THE ELECTRIC WARSHIP VI
by
C G Hodge
and
D J Mattick OBE
A paper for the Institute of Marine Engineers
To be read at 1730 on Tuesday 12 December 2000
The views expressed are those of the authors alone.
BIOGRAPHIES
DAVID MATTICK
David Mattick’s early career resulted from specialisation as a nuclear submarine weapon
engineer. After service as the Assistant Weapon Electrical Officer of HMS WARSPITE
and in the MoD, he rejoined WARSPITE as the Weapon Electrical Officer of HMS
WARSPITE. In 1982 he was appointed as the Marine Engineer Officer of HMS
SWIFTSURE. After promotion to Commander in 1984 he headed the electrical power
systems specialist group within the MoD, subsequently serving with the VANGUARD
Class submarine project and then as a Project Manager with Director General Ship
Refitting. In 1994 he served as the Surface Ship Marine Engineering Desk Officer in
Director Future Projects (Naval), tasked with concept design of future naval vessels and
was appointed as the MoD Electric Ship Programme Manager in February 1996. Since
retiring from the Royal Navy in 1999 he has been employed in the Integrated Propulsion
Systems Division of Rolls-Royce Marine Power as the Electric Ship Manager. He was
awarded the OBE in the 1999 New Year’s Honours List.
CHRISTOPHER HODGE
After initial training as a mechanical engineer, and service as an Assistant Marine
Engineer in HMS WARSPITE, Christopher Hodge joined HMS ORPHEUS as the
Marine Engineer Officer in 1982. He subsequently took an MSc in Electrical Marine
Engineering and served in the MoD as the project officer for electrical ship propulsion.
After promotion to Commander in 1989 he served as the Marine Engineer Officer of
HMS CONQUEROR before returning to MoD as the head of the Nuclear Steam Raising
Plant electrical design authority. He was the head of the electrical power system
specialist group within the MoD until April 1998 after which time he joined Rolls Royce
Marine Power initially in their Integrated Propulsion Systems Division and now in the
Engineering and Technology Division .
THE ELECTRIC WARSHIP VI
C G Hodge BSc MSc CEng FIMarE
D J Mattick OBE BSc CEng FIMarE MIEE MINucE
ABSTRACT
This paper which follows on from five earlier IMarE papers by the same authors, reviews
the progress of the Electric Ship Programme over the last year and focuses on some of the
technological developments in equipment and architectures associated with the Electric
Warship
INTRODUCTION
This is the sixth paper in the Electric Warship series presented at the IMarE and as before
it aims to re-state and update the Electric Warship vision and inform readers of progress
towards realising the concept in a warship and to record significant developments in its
associated technology.
BACKGROUND
The following is a very brief summary of the first five Electric Warship papers,
References 1, 2, 3, 4 and 5
The benefits of employing a common power system for both propulsion and ship’s
services can, in an optimised merchant ship installation, allow Running Cost savings of
up to 25%. In a warship, where the constraint of reducing Unit Purchase Cost as well as
Running Costs does not allow a system to be fully optimised for fuel economy, the
savings may nevertheless be sufficient to allow purchase of one extra ship in a class of
around thirty vessels. However as was explained in the previous papers, the size of
motors and converters needs to be reduced if the equipment is to be installed in current
frigate sized warships.
The Electric Ship (ES) concept was developed from Integrated Full Electric Propulsion
(IFEP) and it was aimed to reduce Unit Purchase Cost (UPC) and yet retain as much as is
practical of the IFEP reduced Running Cost (RC). In the Electric Ship this is achieved
with two main features above those to be found in a traditional IFEP vessel.
Minimum Generator Operation: The UPC and space constraints require fewer but more
highly rated prime movers than would be found in a merchant IFEP vessel. In order to
restore the fuel savings conceded by fitting fewer prime movers the ES runs under a
regime of minimum generator operation, often with only one prime mover operational.
This brings significant gains, partly in fuel consumption but mainly from maintenance
costs due to the minimised engine running hours.
Electrification of Auxiliaries: Additional maintenance and manpower reductions can be
achieved by using electric auxiliaries wherever possible. In addition there will be benefits
to be gained from this policy in terms of overall weight of equipment fitted (central
energy storage and high reliability ship wide electric power systems). The previous
papers described our early vision of an Electric Ship Power System, shown at Figure 1.
GEN WR21
GEN 1-2 MW
SPC
CONV PMPM
En Store
CONV PMPM
SPC
GEN 6-8 MW
GEN WR21
750 V DC
Figure 1: Electric Warship Power System
REVIEW OF CURRENT ISSUES
AC or DC
As has been discussed in may forums, not least the earlier versions of this paper, one of
the principal decisions to be made early in the concept work for an electric warship is
whether AC or DC should be used as the medium for power distribution. The authors
have consistently maintained that the advantages offered by DC in its cost of installation,
efficiency of distribution and control require it to be treated as a serious option for both
the propulsion and ships service duties. In Reference 1 the authors stated that DC should
be used for ships service distribution and AC for the propulsion bus bar. This decision
was based on the belief that technology would limit the power that could be transmitted
safely by a DC system. However industry is now becoming increasingly confident that
that the power for propulsion could be safely controlled using DC. Reference 6 a paper
by PMES and the UK MoD presented at AES 2000, makes a strong case for using DC for
the propulsion system of a frigate sized electric warship and many power electrical
companies are now starting to believe that DC could provide the supply to the Propulsion
Motor Converters. The main issue with DC as a traditional distribution medium is the
expense and mass of DC air circuit breakers and this has been the reason why most
assessments to date the advantages of DC have only been marginal.
