed via the
equilibrium chemical solver CEA [6,7]¶; to take some account of 1800
finite rate chemistry, the flow is assumed to suddenly freeze at the
throat, although, more precisely, the impact of finite rate chemistry ∆v a
depends on the nozzle size [8]. Accordingly, the gain in effective 1200
1000 1500 2000 2500 3000
exhaust velocity adds up to that of a reduced feed system mass, with tbp (s)
an ensuing beneficial effect on the payload mass.
Fig. 6 Velocity increments required for the perigee and apogee
maneuvers of a transfer between a 300-km-altitude LEO and GEO, and
¶
Code available online at www.grc.nasa.gov [retrieved on 26 August 2008]. total
v, as a function of perigee burning time (assuming c
3000 m=s).
 SOLDÀ AND LENTINI 1345

PAYLOAD RATIO FOR GTO INJECTION PAYLOAD RATIO FOR GEO INJECTION
0.34 0.22
0.2
0.18
0.32
0.16
0.14
0.3
0.12

λGEO
λGTO

0.1
0.28
0.08
0.06
0.26 0.04
0.02

0.24 0
500 1000 1500 2000 2500 3000 0 1000 2000 3000
tbp (s) tbp (s)

Fig. 7 Payload ratio for injection into GTO as a function of perigee Fig. 8 Payload ratio for injection into GEO as a function of perigee
burning time (assuming c
3000 m=s) with the proposed feed system. burning time (assuming c
3000 m=s) with the proposed feed system.
Dashed line: mp
1000 kg; long-dashed line: mp
3000 kg; and extra- Single-stage option—dashed line: mp
1000 kg; long-dashed line:
long-dashed line: mp
10; 000 kg. mp
3000 kg; and extra-long-dashed line: mp
10; 000 kg; two-stage
option— dashed-dotted line: mp
1000 kg; long-dashed-dotted line:
mp
3000 kg; and extra-long-dashed-dotted line: mp
10; 000 kg.
altitude of the GEO. However, the two opposing trends do not cancel
out, resulting in a
vtot 
vp 
va still mildly increasing with FEED SYSTEM MASS BREAKDOWN
tbp . In the following paragraphs, the effect of the perigee burning time 900
on two payload ratios, one referring to the first phase of the
maneuver, accordingly defined as 800

mGTO 700
GTO  (40)
mLEO
600
mass (kg)

and a second referring to the whole transfer 500

m
GEO  GEO (41) 400
mLEO
300
is evaluated. Here mLEO is the initial mass in LEO, whereas mGTO and
mGEO clearly stand for the payload mass in the respective orbits. The 200 BATT
ER
P

IE S
former payload ratio is shown in Fig. 7 as a function of the perigee
R
E
S

burning time. The payload ratio considered in Fig. 7 and the figures 100
S
U

ELE CTR IC PU MP
R

S
AN

that follow is recovered by including in the stage structural mass the TANKS
T→

0
feed system mass as detailed earlier and an extra mass indicatively 0 1000 2000 3000 4000
made equal to 4% of the propellant plus pressurant mass to account tb (s)
for other structures. Further, a 3% propellant margin is assumed. Two a)
effects contribute to the determination of the shape of the curve: the
reduction of the feed system mass with the burning time already 100
assessed in Fig. 4 (resulting in a larger payload), and the increase of
the effective
vp with tbp . Their combined effect determines a 90
maximum around a perigee burning time of 1900 s. Such a maximum 80 BAT TERIES
spans a narrow range (from about 0.3316 to 0.3355) depending on
the assumed propellant mass. 70
To arrive at the overall payload ratio GEO for the whole transfer,
60
% mass

two propulsive options are considered: 1) a single-stage, restartable

system, performing both the perigee and the apogee burns; and 2) a 50
two-stage system (one to be fired near the perigee, the other near the ELECTRIC PUMPS

