You are on page 1of 16

Performance Analysis of Pulsed Loads

on Integrated Power Systems


S.D. Sudhoff Phd
Purdue University, USA
B. Kuhn
SmartSpark Energy Systems, Inc., USA
J. Amy Phd
Syntek Corporation, USA
D.E. Delisle
Naval Sea Systems Command, USA

ABSTRACT

This paper examines options for integrating Electro-Magnetic Rail Guns within the DD(X) Integrated Power System,
describes simulation models developed to assess the impact of pulse loads on the electric plant performance, and
describes a reduced power test bed that can be used to validate simulation model approaches. Two basic configurations
of pulse power converters are examined; a load commutated based converter that reflects various power factors on the
electric plant, and a combined uncontrolled rectifier-dc/dc converter based charger that a reflects near-unity power
factor on the electric plant. Each option is studied with and without the option of coordinating operation with the
propulsion system. Potential impacts to plant line-ups and ship-service power are discussed.

INTRODUCTION

The U.S. Navy’s DD(X) Program is in the process of designing a modern, land-attack destroyer. One of the salient
design features of DD(X) is an Integrated Power System (IPS) which will supply electric power both for propulsion and
for ship service loads. The U.S. Navy has recently conducted concept studies to assess the ship impact of integrating EM
rail guns within DD(X) IPS architecture1. The results of the concept studies indicate the baseline DD(X) IPS provides
sufficient installed generation capacity to meet baseline rail gun firing rates while maintaining reasonable propulsion
speeds and ship-service loads. However, detailed analysis is required to address the next level of system integration
issues and trade studies. In a related effort the U.S. Navy has started to develop analysis tools in the form of reduced
power test beds and simulation models to better understand pulse load integration issues within an IPS. Simulation
models of test bed components have been developed and used to begin addressing EM Rail gun integration
considerations and issues. This paper describes various options for integrating an EM rail gun within the DD(X) IPS,
describes the reduced power test bed, and provides the results of simulation models used to address the impact of EM rail
guns on electric plant components and system power quality.

Author’s Biographies

Scott D. Sudhoff received the B.S. (highest distinction), M.S., and Ph.D. degrees in electrical engineering from Purdue University in
1988, 1989, and 1991, respectively. From 1991-1993, he served as a consultant for P.C. Krause and Associates in aerospace power and
actuation systems. From 1993 to 1997, he served as a faculty member at the University of Missouri - Rolla, and in 1997 he joined the
faculty of Purdue University where he currently holds the rank of full professor. His interests include electric machines, power
electronics, and marine and aerospace power systems. He has published over fifty journal papers in these areas, including six prize
papers. He is particularly well known for his work in machine-converter analysis, as well as his work in stability of power electronics
based distribution systems, most of which concerns Naval applications. Much of his current research focuses on genetic algorithms
and their application to power electronic converter and electric machine design.

Brian T. Kuhn received the B.S. and M.S. degrees in electrical engineering from the University of Missouri-Rolla in 1996 and 1997
respectively. From 1998-2003 he served as a research engineer at Purdue University. Presently he is employed as a Senior Engineer
at SmartSpark Energy Systems, Inc., Champaign, IL. His research interests include power electronics and electrical machinery.

Dr. John V. Amy Jr. graduated from the U.S. Naval Academy with a B.S. in E.E. in 1983, and served on USS Boone (FFG 28). He
then reported to M.I.T. where he earned a S.M. in E.E. & C.S. in 1990, a Naval Engineer degree in 1990, and a Ph.D. in naval electric
power systems in 1992. An Engineering Duty Officer, he subsequently was Assistant Project Officer for Aircraft Carrier Overhaul at

Copyright © IMarEST 2004 – all rights reserved


Supervisor of Shipbuilding, Conversion and Repair, USN Newport News, VA. His next assignment was to the Engineering
Directorate at the Naval Sea Systems Command, initially as Assistant Program Manager for Systems Engineering/Ship Integration
within the Advanced Surface Machinery Programs, then subsequently Deputy Program Manager for the Integrated Power Systems
Program (PMS 510). Dr. Amy worked next in the program office of the CVNX Program, the Navy’s Future Aircraft Carrier Program,
as Deputy Assistant Program Manager for Requirements and Test & Evaluation. Following this, Dr. Amy was a member of the M.I.T.
faculty, Associate Professor of Naval Construction and Engineering, where, among other duties, he taught courses covering naval
architecture, marine engineering and naval ship design. Retiring from the U.S. Navy in 2003, Dr. Amy is now Director, Power
Systems at Syntek Technologies, Inc. in Arlington, Virginia

Mr. Dana Delisle is Assistant Program Manager for Future Concepts in the IPS Program (PMS-510) and a Senior Electrical Engineer
for the Total Ships Power Division (SEA05Z3) at Naval Sea Systems Command. For the past 18 years he has investigated power
electronic based electrical power systems for Navy Ship applications. He received his BSEE degree from the University of
Massachusetts in 1984.