The authors now consider that convergence between power electronic converters and
circuit breakers is feasible. Indeed, as is discussed later in this paper, the USA IPS
Programme changed the functional definition of one of their modules from solid-state
circuit breaker to DC-DC converter when the use of the equipment became
predominantly one of DC voltage conversion, nevertheless the circuit protection and fault
current interruption functionality of the module was retained. The authors believe that
DC is the distribution medium of the future and can today be used for ships service
distribution. Further, if the power electronic converters that will necessarily be required
also provide the fault interruption duty then such a system will have clear advantages
over its AC counterpart, including initial cost and mass.
It is worth mentioning that one often quoted advantage of DC systems, namely that of
simpler and more robust stability, needs some qualification. It is true that the same issues
of transient stability do not arise. However if controlled in constant power mode, as is the
case for motor drives, power electronic converters exhibit a negative impedance. That is
if the system voltage falls the current drawn by the converter rises. This effect can cause
voltage instability through interaction with the remaining voltage regulating control
circuits in the complete DC system. Purdue University in the USA has conducted
significant research (led by Dr Scott Sudhoff) into this phenomenon, References 7 and 8
report the latest results of their work, which offers design methods by which stability of
power electronic DC systems can be assured.
MOTORS AND CONVERTERS
The quest for power dense motors suitable for warships continues at an ever-quickening
pace with the Alstom Advanced Induction Motor at the forefront; a motor has been
supplied to the USN Integrated Power System programme and selected for the Royal
Navy Type 45 programme.
In the commercial arena Jeumont Indutrie are providing radial flux permanent magnet
machines to DCN for an export project and they also supply axial flux machines for wind
generation duties. With their experience, they can readily scale and militarise these for
warship propulsion. Siemens are offering radial flux permanent magnet machines in the
Siemens-Schottel propulsors, a commercially available podded drive suitable for
commercial ships. It is likely that development of other commercial permanent magnet
motors is underway.
Podded propulsors are yielding significant benefits to the commercial marine industry. In
particular through reduced fuel consumption – good hydro-dynamic designed pod and
ship back-end can realise propulsion efficiency gains in excess of 10% - and simplified
ship installation - the complete pod can be supplied and fitted in days thus avoiding the
lengthy installation of a traditional system.
In the purely military arena, the Rolls-Royce development of the Transverse Flux
Permanent Magnet Motor, discussed in the last paper of this series, continues with
encouraging results emerging; ABB are offering a radial flux permanent magnet motor
based on their development for the Italian Navy submarine programme; various
companies and groupings in the US are developing permanent magnet motors targeted at
the DD21 programme.
As has been highlighted in many previous papers, the power electronics revolution
continues and various new devices are being developed or are available, which enable
different Converter topologies that can reduce volume, weight, cooling requirements and
cost of the equipment; this thread weaves throughout this paper.
ENERGY STORAGE
There is a continuing debate about the need for and capacity of Energy Storage required
to support Electric Ship propulsion and distribution. In practice, the need must be judged
against the application and be tailored to the specific system requirements. There are a
number of suitable technologies for this Energy Storage including traditional lead acid
batteries widely used in submarines, advanced batteries such as the Zebra system which
was reported at Reference 9, flywheels and, perhaps most recently, a Regenysis system
based on that offered by Innogy Technology Ventures Ltd - previously a part of National
Power. There is a web site dedicated to the Regenysis system and this may be found at
Reference 10.
The UK MoD has work underway to assess the Regenysis potential for a warship
application; indeed it is one of the systems selected for the UK Electric Ship Technology
Demonstrator.
MOTORS
TRANSVERSE FLUX
The development of the TFM continues though not without its difficulties. The 2.5 MW
Technology Demonstrator has been under test at DERA Pyestock with an Alstom Series
IGBT PWM converter providing the electrical supply. Vibration at 33 Hz caused much
early difficulty until a current instability caused by an interaction between the motor and
converter was identified and rectified. The 2.5 MW machine has been built with external
bearings that can be adjusted to investigate the effect of axial misalignment. It appears
that the 33 Hz vibration was sufficient to slacken the bearings and allow the rotor to
move and touch the stator. The damage to the stator C Cores was sufficient to necessitate
an extensive repair. On completion of the repair full power was achieved and accepted by
the UK MoD. After achieving full power the insulation on one of the eight coils failed
and has again necessitated an extensive repair that is still in progress. It is likely - though
not certain - that the insulation failure is associated with the design of the coil lead-outs
and the difficulty of applying the same level of insulation during build as the rest of the
coil. It is even for consideration that the lead-outs were insufficiently supported to
withstand the 33 Hz vibration that occurred early in the testing programme. The design of
the coil lead-outs is being re-assessed as part of the current repair programme.
Because the 2.5 MW TFM, once built, exhibited a lower reactance than expected it
proved impossible to achieve the design rating of the machine and although full power
has been achieved and agreed at 2 MW this is less than originally anticipated. At that
power the machine’s air gap shear stress has only reached 80 kN/m2 rather than the 100
kN/m2 necessary to design a 65 tonne 20 MW 180 RPM machine. Nevertheless the
machine has demonstrated that the topology is not only viable but offers significant
advantages in terms of torque density.
