You are on page 1of 4

IEEE TRANSACTIONS ON MAGNETICS, VOL. 37, NO.

1, JANUARY 2001 485

Simulation-Based Optimal Sizing of Hybrid Electric


Vehicle Components for Specific Combat Missions
Scott Fish, Member, IEEE, and Troy B. Savoie

Abstract—The U.S. Army and the Defense Advanced Research a relatively simple PPU/battery power architecture. A larger en-
Projects Agency (DARPA) are currently supporting the devel- ergy load representative of a medium caliber electromagnetic
opment of conceptual military vehicle designs capitalizing on (EM) railgun is also examined albeit with an associated pulsed
maturing technologies associated with hybrid electric vehicle
propulsion and advanced electrical loads. Preliminary trade alternator/PPU/battery architecture.
studies associated with these concepts often focus on determining Current sizing practice is based on estimates of the peak to
the “best” mix of power system components to achieve the desired average load power ratio expected for the vehicle application.
performance levels at a minimum overall impact on the vehicle. Estimation of these power levels often requires some form of
This paper describes an approach to this optimization through time-based simulation. When one includes pulsed loads, the
the use of vehicle performance simulation. By using simulation,
we can address the complex relationship between the mission mix of equipment ratings becomes even more complex partic-
needs and the required subsystem performance in a dynamic ularly when the power available is a function of the state of
environment. Results for three fictional mission sequences are charge of the energy storage devices. Simulation becomes es-
provided to show how simulations can be used to insure that sential in these cases evaluating the performance throughout the
complex performance goals are met with minimum subsystem mission sequence based on each event’s impact on subsequent
sizes. These minimum sizes can then form a baseline for further
sizing constraints based on arrangements, availability, and other operations.
design considerations while minimizing over design.
Index Terms—Electric weapons, hybrid electric vehicles, A. Simulation Method
optimization, simulation.
To determine the performance of candidate power system
combinations, a time-based simulation of the vehicle and its
I. INTRODUCTION power systems is executed. The simulation routine is written
in MATLAB/Simulink and includes scaleable models for the

V EHICLE designers have always looked for ways to meet


their performance requirements with a minimum system
burden. Because combat vehicles spend very little of their op-
engine, flywheel and battery in the same manner as described in
Fish and Savoie [1]. A control system is implemented to coordi-
nate the use of each power system element as a function of the
eration time at full power conditions, this has become a focus
phase of the mission being simulated. For instance, silent mo-
area for improving fuel efficiency, signature management, and
bility specifications will cause the control system to shut down
overall weight and volume. The underlying rationale for hy-
the engine and draw power only from batteries or flywheel. The
brid electric vehicles is that a smaller power supply (engine
propulsion system architecture assumed here is a series hybrid
or fuel cell) can be coupled with an energy storage compo-
with a permanent magnet generator on the engine and induction
nent during brief periods of high power demand. At other times,
motors driving the axles.
the power supply can be operated more near its most efficient
point at reduced overall power loading. In the cases we examine
here, the power supply will be an engine/generator which we B. Optimization Criteria
will refer to as the prime power unit or PPU. Energy storage The simulation is executed within an optimization shell
systems using flywheels and batteries have received consider- which automates the search through multidimensional pa-
able attention recently, and are included in the examples pro- rameter space for engine/flywheel/battery combinations and
vided here. The availability of the electrical bus has enabled the minimize the objective function input by the user. The objective
simpler integration of other electrical loads in particular pulsed function used in this study places heavy penalty on delays in
power loads. In this paper an electrothermal-chemical (ETC) weapon use to insure that surviving solutions will meet the
gun load is included to show how a small energy loading affects mission requirements in this area. Mobility performance is cal-
culated by the integral of the speed error from the mission input
Manuscript received January 24, 2000. This work was supported in part by the speed profile and has moderate weighting to insure that only
U.S. Army Electric Gun Program and the DARPA CHPS program through the slight performance degradation in acceleration/deceleration can
Army Research Laboratory, Weapons Technology Directorate, DAAA21-93-C- be accommodated. Weight and volume are also estimated for
0101.
S. Fish is with the Institute for Advanced Technology, The University of Texas the vehicle and used to balance the optimization from selecting
at Austin, Austin, TX 78759 USA (e-mail: Scott_Fish@iat.utexas.edu). very large components. The weight and volume of the power
T. B. Savoie was with the Institute for Advanced Technology, The University system components are estimated from nominal energy and
of Texas at Austin, Austin, TX 78759 USA. He is now with MIT, Cambridge,
MA 02139 USA (e-mail: Savoie@mit.edu). power densities, and added to a base vehicle weight in each
Publisher Item Identifier S 0018-9464(01)00209-6. mission category.
0018–9464/01$10.00 © 2001 IEEE
486 IEEE TRANSACTIONS ON MAGNETICS, VOL. 37, NO. 1, JANUARY 2001