DD(X) SYSTEM DESCRIPTION

The baseline DD(X) IPS architecture includes three basic sub-systems; a medium voltage, 13.8 kV, 60 Hz distribution
system for prime power generation and distribution; dc power for propulsion drives, and a lower voltage ship-service
distribution system.

Medium Voltage Distribution

Prime power generation for the DD(X) will be dependent in-part on the ability to achieve sustained speed while
maintaining a level of ship-service power. The requirement for sustained speed is 30 knots. The twenty four hour
average ship service electric load is around 5 MW. These two significant loads, supporting a 14,000+ tonne ship, are to
be supplied from a remarkably few, modest sized gas turbine generators. This efficient use of installed horsepower is
possible only with an integrated power system.

Two large, Main Turbine Generators (MTG) and two smaller Auxiliary Turbine Generators (ATG) shall provide
approximately 80MW of medium voltage, 13.8kV, three-phase, 60 Hz electric power. Fig 1 provides a simplified
diagram of the DD(X) MV distribution system. This medium voltage power would be distributed between the main
machinery rooms and to zonal load centers via six or so switchboards and a reconfigurable, flexible set of
interconnections. The interconnections between the generators and the inputs to the propulsion motors are rated for the
propulsion load. The interconnections between the switchboards and the zonal load centers are rated on the order of
2MW. Hence, ac power, on the order of the capacity of an MTG about 36MW, would be available at the MV
switchboards within the machinery box.

Propulsion Power

The propulsion motors are permanent magnet motors. The 13.8kV, three-phase, 60Hz power is to be transformed by a
phase-shifting step-down propulsion transformer to increase the pulse number seen by the generators and reduce the
voltage which must be rectified. The transformed power will be rectified to dc. Each of the two motors is to be supplied
by two rectifiers for redundancy. Hence, dc power, on the order of one quarter of the generating capacity, would be
available at the output of each of the propulsion rectifiers. The rectified power is subsequently inverted and provided to
the motor windings.

Copyright © IMarEST 2004 – all rights reserved


ATG
Ship Ship
MTG Svc. Svc.
Propulsion XFMR ~ 4 MW,
~ 36 MW, 13.8 kV, 3ph Primary 13.8kV, 3ph
13.8kV, 3ph

Motor Converters

Filters
(if req’d)

Filters
(if req’d)
Converters Motor

Propulsion XFMR
13.8 kV, 3ph Primary

Ship Ship
Svc. MTG
Svc.
ATG
~ 36 MW,
~ 4 MW, 13.8kV, 3ph
13.8kV, 3ph

Fig 1 DD(X) Medium Voltage

STBD MV Bus
UPS 13.8 KV Bus
Step-down Port/ STBD
Uninterrupted
Power for all vital 375 VDC
XFMR
450 VAC
sensitive loads
Load

GTG Elect.
Load
Start Drive

APU
~ 375 VDC
Load
Pwr Pnl

120 VAC
Elex Pwr 48 VDC
Pnl Vital
Motor

Load

375 VDC 450 VAC


Step-down

XFMR
UPS
Port MV Bus

Fig 2 DD(X) Ship-Service System

Copyright © IMarEST 2004 – all rights reserved


Ship-Service Power

The ship service loads on DD(X) are going to be modern, diverse types of equipment. The ship has been divided into
four electric zones. Each zone receives power from two load centers, which are separated to the maximum degree
possible within the zone.

Fig 2 shows the baseline design for ship-service power for a single zone. Available power types within an electric zone
are to be 375Vdc, 450Vac, 115/220Vac and perhaps some lower voltage dc power. No zone has an aggregate load in
excess of several MW. Power quality within the zones for ship service loads is an important consideration.

DESCRIPTION OF OPTIONS FOR INTEGRATING A PULSE POWER SYSTEM WITHIN THE DD(X)
POWER SYSTEM DESIGN.