As the specific torque of the TFM is increased the power factor of the machine reduces.
For the 20 MW 180 RPM application being developed for the UK MoD the power factor
is currently anticipated as being 0.6. This relates to an air gap shear stress of 100 kN/m2
however the machine could be designed for an air gap shear stress of 120 kN/m2 but with
a consequent reduction in power factor to 0.4. A power factor of 0.4 implies a 50 MVA
converter for a 20 MW TFM. With the converter technology currently available this
requirement to provide 50 MVA reduces system efficiency unacceptably because of the
increased switching and conduction losses. Nevertheless the TFM topology has a greater
potential for torque density than is currently being exploited in the MoD development, all
that is necessary to gain further size reductions – perhaps to as low as 40 tonnes for a 20
MW 180 RPM machine - is a more efficient form of power electronic conversion. The
authors believe that this will arise in time as techniques such as resonant conversion and
multi-level conversion are developed. It is of interest to note that the TFM topology is
being actively considered for use in wave energy generation schemes where researchers
anticipate utilising a linear TFM with an air gap shear stress of 120 kN/m2.
AXIAL FLUX
Jeumont Industrie is developing an Axial Flux PMPM under French Defence Ministry
funding. The following tables show the principal parameters of their current machine
development. The authors are grateful to Jeumont Industrie for permission to publish this
information.
Rated voltage V between ( tbd )
phases
Phase Current A (8 disc 8 × ( tbd )
sides each fitted with 3
phase windings)
Frequency Hz 0.8
Overall weight 70 tonnes
External diameter 2.8m
Overall length 3.0m
Efficiency
@ 100% speed 97 %
@ 80% 97.1 %
@ 60% 95.5 %
@ 40% 93 %
@ 20 % 80 %
Table 3: ALSTOM Advanced Induction Motor Parameters
ALSTOM Advanced Induction Motors have been designed and manufactured at a variety
of ratings and phase numbers since their development for industrial applications in the
early 90’s. The machines are fully compliant with Noise and Vibration, EMC/EMI and
Shock requirements in accordance with the appropriate NES or DEFSTAN. Typical
parameters for an H-bridge, PWM inverter fed, 15-phase motor for delivery in 2004 are
shown in Table 3 and the construction of a typical advanced induction motor is shown in
Figure 2.
Figure 2: Alstom Advanced Induction Motor
OPTIMISED CONTROL OF ELECTRICAL MACHINES
Dr Chris French of Newcastle University has been working on a novel and much
improved method of controlling motors with power electronic converters. This work was
reported at the All Electric Ship Conference 2000 in Paris and is at Reference 11. The
method employed by the researchers maximizes the full potential of an electric machine
by optimizing the motors control as applied by the voltage waveform generated by its
associated converter. The technique utilizes the magnetic characteristic of the machine
that is determined during preliminary testing and ensures that the maximum torque is
produced for any given input power. The method is generalized for any separately excited
synchronous, permanent magnet or reluctance machine – including Axial and Transverse
Flux toplogies. The approach is equally applicable to generators where the method then
ensures the maximum electrical output for a given machine size; though, of course, the
generator would need to provide its output via a converter (as is proposed by the authors
for their electric ship power system). The technique ensures that the machine always
receives the optimum voltage waveform and a second advantage of the technique is that
torque oscillations – that may arise through saturation when a machine is driven close to
its magnetic limits (as with the UK MoD PMPM) - can be entirely and predictably
removed at all loads and speeds.
DEVICES AND CONVERTERS
DEVICES
The development of power electronic devices has continued albeit not at as high a rate as
formerly. The USA Office of Naval Research (ONR) funded Power Electronic Building
Block (PEBB) programme previously reported at References 3,4 and 5 has now
concluded with a replacement programme - Advanced Electrical Power Systems (AEPS)
– continuing the overall development of power electronic equipment for the USA electric
ship programme. The main development of power electronic devices has been performed
G1 Cathode
G2
Cathode
G1 G2
N MOSFET
Current Voltage
P
N
P
Anode
Anode
Figure 3: MTO Equivalent Circuits
by SPCO who now, with the acquisition of the development division of Harris Semi-
Conductors, develop both the monolithic devices (GTO, MTO Thyristor based) and the
VLSI (IGBT, MCT, FTO). The main recent success has been the integration of the two
technologies – monolithic and VLSI – into one device now termed by SPCO the Super-
GTO. The authors are pleased that SPCO have chosen this paper to announce this device
publicly for the first time and the previously unpublished short paper, Reference 12
produced by Dr Vic Temple is included in its entirety as an Annex. In order to place the
development of the Super-GTO in context the MOS Turn Off Thyristor (MTO), which
was described in detail in Reference 3, will be reviewed.
The MTO has been developed by SPCO Inc in Pennsylvania USA; an equivalent circuit
is given at Figure 3. The MTO follows the style of commutation found with a MOS
Controlled Thyristor (MCT), where, referring to Figure 3, the upper of the device’s three
junctions is short-circuited by a secondary MOSFET switch. This commutation process
can be simplistically considered to be one of conversion of the four layer GTO into a
three layer Transistor that then commutates by normal base voltage. Another view would
be to imagine the stored charge in the upper middle layer, responsible for the GTO’s
continuing state of conduction being drained away through the short-circuiting MOSFET.