apogee of the GTO) with independent engines. Both options are 40

assumed to be fed via the proposed system. Results are shown in 30
Fig. 8 for both options. In the single-stage option (dashed lines), the TANKS
overall burning time tb is split between the time taken by the perigee 20
burn tbp and that referring to the apogee burn tba ; because of the
10
aforementioned limitations in the discharge rate of the batteries under PRESSURANT
consideration, the cumulative burning time tbp  tba must be greater 0
0 1000 2000 3000 4000
than 750 s (unless oversizing the batteries) and less than 3000 s. tb (s)
However, no constraint applies as far as the split of the total burning
time tb over the burns. It follows that tbp can, in principle, span the b)
whole range of 0–3000 s (although small values will result in high Fig. 9 Mass breakdown of the electric pump feed system for single-
accelerations and, in addition, heavy thrusters). However, in the stage injection into LEO (assuming mp
3000 kg and c
3000 m=s):
present case, it is found that perigee burning times shorter than about a) absolute terms, and b) percentage terms.
1346 SOLDÀ AND LENTINI

550 s (corresponding to a total perigee-plus-apogee burning time of
b p3 mp pc

mb  (43)
about 750 s) result in a zero payload ratio, due to an excessive feed ep E
system mass for short burning times (see also the comment on Fig. 9
below). Notice that, for tbp longer than about 2190 s (indicated by the where the only dependence on the burning time stems from the
dotted line in Fig. 8), the overall burning time tb exceeds the limit of energy density. It can be seen from Fig. 1 that E actually slightly
3000 s, which, at any rate, has been chosen merely to identify a decreases as the burning time increases (leading to a somewhat
burning-time range and does not imply any physical bound. The higher battery mass), but the effect becomes appreciable only for
maximum payload ratio is obtained for a perigee burning time of burning times exceeding a few hours. The mass breakdown is
around 2400 s and spans the narrow range of 0.171–0.176 as the total reported in percentage terms in Fig. 9b, indicating that, although the
propellant mass (burnt in the two kicks) ranges from 1000 to battery mass is the dominating item for relatively short burning
10,000 kg. times, it plays a less crucial role for the long values of the burning
In the two-stage option (dashed-dotted lines), both tbp and tba time as advocated.
contribute in principle to determining the overall payload ratio.
However, because the apogee kick gravitational losses are assumed IV. Conclusions
to be negligible (at least for burning times limited to a maximum of
3000 s) and, accordingly, the
va to be provided by the second stage The plots worked out indicate that there is an interesting field of
is held fixed, it is apparent that it is convenient to choose as long a tba application, typified by relatively high chamber pressures and
as possible (i.e., 3000 s under the present assumptions), as this burning times, for a feed system for liquid-propellant rockets based
minimizes the feed system mass (see Fig. 4). Accordingly, only tbp , on the use of electric pumps powered by batteries. Although the
that is, the burning time of the first stage, remains as a free parameter proposed system cannot compete with more complex turbopump
(in this case limited to a minimum of 750 s because of the systems for high-performance boosters, it nonetheless represents a
aforementioned limit in the discharge rate of the batteries). Because viable alternative to pressure-fed systems; it presents an advantage
the payload ratio of the second stage is roughly fixed (having fixed not only in terms of the feed system mass, but also of the higher
the
va required and the burning time, it spans the range of 0.568– effective exhaust velocity that can be attained by raising the level of
0.602 depending on the propellant mass of the second stage, which in the chamber pressure, an option inapplicable to pressure-fed
turn depends on tbp and the propellant mass of the first stage, as systems. Like the pressure-gas system, the proposed system ensures