A first option for supplying a pulse power system is to install an additional, dedicated generator on the ship for the pulse
power system. The advantages and disadvantages of this approach are obvious; hitherto, the disadvantages of this option
have precluded the development of high powered, pulse power systems.

A second option for supplying a pulse power system is the approach implemented by the Integrated Power System for
supplying both propulsion and ship service loads; namely, the power from the generators is shared. This option is
considered here. Implementation and the design of the interface between the pulse power system and the ship’s power
system hinges on two related issues. First, what is the magnitude of the maximum power to be drawn from the ship’s
power system? Second, what is the pulse repetition rate of the pulse power system?

The first issue, the magnitude of the maximum power to be drawn from the ship’s power system, is influenced by several
considerations. What is the capacity of the ship’s power system? How much power does the pulse load require? What
type of power is to be drawn from the ship’s power system? Are there constraints on the dynamical application of the
pulse load owing to the necessity of avoiding instability or other degradation of power quality with respect to the balance
of the ship’s loads? The first two considerations are important relative to each other. A pulse load that draws more
power than the ship’s capacity is infeasible. Further, a pulse load that draws so much power that it requires all generators
to be online is probably also infeasible. Hence, a reasonable limit on the maximum pulse load power level would be that
which is available with one generator unavailable, for whatever reason, less an engineered and managed minimum
combined ship’s service and propulsion load. For a system such as that in Fig 1, this limit would be on the order of the
rating of one of the larger generators; let us consider something on the order of 30MW.

The third consideration, the type of power to be drawn from the ship’s power system, addresses two things. First, the
amount of power dictates, to a degree, the type of power. 30MW is more easily distributed at medium voltages than at
low voltages. Second, the inherent physics of the pulse load dictate the type of power to achieve its effect. Herein lies
the function of, what is called by some, a pulse forming network; it changes the type of power to that which the physics
of the pulse load demands. Where the ship’s generators supply power in a different form than that required by the
physics of the pulse load, a means to convert that power, a pulse forming network, is necessary. Here, it is assumed that
such a pulse forming network is necessary. Only the input to that pulse forming network is considered. In this instance,
30MW of power would be provided at a medium voltage, between 1000V and 13.8kV, to a pulse forming network.

The fourth consideration, constraints on the dynamical application of the pulse load owing to the necessity of avoiding
instability or other degradation of power quality with respect to the balance of the ship’s loads, is concerned with the
implementation of the pulse forming network. Here, for example, the rate at which the pulse load is increased to its
maximum level must be such that voltage or frequency responses within the balance of the ship’s service distribution
system remain within an acceptable range of values. Detailed simulations of specific configurations would provide
assessments of the maximum ‘ramp rates’ for the pulse load for ensuring acceptable frequency and voltage behaviour
within the balance of the ship’s service system. A further limit on the behaviour of the pulse forming network is
concerned with the injection of load current harmonics into the ship’s service distribution system. A straightforward,
though ‘expensive’, way to segregate the pulse load from the ship’s service and propulsion loads would be to provide the
pulse forming network from a split, dedicated generator during operation of the pulse load.

The second issue, the pulse repetition rate of the pulse power system, is related to the first and is a first order driver of the
pulse forming network design. This issue is based upon several considerations. How much energy does the pulse load
require for each pulse? How much energy can be stored in the pulse forming network? How frequently will the load
pulse? The most appealing aspect of the engine-as-a-weapon is the possibility of providing a virtually unlimited number
of these pulses.

Copyright © IMarEST 2004 – all rights reserved


The first consideration, how much energy each pulse requires, is related to the maximum pulse power level seen by the
ship’s distribution system and the pulse repetition rate, the third consideration. Two of the three characteristics can be
chosen independently; the third follows the choice of the two preceding. Here, the maximum pulse power level has been
fixed, the 30MW example. Hence, the energy per pulse, once known, gives rise to the maximum pulse repetition rate.
Complementary approaches would lead to somewhat different conclusions.

The second consideration, how much energy can be stored in the pulse forming network, is determined by energy storage
technologies and by how it is desired, temporally, to operate the pulse load and the pulse forming network, taken
together.