In either case the need for externally produced, stored and injected current – necessary
for a GTO or force commutated Thyristor and part of the external circuits of an IGCT - is
removed. The commutation process becomes entirely internal to the MTO. With the
MCT these MOSFETs are integrated fully within the structure of the MCT itself.
Conversely the MTO employs MOSFETs that are external to the GTO.
Figure 4 shows the construction of a Super-GTO. An edited extract of Reference 12
follows to provide an indication of the devices advantages.
The hybrid MOS Controlled GTO – or Super-GTO - is a planar processed, fine-line GTO
EST lid top lid
MTO lid
layout bottom
top with
mounted
Pebb insulating off-FET’s
base or “sled””
Solder screened copper
bonded to “sled”
Figure 4: Super-GTO Construction
combined with a thinPak lid that simultaneously packages the FET control element. This
appears to provide both better GTO function and improved gate impedance. A list of the
advantages of this approach are summarized as follows and include:
• No costly single die special handling
• Many die per wafer is simple, leading to improved yields
• ThinPak is a lower cost package
• ThinPak is many time smaller and lighter
• No dry interfaces and only moderate mounting force needed
• Very low inductance and resistance with multiple gate and cathode contacts to
reduce GTO current non-uniformity
• 3 times higher switching frequency
• 10 times (or more) higher cell turn-off current capability
• Much lower forward drop
• Very uniform on-state and transient current distribution.
• Flexible heat removal.
CONVERTERS
The advantages of multi level converters, as reported at Reference 5, are now being
widely recognised this is mainly due to the fact that for the first time IGBTs are available
at sufficiently high voltage levels so that series connection is avoided for a converter in
the 10s of MW range. Both Ultra PMES in the UK and Power Paragon in the USA are
actively developing these converters.
In the case of the PMES converter the design is a Neutral Point Clamped topology which
effectively uses a capacitor network to create an effective neutral voltage which can be
used to provide a three level converter. When this is implemented with the 6.5 kV IGBTs
now available series connection is not required. A schematic for a single phase NPC
converter is shown at Figure 5.
DCPOS
IGBT Gate IGBT Gate
Drive PEC Drive PEC
IGBT Gate IGBT Gate
Drive PEC Drive PEC
OUTPUT
IGBT Gate IGBT Gate
Drive PEC Drive PEC
IGBT Gate IGBT Gate
Drive PEC Drive PEC
DCNEG
Figure 5: Ultra-PMES NPC Converter Topology
The NPC topology can also be applied to a poly-phase arrangement which will have
space and weight advantages although, in the marine environment, control of earth
circulating currents becomes more difficult.
The creation of intermediate voltages to enable multi-level conversion is not limited to
the neutral voltage; any other number of intermediate voltages can be developed through
a capacitor network. The same technique is used, although on a larger scale, with diodes
Figure 6: Power Paragon Diode Clamped Converter
being used to clamp circuit points to desired voltages. The topology may then be referred
to as a Diode Clamped Converter. The advantage is that with more intermediate voltage
levels in play the harmonic distortion of the converter is reduced, as are the overall
switching losses since each device switches at reduced voltage. Figure 6 shows a
schematic for the Diode Clamped Converter being developed by Power Paragon in the
USA.
The Multi Port Converter, reported at References, 3, 4 and 5 continues to be developed
by SAIC in the USA. The topology was recently recognised by ONR as having particular
merit and they have funded a $2M development programme aimed at producing a
working prototype of around 300 kW. The programme has only just started and the
authors hope to be able to report more on this topology in the future. The UK MoD
assessed the use of the converter for application in an Electric Warship power system and
it was noted that its inherent flexibility allowed its use to be contemplated in all areas of
the electric ship power system. Its inherent capability to integrate disparate electrical
power systems offers much to the power system designer.
GENERAL
HARMONICS
Power electronic devices are, by necessity, used as a switch and they have, by design,
very fast switching times and provide complete electrical isolation. As such they interrupt
current flow virtually instantaneously but, as they usually operate repetitively, they cause
significant steady state disturbances to both the current and voltage waveforms of their
power supply system; known as harmonic distortion.
Harmonic distortion has several deleterious effects including causing insulation system
degradation and heating of generators windings - thus a good knowledge of the harmonic
burden imposed by all power electronics equipments is necessary when designing
systems and specifying equipments.
One way of visualising the Electro-Magnetic Interference (EMI) problem caused by
harmonics is to undertake a Fourier Transform of the distorted waveform. Generally,
there is a large number of odd harmonics based on the power electronic switching
frequency, often 2kHz or higher. The lower order harmonics can be difficult to filter and
can propagate around the galvanically connected system and can feed into sensitive
equipment such as lighting, broadcast and telephony equipments. The higher order
harmonics are in the radio frequency and can easily transmit into the ship and, in the case
of a warship, can be received at the inputs to combat systems. As a typical combat system
has a high gain amplifier at the front end, severe degradation or even failure of the
combat system can result. In general, there are military specifications for the quality of
the power supply to equipment, which controls this interface. However, the cost and size
of suitable filtering to meet these specifications can be considerable and system design
must undertake trade-offs to determine where and how best to achieve an acceptable
harmonic distortion level.
The design of the filtering systems is never simple and one method becoming available is
that of system simulation. If accurate results are to be obtained the problem of simulating
harmonics is complicated. Due to the need to accurately take account of multiple
distorting loads interacting with each other it is necessary to work in the time domain and
at an integration step interval consistent with the switching frequency of the converters.