Fig. 1. Terrain elevation.

The power system parameters varied for this study were en-
gine power rating, battery capacity, and flywheel capacity. The
engine data are derived from a baseline 1000 hp diesel design
[2]. The ratio of flywheel power to energy was fixed for sim-
plicity at 0.01 W/J, which is within the demonstrated design
space of existing flywheels [3]. The flywheel is assumed to have
Fig. 2. Objective function topology for reconnaissance mission.
a total rotor kinetic energy density of 40 kJ/kg which is also
within the state of the art when including power electronics as
well as the machine. The battery technology assumed is lithium The second mission is a strike mission for a medium weight
ion with an energy density of 98 Whr/kg commensurate with the vehicle of approximately 20 000 kg with a mobility sequence
state of the art demonstrated at the cell level. Additional devel- of 20 km over the same non level terrain followed by a 2 km
opment would be required to achieve this in a full pack config- silent mobility segment. The vehicle carries an ETC weapon re-
uration with thermal management [4]. quiring 0.5 MJ of energy from a capacitor based power supply
The optimization scheme includes constraints on the param- for each shot. The capacitor is recharged prior to each shot at a
eters to limit the space of the search to regions of interest. In rate of approximately 500 kW (complete charge supplied in one
practice, however, we insure that the optimum solutions do not second) from both the engine and battery. The firing sequence
lie on the boundaries of the parameter space. The optimiza- begins at the end of the silent mobility segment during an addi-
tion algorithm is based on sequential quadratic programming tional 1 km movement. Five volleys of 3 shots with 2 seconds
and implemented in MATLAB. This method is highly efficient between shots, and 15 seconds between volleys are executed.
while closely mimicking Newton’s method for unconstrained The last mission illustrated here is the same strike mission
optimization as described by Fletcher [5]. with an EM gun instead of the ETC gun. Because the EM gun
requires additional energy for each shot a flywheel based pulsed
II. MISSION SEQUENCES power supply has been assumed. The flywheel rotor speed is
reduced commensurate with an 8 MJ kinetic energy loss for each
To illustrate the simulation based optimization technique,
shot. Flywheel recharge from the battery and engine is invoked
three different mission sequences were composed. These
between shots.
mission sequences are fictional to avoid confusion with any
existing or future vehicle program or technology demonstra-
tion. The terrain over which these missions were executed is III. RESULTS
derived from real terrain in the central Texas region (elevation is 1) Mission 1—Reconnaissance: The reconnaissance mission
amplified by a factor of 4 to better represent difficult climbing objective was found to be optimized with a vehicle containing
missions) and is decomposed into an elevation as a function of the following ratings: engine 50 kW, battery 50 MJ (22 kWhr).
distance as shown in Fig. 1. The highest grades found in this The actual battery would be approximately 80 MJ to retain suf-
profile are 8% but most of the operation is in terrain of 2%. ficient capacity to operate in silent watch for 24 hours with
The movement from start to observation or firing positions an average load of approximately 300 W. Convergence of the
involves an average grade of 0.8%. For simplicity, the rolling optimization routine required 7 iterations with a resolution of
resistance is fixed at 0.25 N/kg (an average value for wheeled 10 kW on the engine power and 10 MJ on the battery pack. This
vehicles on hard soil) over the complete length of the mission. optimum was confirmed in a subsequent mapping of the engine
The grade force on the vehicle is calculated from the local and battery size parameter space as shown in Fig. 2.
terrain grade as a function of vehicle position. It should be noted that the engine size is quite a bit smaller
The first mission is assumed to be for a light vehicle of than one would customarily consider for a vehicle of this weight
approximately 3000 kg in a reconnaissance type operation. It class. Nevertheless, this size, in conjunction with the battery
begins with a 20 km movement over the terrain before shifting pack, can perform the mission sequence as shown. Should addi-
into a silent (engine off) mode of mobility for an additional tional constraints be placed on the engine, such as acceleration
5 km. The vehicle then makes the return trip in the same silent requirements or hill climbing without the battery, the engine size
and engine on modes. A battery margin of 10 kWhr is added would certainly increase. Doing so, however, negates some of
after the simulation to account for consumption in the silent the benefit of having a hybrid architecture in the first place. The
stationary mode of observation which is not simulated. performance of the best system is shown in Fig. 3 highlighting
FISH AND SAVOIE: SIMULATION-BASED OPTIMAL SIZING OF HYBRID ELECTRIC VEHICLE COMPONENTS 487