The two issues described, with their supporting considerations, provides a reasoned approach to assessing an appropriate
point, within such a power system as Fig 1, from which to draw power for a pulse load. Lacking all other information
about the pulse load, the direction of design ‘goodness’ is assumed to be in the direction of greater pulse load power
levels which leads to the greater pulse repetition rates. Therefore, a pulse load could be provided around 30MW, from
the medium voltage distribution system, or possibly the output terminals of the propulsion rectifiers, if dc is desired.
Provision of this much power to the pulse load requires active power management to reduce to a minimum the combined
ship’s service and propulsion load. What this portends for the pulse forming network follows. 1) It must convert either
13.8kV AC power, or propulsion rectifier output DC power, to the type of power inherently required for the pulse load
effect. 2) It must limit its ‘ramp rates’ and load current waveform to be compatible with the balance of the ship’s power
system. 3) It must store sufficient energy to provide pulses as the pulse load is planned to operate.

ANALYSIS TOOL DESCRIPTION

The U.S. Navy has started to develop analysis tools to better understand pulse load integration issues and address the
aforementioned issues and considerations. A reduced power generation, propulsion, and pulse power test bed has been
developed at Purdue University. Simulation models of test bed components have been developed and used to begin
addressing EM Rail gun integration considerations and issues.

Fig 3 is a one-line diagram of the reduced power generation, propulsion, and pulse power test bed. This test bed is
designed to emulate the principal features of the medium voltage portion of an IPS. In this system, a turbine emulator /
governor acts as the prime mover for the system. Physically, the turbine emulator is implemented using a 120 kW 4-
quadrant speed-controlled dynamometer. The prime mover drives a 59 kW 3-phase wound rotor synchronous machine
with a brushless excitation system2.

The most significant load on the system is the propulsion load. In this system, the propulsion load consists of a rectifier,
dc link, inverter based induction motor drive, with a load emulator. The power level of the propulsion load is 37 kW, or
about 63 % of the system generation3.

Other loads on the system include a harmonic filter and a power supply. The harmonic filter is designed to reduce the
bus distortion due to the propulsion rectifier; it consists of a wye-connected LC circuit. Parameters are listed in Table I.
The 15 kW power supply feeds the dc portion of the system and emulates a feed to a dc based ship-service distribution
system4. For the studies herein it is operated as an uncontrolled rectifier. At peak load, it utilizes roughly 22 % of total
system generation.

Copyright © IMarEST 2004 – all rights reserved


Fig 3 One-line Diagram of the Reduced Power Generation, Propulsion, and Pulse Power Test Bed

Table I. Harmonic filter parameters.


Symbol Description Value
L Per-phase inductance 5.6 mH
C Per-phase capacitance 50 µ F
r Per-phase effective series 39 m Ω
resistance

Table II. Configuration 1 parameters.


Symbol Description Value
n sp Transformer secondary to primary 0.79
turns ratio (wye-equivalent)
Llp Transformer primary leakage 1.6 mH
rp Transformer primary resistance 61 m Ω
Lls ' Referred transformer secondary 1.6 mH
leakage
rs ' Referred transformer secondary 61 m Ω
resistance
LM Transformer magnetizing 0.43 H
inductance
C es Energy storage capacitance. 2.02 F
rin Input inductor series resistance 0.1 Ω
Lin Input inductor series inductance 11 mH

Table III. Configuration 2 parameters.


Symbol Description Value
rin Input inductor series resistance 0.1 Ω
Lin Input inductor series inductance 4 mH
Cin Input capacitance 500 µ F
rout Output inductor series resistance 0.1 Ω
Lout Output inductor inductance 3 mH

Copyright © IMarEST 2004 – all rights reserved


Pulsed Load

Two pulse-forming network charging configurations are considered herein as depicted in Figures 4 and
5. In configuration 1, a transformer line-commutated converter is used to charge the energy storage
capacitor C stg . In configuration 2, a transformer-uncontrolled rectifier-buck converter combination is
used. Note in Configuration 2, the energy storage portion of the pulse forming network is represented
by a capacitor however, this could also represent a rotating machine based pulse forming network.
Configuration 2 represents a 2-stage power converter where the first stage (transformer-rectifier) is not
unlike that required for the propulsion system drives. The parameters of the passive components of
each of the configuration 1 and configuration 2 systems are set forth in Table II and Table III,
respectively. Note that the transformer parameters are wye-equivalent. Also, the transformer and
energy storage capacitors are identical for both configurations and so are listed only once in Table II.