Even with today’s computing power, with increasing system size and numbers of power
electronic converters, this rapidly becomes almost intractable and the problem is
exacerbated in the marine field where high impedance power systems exhibit high
susceptibility to voltage distortion. While it is now feasible – though time consuming - to
conduct time domain simulations of limited power electronic esystems the overall scope
of the system is still limited by computing power, in addition the wide disparity between
the time constants of power electronics, the electro-mechanical generation and
distribution system and of course the ship dynamics itself effectively prevents a unified
simulation being developed. Indeed this may always prove impossible even with
increased computing power dude to the conflict between rounding errors (which prevent
small integration steps being used over large time intervals) and the short transient time
constants (which force small integration intervals to track rapid changes).
As a result it is likely that a dual approach will be required for the simulation and
modelling of marine electrical power systems for the foreseeable future. The challenge is
perhaps to form open simulation architectures that allow simple integration of the results
from time domain simulation using packages such as Power System Blockset into the
more traditional electrical power system frequency domain analysis packages such as
Viper.
In the particular case of harmonic distortion in a marine power system it should be
possible to 'map' the harmonic distortion of the power system over a range of operating
conditions generated through frequency domain analysis. These results could also be fed
into models of the weapon systems' detectors for assessment of military effectiveness.
The authors are grateful for the advice and assistance of The Mathworks (formerly
Cambridge Control) in the development of this section.
PROTECTION
As the capability of the power electronics and the intelligence that can be embedded in
the control increases, the option of using the power electronics for system protection
becomes more practicable. The issues involved are seen as galvanic isolation and losses.
Power electronics can fail short circuit and conducting maintenance where only silicon
isolates the maintainer from a potentially lethal power supply is undesirable. The solution
is relatively straight forward as off-load isolators can provide the galvanic isolation,
however, these all add cost, volume and weight to a system. This impact can be
minimised by allowing a larger proportion of the system to be de-energised when
maintenance is underway and in many respects the traditional system using switchgear
resolves this.
The other facet of power electronics is that when a silicon junction is conducting physics
demands that there is a volt drop across it. This in turn means that power is dissipated
across the junction whenever the device is turned on. At the current ratings of devices
with a full duty function, these losses are significant, as much as 2% of the power being
handled in a particular application. These losses are allowed for when designing the
equipment and are acceptable for such applications as Converters where the functionality
they offer is essential. However, to introduce another device in series such as solid states
switchgear, can double the losses adding significant inefficiency to the system.
In principal then, traditional switchgear, which has very little volt drop when conducting
and good galvanic isolation when open, remains the preferred choice for purely
protection functions. However, hybrid switchgear based on solid-state devices is being
developed and will become available at some stage. These are likely to use power
electronics to make and break the supply followed up by a mechanical conductor to
minimise conduction losses. Once available one of the disadvantages of electrical DC
distribution will vanish as the difficulty of extinguishing the arc drawn when DC contacts
open will have been resolved.
SECURITY OF SUPPLY
In today’s world, even outside of essential military applications, losses of electrical
supply are becoming ever more unacceptable. Indeed, the situation is exacerbated by the
fact that even a very short interruption or disruption of the supply waveform can cause
computer based systems to crash.
Even if single generator operation is not adopted with a particular Electric Ship system
there is a need to thoroughly address security of supply. The UK MoD Electric Ship
Technology Demonstrator is assessing a Zonal Power Supply Unit concept, which has
been outlined in previous papers and is based on energy storage.
The US Navy IPS programme has successfully demonstrated a similar capability to
ensure the continued provision of high quality supplies despite severe disruption from
failures or action damage – known as ‘fight through’ and this will be discussed later in
the paper. This concept is based on immediate switching between two separate
distributed supplies.
AES on the World Stage
GENERAL
Electric warship concepts are now being developed by several nations. The USA has a
fully funded development programme aimed at proving the technology for their
1100 VDC
Port 4160 VAC
Zone A Zone B
PMM-1
PCM-4 PCM-1 PCM-1
900 VDC Swbd 860 VDC
PCM-2 PCM-2
860 VDC
900 VDC
PCM-4 PCM-1 PCM-1
PMM-1
Stbd 4160 VAC 1100 VDC
Figure 7: USA IPS Schematic
implementation of an electric warship – this is discussed below – and have stated that
their new surface combatant, DD21, will have an integrated electric propulsion system.
As reported at Reference 13 the UK continues to further develop the electric warship
concept through their Electric Ship Technology Demonstrator and the Type 45 class of
destroyer will have an integrated full electric propulsion system. In addition France, The
Netherlands and Germany are developing equipment for an electric warship with Italy,
Spain and Japan, at least, undertaking concept studies.
THE USA
As has already been stated in this paper the concept of the Electric Warship is now
receiving significant attention in many countries. The USA is one of the leaders and the
authors are pleased to be able to report on their recent activities in this paper. The Electric
Warship Programme in the United States of America has two distinct facets. The
acquisition programme office for the replacement surface combatant, DD-21, has
announced that Integrated Electric Propulsion will be used in the new warship. In
addition the Naval Sea Systems Command has an active development programme
working towards implementing not only integrated propulsion and power systems (IPS)
but also of a topology and control regime that allows, in their terms, “Zonal Fight
Through”. A system exhibiting Zonal Fight Through will be resilient to damage and self-
healing to the extent that zones on either side of the damaged section will suffer neither
interruption of degradation of their power supply. It is important to note that although the
DD-21 acquisition programme office and the Naval Sea Systems Command IPS team
both refer to Integrated Propulsion Systems the system actually referred to in each case is
different with the acquisition programme office’s definition of IPS being much more
loosely defined. Indeed it is possible that the DD-21 final IPS system will not be that
being developed by the Naval Sea Systems Team.