extended restartability (basically limited solely by battery life,
inferred by Fig. 7), the trend of the relevant curves in Fig. 8 follow
which, however, is on the order of several years). It is also expected
that of the GTO payload ratio given in Fig. 7 (notice the different
that the weaker coupling between the feed system and the chamber
scale of the two figures). The maximum values of the payload are
may alleviate combustion instabilities. Further, if secondary (i.e.,
obtained for burning times around 2300 s and span the range of
rechargeable, albeit with a lower performance) cells, rather than
0.194–0.197 as the propellant mass of the first stage ranges from
primary cells as assumed in the present study, are used, the battery
1000 to 10,000 kg. Accordingly, the two-stage option ensures a
mass can be accounted for as payload mass in missions requiring
limited advantage with respect to the single-stage one on the order of
extensive energy storage, such as in the case of a lunar surface probe
12–13% on a relative basis, which has to be weighted against the
(owing to the lengthy lunar night).
accompanying increase in complexity and cost, and the reduced
reliability.
Finally, a mass breakdown of the electric pump system is shown in References
Fig. 9 for the case of the single-stage LEO to GEO transfer with a [1] Humble, R. W., Lewis, D., and Sackheim, R., “Liquid Rocket
propellant mass of 3000 kg. In this case, results are plotted as a Propulsion Systems,” Space Propulsion Analysis and Design, edited by
function of the overall burning time (perigee plus apogee) to R. W. Humble, G. N. Henry, and W. J. Larson, McGraw–Hill, New
highlight the impact of the cumulative burning time. The masses of York, 1995.
the pressurant gas and of the tanks (for the propellant and pressurant) [2] Sutton, G. P., and Biblarz, O., Rocket Propulsion Elements, 7th ed.,
are fixed at 9.3 and 43.7 kg, respectively, for a total of 53 kg. The Wiley, New York, 2001, pp. 203–226.
[3] Bockris, J. O’M., and Reddy, A. K. N., Modern Electrochemistry 2B,
electric pump mass decreases from 136 to 25 kg as the burning time 2nd ed., Kluwer/Plenum, New York, 2000, p. 1856.
spans the range of 750–4070 s (which, incidentally, corresponds to [4] Christen, T., and Carlen, M. W., “Theory of Ragone Plots,” Journal of
the aforementioned range of perigee burning time of 548–3000 s), Power Sources, Vol. 91, No. 2, 2000, pp. 210–216.
consistent with Eq. (29). The mass of the batteries also decreases doi:10.1016/S0378-7753(00)00474-2
from 629 to as few as 29 kg in the same overall burning-time range. [5] Burke, K. A., “Fuel Cells for Space Science Applications,” NASA TM-
This striking result is due to the fact that, up to a total burning time of 2003-212730, 2003; also AIAA Paper 2003-5938, 2003.
about 3700 s, the battery mass turns out to be power constrained, so [6] Gordon, S., and McBride, B., “Computer Program for Calculation of
that, in view of Eqs. (27), (28), and (30), it results in Complex Chemical Equilibrium Compositions and Applications—I.
Analysis,” NASA RP 1311-I, 1994.

b p3 mp pc [7] McBride, B., and Gordon, S., “Computer Program for Calculation of
mb  (42) Complex Chemical Equilibrium Compositions and Applications—II.
ep P tb
Users Manual and Program Description,” NASA RP 1311-II, 1996.
The first term is a constant, whereas the denominator of the second [8] Lentini, D., “Identification of Chemical and Vibrational Relaxation
Regimes in Rocket Nozzle Flow via a Quasi-Linear Formulation,”
term features, in addition to a direct dependence upon the overall Proceedings of the Institution of Mechanical Engineers. Part G,
burning time, an indirect dependence through the power density P , Journal of Aerospace Engineering, Vol. 215, No. 2, 2001, pp. 79–87.
which is seen from the Ragone plot in Fig. 1 to greatly increase with doi:10.1243/0954410011531781
discharge time. This further emphasizes that the proposed system is
particularly advantageous for long burning times. However, for a tb
greater than about 3700 s, the battery mass becomes energy D. Talley
constrained, resulting in the following expression: Associate Editor