Fig 4 Configuration 1 Power Converter

run
rin Lin rout Lout

+
rstart
AC Pulsed
Ci Cstg vc Load
BUS

1: nsp _

Fig 5 Configuration 2 Power Converter

Pulse-Load Characteristics

The specific energy storage requirements for an EM rail gun pulse forming network will be dependent
on a number of factors including projectile weight, launch velocities, firing rates all which equate to
energy demands from the distribution system. For the purposes of this study, the pulsed power
charging network will store a peak energy of 200 kJ. Each pulsed load will remove 128 kJ from the
capacitor bank and the peak pulse rate for this load will be 0.1 Hz (6 cycles per minute).

Pulse Forming Network Control Algorithms

The controls of the two configurations have much in common. At the highest level is the charge /
discharge control shown in Fig 6. The inputs to this control are a command to charge the capacitor, ec*
(high to charge), a command to discharge, ed* (high to discharge), the filtered voltage across the energy
storage capacitor, v cf , and the minimum capacitor voltage at which firing is allowed, v c*2 . The
outputs of this control are the actual charging status, ec (high to charge), and actual discharge status,
ed (high to discharge).

Copyright © IMarEST 2004 – all rights reserved


Provided that a discharge sequence is not underway, setting ec* high will cause the charge status, ec , to
go high, whereupon the capacitor will be charged. If the energy storage capacitor voltage is above the
threshold and ed* goes high (indicating a request for discharge), or ed* is already high and the filtered
capacitor voltage becomes greater than the minimum firing voltage, v c*2 , then eds will immediately go
high to indicate that a discharge sequence has been initiated. On the rising edge of this signal two
pulses are generated. The first of these has a duration of t dc , the time interval over which charging is
disabled. The second of these has a duration of the total discharge cycle t dc less the time required to
physically discharge the capacitor t dis . The result is that the charge status ec goes low for the entire
discharge cycle of length t dc and ed goes high (and the actual discharge occurs) during the last t dis of
the discharge interval of length t dc .

The next level of the control is the synthesis of a capacitor current command ic* . This control is
diagrammed in Fig 7. The basic philosophy of the control is to charge the capacitor as rapidly as
possible subject to a peak capacitor current limit, icmax , and a peak power limit, Pcmax . Inputs to the
control are the target final capacitor voltage, v c*1 , the measured capacitor voltage, v̂ c , and the charge
status, ec . As can be seen, the measured capacitor voltage is first filtered by a simple low pass filter
with time constant τ inf and then subtracted from the command. The voltage error is then multiplied
by a proportional gain K sf and limited to a dynamic limit iclimit , which is calculated so as to insure
both the power and peak current limit will be observed.

The commanded capacitor current is then the filtered output (time constant of τ of ) of either this
gain/limiter (if ec is high) or zero (if ec is low).

Fig 6 Charge/Discharge Control

Fig 7 Current Command Synthesizer

Copyright © IMarEST 2004 – all rights reserved


The gain K sf is selected to be large enough that the limit is almost always in effect until the point
where v cf becomes very close to v c*1 ; after this point the capacitor voltage approaches the target
voltage asymptotically. For this reason, the target voltage, v c*1 , is slightly higher than the minimum
voltage to fire, v c*2 . Parameters for the charge/discharge and capacitor current controls are common
for the two systems and are listed in Table IV.

Table IV. Charge/discharge and capacitor current control parameters.


Symbol Description Value
vc*1 Target capacitor charging voltage 450 V

vc*2 Minimum capacitor firing voltage 445 V


Pcmax Maximum allowed power into 15.5
storage capacitor kW
icmax Maximum allowed current into 40.5 A
storage capacitor
tdis Time over which storage capacitor 0.1 s
is discharged
tdc Time of discharge cycle 0.15 s
τ inf Time constant of input filter 5 ms
τ of Time constant of output filter 5 ms

The objective of the next level of the control is to insure that the capacitor current accurately tracks the
desired current. The strategy for doing this is a function of the power converter topology.

Fig 8 shows the current control for configuration 1. In this case, the inputs to the control are the
capacitor current command, i* , the measured rectifier current, iˆ , and the filtered capacitor voltage,
c r
vcf . The output of the control is the cosine of the phase delay of the load commutated converter,

cosα . The basic strategy of this control is the formulation of a rectifier voltage command, v*r , which
is essentially equal to the output of a PI control operating on the current error (the difference between
the capacitor current command ic* and the filtered rectifier current ir , which is in turn equal to the
capacitor current during the charge cycle) plus a feed-forward term equal to the filtered capacitor
voltage. This command is then translated to a phase delay by dividing by the nominal rectifier voltage
and limiting. Note that when the limits are in effect the integrator of the PI control is prevented from
integrating in order to prevent wind up. Parameters of this control are listed in Table V.