PGM Power Generation Module
PDM Power Distribution Module
PCM Power Conversion Module
PMM Propulsion Motor Module
ESM Energy Storage Module
PCON Power Control Module
Notes:
• Where two differing equipments with the same function are used they are
discriminated by their numbering: hence PCM-1 and PCM-4.
• The PMM includes the Propulsion Motor and its converter.
• Not all the modules referred to in this table appear in the current IPS scheme –
ESM for example.
Table 4: USA IPS Component Designations
A diagram of the Naval Sea Systems Command IPS concept is at Figure 7 and shows two
zones though in practice there would be more, perhaps as many as six. The design and
operation of the IPS system is significantly different to the IFEP system originally
proposed by the authors at Reference 1 and so some explanation is worthwhile. The
system is conceived as having three 20 MW Gas Turbine Alternators and this increase in
propulsion power over the original UK concept is simply due to the different and much
larger ship displacement: perhaps 8 or 9 thousand tonnes for the USA ship against 3 of 4
thousand tonnes for the UK. In a similar fashion to the UK concept the propulsion and
ships service systems are segregated and AC and DC respectively with power electronics
performing the integration. However in the USA system all power generation is
performed on the (in the USA case) 4.16 kV AC 60 Hz system and therefore the
interconnecting power electronics need not be bi-directional.
The diagram at Figure 1 uses the Naval Sea Systems Command nomenclature and
abbreviations. These are explained in Table 4. At first inspection the system seems to be
inherently inefficient with two Power Conversion Modules in simple series where one
would suffice. The answer lies in the desire to implement robust Zonal Fight Through. By
using both PCM-4s and PCM-1s it is possible to choose which of the Starboard or Port
main 1100 V busbars provide power to individual zones. Thus in Figure 7 the Zone A
would receive its power from the Port DC busbar because its PCM-1 output voltage is set
at 900 V – higher that the Starboard PCM-1 (860 V). Conversely Zone B is receiving its
power from the Starboard busbar. In this way it is possible to control the loads seen by
the Port and Starboard busbars whilst at the same time guaranteeing uninterrupted supply
should one of the main 1100 V DC busbars fail. It is of interest that the PCM-1 module
Full Zone 1/2 Zone
Load
PCM-4 PCM-1 PCM-1
PMM-1 Bank
PCM-2 PCM-2
PGM-1
PDM-1 AC
Swbd Load
PGM-3 Bank
Load
Bank
PCM-4 PCM-1 Prototypical
Functionally Equivalent
LBES Use only
Figure 8: Philadelphia IPS Test System Schematic
started as a solid state DC circuit breaker, but was re-used when the principal use became
voltage conversion to achieve zonal fight through.
This concept is being extensively tested at the USA shore test facility at Philadelphia.
This facility is part of the Naval Surface Warfare Centre – Carderock Division
(NSWCCD) and is termed the Advanced Propulsion and Power Generation Test Site
(APPGTS). The APPGTS has been under development since January 1993 and although
originally planned for testing of the US Navy’s Intercooled Recuperated (ICR) gas
turbine engine development program, the facility has proven its versatility, having
provided a venue for performance testing of the Advanced Turbine System (ATS)
compressor for Westinghouse Electric Corporation prior to being converted for testing of
the IPS.
In order to limit costs the IPS system is reduced in scope and in addition some
components are not fully representative of equipment that would be used in a warship.
3000 VOLTAGE PGM-1 VOLTAGE
5000
2800 3500
2600 4500
2400 3000
4000
2200
LM2500 POWER TURBINE SPEED
2000 3500 2500
1800
3000
1600 2000
1400 2500
1200 1500
2000
1000
PGM-1 CURRENT
1500
800 1000
600 1000
400 PCM-2 VOLTAGE 500
500
200
0 0 0
200 205 210 215 220 225 230
SPEED Time Elapsed (sec.) CURRENT
Figure 9: Philadelphia IPS Results: Load Transients
However there is in general at least one fully specified example of each of the equipment
that would be required to implement IPS in a warship. It is of note that the Propulsion
Motor Module is an Alstom 15 phase Advanced Induction Motor rated at 20 MW and
150 RPM together with its associated Alstom 20 MW PWM series connected IGBT
converter. The initial testing results from Philadelphia have been reported at Reference
14 and it is clear that the system is indeed extremely robust to damage and failure.
Testing continues but early results are extremely encouraging. The total harmonic
distortion levels are higher than the original target but nevertheless the harmonic
performance is reasonable and the original target for total harmonic distortion was
extremely demanding. A schematic of the Philadelphia IPS test system is shown at Figure
8. The testing to date has included efficiency measurements and system transient
responses. Perhaps the most remarkable result is the stability of the in zone 450 V 60 Hz
three phase supply during loss of one 1100 V DC busbar – no transient appears on the
converted supply. Figures 9, 10 and 11 show some of the results obtained for the
Philadelphia test programme.