Fig. 3. Engine (top) and battery (bottom) usage in reconnaissance mission. Fig. 4. Engine (top) and battery (bottom) usage in ETC mission.

the tradeoffs in use of engine and battery at various points in the


mission.
The first graph of Fig. 3 shows how the PPU normally
operated at a constant power level. When the drive load is
greater than the PPU power, the battery is depleted. When the
drive load is less than the PPU power, the excess PPU power
is diverted into the battery to recharge. The simulation also
accounts for auxiliary loads such as cooling pumps and fans in
determining what power level the engine must operate. When
the vehicle goes into silent mobility mode at approximately
31 minutes, the engine is turned off, and all of the mobility
power must come from the battery. At 92 minutes, the engine is
restarted and the speed is increased from 10 km/hr to 40 km/hr.
The reduced drive power displayed on the return trip is a
reflection of the vehicle moving generally downhill to the
mission start point in the terrain.
The restarted engine accommodates some battery recharging Fig. 5. Battery voltage during ETC firing.
as well as the average mobility load as shown in the lower plot
of Fig. 3. The minimum battery SOC of is a reflection of In this case, we assume that the power to fire the gun is ex-
an operational constraint we place on the battery control system tracted from the kinetic energy in the rotor of a pulsed flywheel.
to retain good cycle life. The energy extracted per shot is 8 MJ. The recharge of the
2) Mission 2—ETC Gun: In the second mission, the battery flywheel is done with an electric motor powered by both the
and engine are required to not only satisfy mobility require- engine/generator and the battery between shots. Fig. 6 shows
ments, but also to fire a ETC weapon. The optimization for this the objective function progression, with a solution containing an
case returns a best engine size of 300 kW with a battery size engine rated for 650 kW, a battery rated for 145 MJ (40 kWhr),
of 55 MJ (15 kWhr). Fig. 4 shows how the battery and engine and a flywheel storing 90 MJ.
work together in this more complex mission sequence. Fig. 5 The difference in required drive motor power and PPU power
shows the detail of the battery voltage during the weapon oper- shown in the first graph of Fig. 6 shows that the control system
ations. Note from Fig. 4 that the engine in this period is moving is causing the battery to be used to moderate the load placed
into full power mode of operation. Although the battery output on the PPU. The effect of this load leveling on the battery state
power will normally be limited to prevent battery voltage drop of charge is clear from the second graph in Fig. 6. When the
below 225 volts, for pulse discharge (one second duration) we drive load power is greater than the PPU, the battery is depleted
have allowed the voltage to drop to 203 volts. If one wants to to compensate. When the PPU power is greater than the load
increase this minimum battery voltage, additional capacity will power, this energy is diverted to charge the battery. After ap-
be required. An estimate of the required capacity growth for proximately 31 minutes, the silent mobility phase begins and
this case based on an assumption of similar internal impedance all of the mobility power (as shown previously) must come from
would be 10%. the battery. At 44 minutes, the first shot is fired, and the engine
3) Mission 3—EM Gun: In the third mission, the vehicle size, is started for the combat phase. PPU power in this phase goes to
mobility, and weapon firing sequence are identical to Mission 2, full output to combine with the battery to provide the high rate
but the ETC gun is replaced with a medium caliber EM gun. of recharge power needed by the flywheel to meet the shooting
488 IEEE TRANSACTIONS ON MAGNETICS, VOL. 37, NO. 1, JANUARY 2001