The current control for configuration 2 is depicted in Fig 9. In this control, an inductor current
command il* is formulated which is equal to the capacitor current command ic* plus the limited (to
avoid wind-up) integral of the error between the capacitor current command and inductor current (the
capacitor current and inductor current are synonymous during the charge cycle). The switching state of
the active switch, sas (high is on, low is off), is then determined from this command and the measured
inductor current using a hysteresis modulator. The purpose of the integral term is to remove any
tracking error in the hysteresis modulator. Parameters of the configuration 2 current control parameters
are listed in Table VI.
Table V. Configuration 1 current control parameters
Description Value
PI control gain 4 V/A
PI control time constant 16.8 ms
Nominal rectifier voltage 599 V
Input filter time constant Table III

Copyright © IMarEST 2004 – all rights reserved


Table VI. Configuration 2 current control parameters.
Description Value
Time constant of integral feedback 1 ms
Integral feedback limit 2.62 A
Hysteresis level of hysteresis modulator 2.62 A

Fig 8 Current Control for Configuration 1

Fig 9 Current Control for Configuration 2

This completes the description of the controls directly associated with the pulsed power load.
However, there is one additional system control that will be considered, which is associated with the
coordination of the pulsed power load with the propulsion load. This control algorithm is depicted in
Fig 10. Therein, inputs are the mechanical rotor speed of the propulsion drive, ω rm , the torque
command to the propulsion drive, Te* , and the effective instantaneous power command into the pulsed
power load, P* , which is equal to the commanded capacitor current, ic* , times the filtered measured
capacitor voltage, vcf . Based on these quantities a modified torque command, Te** , for the propulsion
drive is produced such as to keep the overall system power requirements constant, subject to a
maximum torque perturbation ∆Te, mx and that the final torque command has to be positive. For the
studies to be considered, ∆Te, mx was set equal to the maximum allowed torque command of 200 Nm.

Fig 10 Propulsion Coordination Control

Copyright © IMarEST 2004 – all rights reserved


RELATIVE PERFORMANCE

Four configurations of pulsed power loads have been set forth (two basic configurations, each with the
variation of whether or not propulsion coordination is utilized). In this section, the system impact of
each of these configurations is compared. Two scenarios will be considered – one with the system
initially in a lightly loaded condition and the other with the system in a heavily loaded condition.

In the first scenario, the system is at very light load. The propulsion drive is at 0.2 p.u. speed and 0.05
p.u. torque for 0.01 p.u. load on the drive base (per unit definitions appear in Table VII). The power
supply is operating with a 0.1 p.u constant power load (power supply base). The system is initially
operating in the steady state with zero energy storage capacitor voltage. Then, both the charge
command, ec* , and discharge command, ed* , are set high. The result is the pulsed power load charges
and discharges as fast as possible.

In the second scenario to be studied, the situation is identical except that the system is heavily loaded.
In particular, the propulsion drive is at 1 p.u. speed and 1 p.u. torque. The power supply is operating
with a 0.1 p.u. constant power load (power supply base).

The simulation studies are based upon a detailed simulation model in which the switching of every
power semiconductor in the system is represented on an individual basis. Because of the prolonged
duration of the study, the system variables will be processed to a form that lends itself to a more
insightful representation on the time scale considered. Variables will also be shown in per unit (see
Table VII for base values).

In regard to processing, variables were filtered prior to plotting in order to facilitate plotting on a
compressed time-scale. In all cases the filter used consisted of a cascade of two first-order low-pass
filters with a cut-off frequency of 36 Hz. Since this can be crudely viewed as an averaging operation, a
bar over a variable or quantity is used to indicate that it has been filtered with this filter. This bar does
not appear in the figures.

Instantaneous magnitudes of ac quantities are defined herein by the following algorithm. First abc
phase variables are transformed to qd0 variables in a synchronous reference frame5. Next, the
magnitude is computed as

f mg
= (f ) + (f )
e 2
q d
e 2
(1)

where f qe and f de are the filtered q- and d- axis quantities of interest in a synchronous (in this case
rotor) reference frame. Throughout this work mg
will designate the magnitude of an ac quantity as
calculated using (1).