45
ORIGINAL PREDICTION
TOTAL HARMONIC DISTORTION (%)
40
REVISED PREDICTION
35 MEASURED DATA
30
25
20
15
10
5
0
0 20 40 60 80 100
PERCENT OF RATED POWER
Figure 10: Philadelphia IPS Results : Harmonic Distortion
As can be seen in Figure 9 the voltages remain stable during load shedding but most
notably the Gas Turbine speed is controlled satisfactorily during the load reduction.
Figure 10 illustrates the level of harmonic distortion present in the 4.16 kV system and
100
90
80
PERCENT EFFICIENCY
70
60
50
40
30
20
10
0
0 20 40 60 80 100
PERCENT POWER
Figure 11: Philadelphia IPS Results : System Efficiency
Figure 11 the variation of efficiency with system load and this is particularly noteworthy
as it demonstrates the way that the overall system efficiency for an Integrated Electrical
Propulsion System remains high throughout the majority of the load range.
The following tables give a more analytical assessment of the success of the IPS system.
CRITERIA NOMINAL TARGET RESULTS
VALUE
Frequency Level 60 HZ +/- 5 % + 0 % / - .1 %
Frequency Droop 3.3 % @ rated +/- 1 % 3.3 % @ rated power
power
Voltage Level 4160 V +/- 10 % +2%/-0%
Voltage Droop 3.0 % @ rated +/- 5 % Within – 1.5 %
power Over power range
Current Harmonics NA 291 A or 8 % of 185 A or 7.2 %
(PGM-1 design 5th IHD @ rated power @ 5th IHD
limit)
Voltage Harmonics NA Any IHD 4.5KV GTO’s on a 6
inch float zone wafer running an IC foundry whose highest voltage prior device was of the order of 100
volts. Of course, the immediate advantages of small size and fine geometries at extremely large process
yields was one of the advantages achieved. In fact we chose to a 2 mil cell repeat distance.
The next problem is how to interdigitate high current gate and cathode contacts with very small
(microns) separation. The solution here was in part the use of a second metal and in part SPCO’s
thinPak package technology which is described in some detail below. The final problem was how to
achieve stable, high voltage breakdown voltage without the normal deep, closed tube diffusion. For this
SPCO turned to its JTE approach.. Given that the combination of fine geometry, thinPak package and
JTE termination worked we would expect a rather remarkable turn-off capability of well over
1000A/cm2.
Figure 1 shows how such a device could be incorporated, using a thinPak lid approach to minimize
inductance, to make various MOS controlled or gated alternative high performance GTO-like or based
devices. Note that a small MCT can be used as the MOS gated turn-on amplifying gate with the unusual
advantage that it can remain gated on and will re-gate the GTO as necessary if the current in the main
circuit oscillates through zero, for example.
Standard MTO MTO Block w/ high Standard IGCT
performance turn-on
Advanced MTO Block Emitter Switched IGCT Block w/ high
w/ high performance Thyristor Block w/ high performance turn-on
turn-on, ultra-high performance turn-on,
current turn-off and ultra-high current
current sense turn-off, current sense
and current limiting
Figure 1. Advanced hybrid GTO-like devices with MOS control.
Figure 2 shows how one might make an MTO power module using one of these “superGTO’s”. The
base is aluminum nitride upon which are mounted several S-GTO’s each lidded with a thinPak lid. The
thinPak lid top sides are patterned to mount the gate-cathode FET die or surface mount pre-packaged
FET’s as the case may be. Also shown is the MCT turn-on device with a series diode that is intended to
allow the pilot MCT to run at low current in the on-state, ie on-state MCR/diode drop is to be
considerably larger than the S-GTO..
insulating substrate
metal matrix heat spreader
low R-on FET diode (or resistor)
device metal thinPak lid
planar super-GTO high voltage MCT
power electrode
Figure 2. Hybrid MTO example.
Figure 3 shows the edge regions, including the JTE termination of our high voltage MCT and S-GTO
devices. In the figure the JTE zone is the varying doping P- region at the edge of the device whose goal
is to allow us to rate at a very high percent of theoretical one-D breakdown voltage and to
simultaneously reduce surface field to less than one half the peak bulk electric field in order to enhance
breakdown voltage reliability.
Figure 4 shows experimental data from an 18 mil thick set of float zone devices with a 4KV one-D
theoretical BV. 75% of our tested devices with the optimum JTE implant dose made our 90% design
target. The inset table provides an insight into another of the advantages of JTE, namely the much
smaller wasted device area which is typically reduced factors of 3 or higher. For example a 5 mil deep
one degree bevel uses an edge width of just under 400 mils and achieves 5340 volts (see table) while
about 35 mils, 10 times less, are needed for an equivalent BV using JTE.
SGTO termination and cell
MCT termination and cells
Figure 3. SPCO JTE terminated high voltage devices.
5 KV standard edge bevel 5 KV JTE terminated device
Ideal BV 6000 V Ideal BV 6000 V
1 degree bevel 5340 V 30 mils 5180 V
2 degree bevel 5040 V 45 mils 5690 V
3 degree bevel 4740 V 60 mils 5890 V
Pebb device BV vs JTE implant step dose
Breakdown voltage
4,000
75%
TRP phase 1:
3,500
25% thinPak packaged
3,000
JTE terminated
device breakdown
2,500 voltage results for an
18 mil thick substrate
2,000
1.5 2 2.5 3 3.5
JTE dose (e12/cm2)
Figure 4. Top: comparison (modeled) between bevel and JTE terminations. Bottom: experimental BV
results for 18 mil thick devices as a function of JTE implant dose.