as the use of ETC guns) in addition to that found in the recon-


naissance mission. The battery also reduced the required engine
power for weapon response. The short duration high power level
of 450 kW out of the battery is combined with the PPU output of
150 kW to provide the 500 kW recharge power for the ETC ca-
pacitor prior to each shot and the approximately 100 kW of drive
power for mobility at the time. Note that this study assumed a
lithium ion battery with significantly higher power density than
the typical lead acid battery.
In the EM weapon case, the optimization is increased in
importance due to the existence of two energy storage media
(the battery and the flywheel). In the implementation here,
the flywheel is focused on the weapon duty and the battery
does dual function as a mobility load leveler and as an aid in
recharge of the flywheel. Accurate sizing of the engine, battery
and flywheel under these conditions is greatly aided by the
Fig. 6. Engine (top) and battery and flywheel (bottom) usage in EM mission. use of simulation based optimization. A side benefit of this
technique is the highlighting of periods where components
are underutilized and thought can be given to improving the
utility of the overall system. In the EM case for instance, we
found that the flywheel was underutilized during the mobility
segments prior to the weapon usage. With good situational
awareness, one could use a small portion of the flywheel energy
to reduce transients on the batteries and thus increase their life
expectancy.
Clearly, selection of component sizes for actual design will
require more than one mission sequence for simulation. In fact,
the component ratings derived from these mission sequences
are much lower than those found in similar weight class
Fig. 7. Detail view of flywheel and battery for EM firing. vehicles in the field today. It is likely that the vehicle designer
will want to provide some level of margin to the minimums
sequence. An expanded view of the flywheel and battery state
suggested by simulation based optimizations to account for
of charge history during the firing sequence is shown in Fig. 7.
variations not covered in the mission sequences selected. The
Note the decrement of approximately 0.9 of full state of
simulation based method described here nevertheless provides
charge for each shot. Between shots, the power from the battery
good guidance from which to implement margins. The impor-
and PPU are used to “spin up” the flywheel and prepare it for
tance of having these baselines will ultimately be realized in the
subsequent firings. Because the flywheel is a relatively heavy
balance of costs associated with over sizing components and
element in the power system, the optimization process attempts
subsystems with the associated increase in mission flexibility.
to minimize its size. Its success in doing so can be seen by the
minimum SOC of 0.5 which is a control system limit imposed
to insure adequate speed in the flywheel to produce the high ACKNOWLEDGMENT
power pulses for firing. Use of the energy below 0.5 is feasible
for loads at lower power levels (like mobility load leveling) but The Institute for Advanced Technology thanks Dr. E.
were not investigated in this study. Schmidt, LTC Buck Tanner, and Dr. M. Freeman for reviewing
and providing technical comments on this paper.
IV. CONCLUSIONS
The results obtained for the three different missions indicate REFERENCES
the different modes of operation of a hybrid electric combat [1] S. Fish and E. Redding, “Prime power and pulsed energy storage for EM
vehicle in different applications. In all three missions studied gun equipped tank combat missions,” IEEE Trans. Magn., vol. 33, no.
here, the battery was used to load level the mobility system 1, January 1997.
[2] J. Hitchcock, “U.S. Army TACOM AIPS Technology as communicated
during “engine on” operations while providing silent mobility directly from TACOM,”, 1998.
power over periods when the vehicle is moving into a stealthy [3] G. Reiner, P. Ehrhart, M. Heeg, G. Heidelberg, J. Steffen, and W. Weck,
observation position. This use of the battery results in a sig- “Integration of a magnetrodynamic storage in an electric combat vehicle
and resulting system features,” in Proceedings of AECV Conf., Detroit,
nificant reduction in engine power required compared with a MI, 1997.
mechanical drive equivalent. Note that the mechanical drive ve- [4] G. Chagnon, S. Oweis, T. Sack, A. Romero, and L. d’Ussel, “High
hicle would not have had the option of silent mobility. power lithium-ion batteries,” in 15th Intl. Elect. Veh. Symp., Brussels,
September 1999.
The ETC strike mission showed an additional role of the bat- [5] R. Fletcher, Practical Methods of Optimization: John Wiley and Sons,
tery for relatively low energy and moderate power loads (such 1980, vol. 2, Constrained Optimization.

You might also like