Another quantity of interest will be the instantaneous power factor. This is determined by first
transforming the voltage and current to a synchronous reference frame and then computing the
instantaneous real (2) and reactive (3) power as

P=
3 ee
2
(
vqiq + vde ide ) (2)

Q=
3 ee
2
(
vqid − vde iqe ) (3)

These quantities are then filtered to yield P and Q . The power factor is then taken as (4)

P
v, i pf = (4)
P2 +Q 2

Copyright © IMarEST 2004 – all rights reserved


where the indicates absolute value.

A final conditioned property of interest will be the instantaneous total harmonic distortion of the bus
voltage. For this three-phase system the dynamic total harmonic distortion is defined as (5).

v thd =
(v qe − vqe )2 + (vde − vde )2 (5)
(v qe )2 + (vde )2
In the steady state, this definition of 3-phase dynamic total harmonic distortion coincides with the
classical steady-state definition if the harmonics in the three-phase system occur in balanced sets.

Now that the basic processing definitions have been set forth, the reader is referred to Table VII. for a
complete listing of variables examined and their base values.
Table VII. System Variables.
Symbol Description Base
vb mg Magnitude of bus voltage 457 V

vb thd Total harmonic distortion of the -


bus voltage
i sm mg Synchronous machine current 86 A

vb , ism Synchronous machine power -


pf
factor
Tesm Synchronous machine torque; 313 Nm
negative for generator operation.
ibe Brushless exciter field current 0.8 A
ω rsm Synchronous machine 188.5
(generator) speed rad/s
i pp Current into pulsed power 18.7 A
mg converter
vb , i pp Power factor of pulsed power -
pf converter
ir , pp Pulsed power converter rectifier 28.4
current
vc, pp Pulsed power energy storage 450 V
capacitor voltage
i pd Current into propulsion drive 54.4 A
mg power converter
vb , i pd Propulsion drive power factor -
pf

vdc, pd Propulsion drive DC link voltage 756 V

Te, pd Propulsion drive torque; positive 198 Nm


for motor operation
i ps Current into power supply 21.9 A
mg

vb , i ps Power supply power factor -


pf
v dc, ps Power supply dc output voltage 500 V

Copyright © IMarEST 2004 – all rights reserved


Computer Simulation Results

The results of the scenario 1 study (light system load), are depicted in Fig 11, and the results of the
scenario 2 study (heavy system load) are illustrated in Fig 12. Configuration 1B and 2B include
propulsion coordination control. Between 0 and 5 seconds, the system is in a steady-state condition at
base load. The first pulse begins at T = 5 seconds at which point the charging network emulates
charging the capacitor to store 200kJ as fast as possible but within the established power and current
limits. By T = 28 seconds, the charging network has stored approximately 200kJ of energy. The
capacitor is then isolated and approximately 128kJ is immediately removed from the energy storage
network. Once the energy is removed, the charging network begins recharging the energy storage as
fast as possible but within the power and current limits. Within 10 seconds, the 128kJ has been
replenished. At T = 38 seconds, 128 kJ is again immediately removed from the energy storage network
and the cycle repeats. The first observation that can be made is that the first charging cycle requires
much more time that subsequent cycles. This is because on the first cycle there is initially no capacitor
voltage on the energy storage capacitor; since the discharge is not complete subsequent charges require
much less time.

The ac bus voltage during these studies is illustrated in the first trace of Figures 11 and 12. As can be
seen, the perturbation of bus voltage is similar for the two studies. It can also be seen that
configurations 1A and 1B undergo a much more significant initial voltage dip than configurations 2A
and 2B. On subsequent charge cycles, it can be see the configuration 2B produces the smallest voltage
excursion. The total harmonic distortion of the ac bus is depicted in the second trace of Figures 11 and
12. In all cases, it can be seen that from this perspective configurations 2A and 2B are superior to
configurations 1A and 1B. In scenario 2, in which the system is heavily loaded, it can be seen that
configuration 2B is superior to 2A, and configuration 1B is superior to configuration 1A.