One of the key advantages of JTE is that the upper p-base junction need no longer be as deep. In the
case of the devices shown in figure 4 we chose a depth of the order of 10 microns, about an order of
magnitude thinner than in standard GTO’s. With the cathode finger width being so narrow we were also
able to increase upperbase sheet resistance and still expect to turn off very high current densities. The
net result is that the upper transistor gain in our S-GTO is very high. This allows us to make the gain of
the lower transistor much smaller than usual. The net result is a vastly improved tradeoff between
forward drop and turn-off energy. In our detailed modeling turn-off energy is greatly reduced and
forward drop decreased. Part of the advantage comes from the very density of the fingers which results
in uniform current density through all but the upper 10 or 20 microns of the S-GTO. Forward drops of
under 2 volts are easily achieveable at current densities of 100A/cm2, a current density that is very much
higher than the rated RMS current densities of standard GTO’s. Figure 5 shows the sub 2 volt forward
drops of the 3600 volt devices of figure 4 at 25C and at 125C for un-irradiated devices (slow) and
irradiated devices of about half the turn-off loss.
Thick, substrate based M’top Pebb diode with sub-1.5V Vf at 100A/cm2
Pebb device forward drop at 25C and 125C
Current
400
Dev A
300 25 C
200 Dev B
25 C
100
0 Dev A
.75 1 1.25 1.5 1.75 2 2.25 2.5 2.75 125 C
Forward drop
Device A: no radiation Dev B
Device B: 0.2 MR radiation 125 C
Active area: 1 cm2
Experimental Phase 1 TRP results for 18 mil
thick, ~4KV device.
Figure 5. Demonstration of low forward drop at 100A/cm2. The device shown has a 1 cm2 active
device area.
Status
What has been presented above are good arguments for expecting a JTE terminated, thinPak packaged
S-GTO to provide superior hybrid GTO-based devices and modules. Presently, SPCO has generated
masks and run several lots of S-GTO’s at its California foundry. Before agreeing to contribute to this
paper we met the key milestone of achieving high breakdown voltage, ie >4500 volts, on >50% of the
tested die on our most recent lot. Figure 6 below shows a completed die.
SGTO devices on a 6”
Figure 6 34 S-GTO’s on a completed wafer.
The metalization pattern that is visible is the top layer of a two metal layer contact scheme. The stripes
are alternatively gate and cathode with each stripe making tens of contacts to each of the fingers which
run at right angles. The ceramic thinPak lid mates exactly, on its underside, to the device metal. On its
top the gates and cathodes are combined into two metal contacts as seen in figure 7 where the lid for a
top mounted series FET for an emitted switched GTO is shown on the left and the lid for top mounted
MTO FET’s is shown on the right. For the MTO lid, the metal contacts from left-to right are 1) GTO
cathode, 2) GTO gate, 3) FET gate, 4) FET gate return, 5) FET gate, 6) GTO gate and 7) GTO cathode.
With the 8 FET’s installed, each rated at 4 milliohms and < 2 nH inductance, we would expect to have <
0.5 uH and about 0.5 mOhms of gate-cathode turn-off path impedance for about 2 cm2 of active GTO
silicon.
If one tracks the various gate turn-off current paths one finds that all parts of all fingers are similar and
very short. Therefore, we have the expectation of very uniform current in both on-state and transient
condition. This is very different from the mechanically contacted fingers, all of which are contacted at
different local pressures at random high points and with varying contact pressures.
The lid provides another function in that it provides additional thermal mass that enhances 60 hertz
surge current rating. This is because the lid is so well thermally connected to the die.
Although the cooling of the figure 2 module is only one side, the module form allows the heat to be
spread over a much larger effective area than does, say, the center square centimeter of the pressPak
packaged GTO with a corresponding advantage in thermal resistance which more than off-sets (there
also are no dry interfaces.) the 2-side cooling advantage of the pressPak.
EST lid top lid
MTO lid
layout bottom
top with
mounted
Pebb insulating off-FET’s
base or “sled””
Solder screened copper
bonded to “sled”
Figure 7. ThinPak lids for MTO and EST hybrids. The designed surface mount FET’s for our initial
experiments are shown on the bottom right.
Summary
This paper has described the advantages of a hybrid MOS Controlled GTO. Its planar processed, fine-
line GTO combined with a thinPak lid that simultaneously packages the FET control elements appears
to provide both better GTO function and improved gate impedance. A list of advantages of this
approach are summarized as follows and include:
• No costly single die special handling
• Many die per wafer is simple, leading to improved yields
• ThinPak is a lower cost package
• ThinPak is many time smaller and lighter
• No dry interfaces and only moderate mounting force needed
• Very low inductance and resistance with multiple gate and cathode contacts to reduce GTO
current non-uniformity
• 3 times higher switching frequency
• 10 times (or more) higher cell turn-off current capability
• Much lower forward drop
• Very uniform on-state and transient current distribution.
• Flexible heat removal.
Acknowledements: To Steve Arthur for key contributions to high voltage diodes and MCT’s and to
Dante Piccone for SPCO’s first hybrid FET/GTO devices. To Clarence Severt (Wright Patterson AFB)
and to Terry Ericsen (ONR) for their support of advanced devices and packaging through the TRP and
Pebb contract support. To Sabih Al-Marayati and Forrest Holroyd for their key roles in device design
and fabrication.