The synchronous machine current and synchronous machine power factor are illustrated as the third
and fourth traces of Figures 11 and 12 respectively. In scenario 1, it can be seen that configurations 2A
and 2B draw considerably less current at a much higher factor than configurations 1A and 1B. In this
case, since there is little propulsion power, the difference between the A and B configurations is
insignificant. For scenario 2, configuration 1A requires the most current from the generator,
configuration 2A the next largest, then configuration 1B (except for the initial part of the initial charge
cycle), and finally configuration 2B results in the least generator current. Configuration 2A results in a
higher generator current than configuration 1B because it results in a higher system load then
configuration 1B, even though the synchronous machine power factor is much better.

The synchronous machine electromagnetic torque and brushless exciter field current are depicted the
fifth and sixth traces, respectively, of Figures 11 and 12. Note that the torque is negative for generation
operation. During scenario 1, in which the system is under light load causing the propulsion
coordination control to be ineffective, it can be seen that the synchronous machine torque is similar for
all configurations (which follows from the fact that the real power demand is similar). During scenario
2, the synchronous machine torque for the B configurations has a smaller magnitude than for the A
configurations since the total system load is smaller in this case. In scenario 1, it can also be seen that
configuration 2A and 2B result in much less excitation current than configurations 1A and 1B. During
scenario 2, the situation is more complicated. Configuration 1A always requires the greatest field
current and configuration 2B always requires the least field current. However, the ranking between
configuration 1B and configuration 2A varies depending upon the portion of the charge cycle as the
field current varies both with real and reactive power requirements.

Copyright © IMarEST 2004 – all rights reserved


Fig 11 Scenario 1 Simulation Results

Copyright © IMarEST 2004 – all rights reserved


Fig 12 Scenario 2 Simulation Results

Copyright © IMarEST 2004 – all rights reserved


CONCLUSIONS

A number of options exist to integrate an EM gun within the DD(X) IPS architecture. A systems
engineering solution that includes detailed simulations of the electric system and components can be
used to optimize the integration and ensure performance requirements are achieved.

Looking only at the pulse forming network, the line-commutated converter (configuration 1) appears to
be the simplest to implement. For the DD(X) configuration, the converter could be connected directly
across the MV bus and SRCs can be phased-back as required to charge the energy storage element.
The results of this simulation however show such a converter can introduce higher levels of harmonic
distortion and voltage fluctuations on the MV bus, and will require the generation system to produce
higher currents then configuration 2.

A transformer, uncontrolled rectifier buck converter combination (configuration 2) appears to be a


more expensive power conversion option in comparison to configuration 1. However, the additional
costs might be off-set by the cost of approaches to provide sufficient quality power to other connected
loads and and/or sizing the generators to accommodate the additional harmonic currents and low power
higher currents. One option to offset the additional size/weight of configuration 2 might be to locate
the pulse forming network charging circuit on the dc side of the propulsion transformer(s). The dc
voltage level and potential impacts to dc bus power quality would have to be addressed to assess this
option.

Generator torque plots show that the real power delivered by the generator is approximately the same
regardless of which configuration is used to charge the pulse forming network. Shaft torque transients
can result in frequency transients on the MV bus. The magnitude of frequency transients and the
impact of those transients on other connected loads would need to be assessed. With respect to peak
shaft torque, power management can be used to ensure peak torque levels do not exceed machine
ratings. Automated power management approaches could allow EM gun firing rates to be adjusted and
maximized based on mission requirements and load priorities.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge financial and technical support of this project from the U.S. Navy
under the Naval Combat Survivability effort; contract N00024-02-NR-60427.

REFERENCES

1
Chaboki et al., “Integration of Electromagnetic Rail Gun into Future Electric Warships”, Engine as a
Weapon Conference, 10 June 2004.
2
D.C. Aliprantis, S.D. Sudhoff, B.T. Kuhn, J.D. Sauer, D.E. Delisle, T.J. McCoy, “A Detailed
Synchronous Machine Model,” SAE2002 Power Systems Conference, October 29-31, 2002
3
S.D. Sudhoff, B.T. Kuhn, “Advanced Propulsion System Report,” report to Naval Sea Systems
Command, 24 January 2002.
4
S. Pekarek et. al., “A Hardware Power Electronic-Based Distribution and Propulsion Testbed,” Sixth
IASTED International Multi-Conference On Power and Energy System, May 12-15, 2002, Marina del
Rey, California, USA
5
P C. Krause et al, “Analysis of Electric Machinery”, IEEE Press, 1995

Copyright © IMarEST 2004 – all rights reserved

You might also like