Electric propulsion

The electric warship Integratedfull electric propulsion has become a normal first choice for several commercial operators, in particularfor cruise ships where it proves to be afuel efficient transmission system. Operating profiles for warships offrigate sire and above are similar to cmise ships with long periods at well belowfull power Howeve? the size, weight and initial cost of electric propulsion equipment has generally precluded selection for warships. With the application of modem technologies, equipment power densities are being increased such that integrated full electric propulsion has now become a serious competitor to the mechanical transmission systems traditionally adoptedfor warships.

by D. S. Parker and C. G. Hodge

arships are generally designed with a maximum speed around 30 knots, yet spend much of their W time at slower speeds, typically

80% below 18 knots. The concept of ‘cruise’ and ‘boost’ propulsion capability allows machinery alignments that minimise fuel usage. Discussion is confined to a frigate of some 4000 tonnes and this requires development of compact high power motors and associated converters if the equipment is to be suitable for installation in that size of hull.

Electrical transmission With electrical transmission, generators convert prime mover rotational power into electricity which is transmitted to motors which convert electricity back into rotational power. The two conversions in the transmission train result in greater losses than that of a geared mechanical system; however, the efficiency of electrical transmission does not fall off as quickly with shaft speed as mechanical transmission and it can be more efficient at lower speeds.

Efficiency comparisons cannot be made without considering all aspects of power usage and, while in a conventional mechanical system propulsion and services systems have separate prime movers, the electrical system is designed such that all prime movers can be used for both duties. This leads to a flexible, efficient system

known as integrated full electric propulsion (IFEP). The flexibility of such a system with an optimised prime mover arrangement leads to greater efficiency at all power levels and lower unit procurement costs (UPC) and running costs (RC). In the case of RN frigates, a conventional system with mechanical transmission has eight separate prime movers, a hybrid system has six and a future frigate with IFEP might have four.

The benefit of IFEP is maximised when much of the operating profile is well below maximum speed and when the ship service load is a significant proportion of the usual propulsion load. The prime mover running costs for a frigate with mechanical transmission and IFEP are compared in Fig.1 for the three

operating profiles of Fig.2.’ 1 Reduction of prime Studies of IFEP systems suitable for warships

undertaken by MOD2 indicate that to make IFEP the natural selection in a cost constrained

running costs for IFEP over mechanical transmission

operating pmflls saving mrceot fik per annum

anti submanne warfare 374 458 36 0 G? 1 gensrd p c pose

carrier escort 29.13 765

POWER ENGINEERING JOURNAL FEBRUARY 1998 5

Electric prvpulsivn

2 Operating profiles

key: - ASW -GP - CBG 10

8

.: - 6 ”.- - S

E 4 a

2

0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

speed

environment requires the motor and converter cost, weight and volume to be reduced.

Increasing the power density of motors and converters This part of the article reviews the recent advances, principally in permanent magnet technology, motor design and power semi- conductor devices and variable speed drives, that are ready to reduce the physical size and mass of an IFEP system such that its bulk compares favourably to that of its mechanical equivalent and, further, offers a much more flexible approach to the physical location of equipment with consequent benefits to the overall design of the ship itself.

Permanent magnet technology Recent advances in rare earth permanent

3 Radial flux PM magnet materials have led to the introduction machine of neodymium iron boron (NdFeB) and

samarium cobalt (SmCo) which have the necessary magnetic properties required for high power density marine propulsion applications. While both NdFeB and SinCo possess the required magnetic properties, they differ greatly in cost, mechanical strength, corrosion resistance and temperature performance. The choice of material is dependent on the exact application, and this primarily concerns the mechanical arrangement and thermal profile within the machine.

Motor design Conventional machines appear to be approaching full development and whilst considerable improvements in power density are being achieved with direct water cooling these are gained at the expense of efficiency and at higher initial cost. Although they are available to suit the majority of drive applications they do not suit the efficiency and power density demands of warship main direct drive propulsion motors. Furthermore the traditional topology constrains the machine’s torque capacity: there are two

reasons for this:

(i) The armature conductors and the machine’s magnetic flux compete for the same space around its circumference. If the flux is increased by designing more iron into the magnetic circuit then the electrical loading of the armature winding must be reduced since it is now allowed less space for its conductors, and vice versa.

6 POWER ENGINEERING JOURNAL FEBRUARY 1998

Electric propulsion

(ii) The torque in a conventional machine is developed in the annular cylinder of the air gap and the armature and field windings. This represents a small percentage of the overall size of the machine and offers scope for torque density improvement if the packing density of the active components could be improved.

To achieve performance and operating characteristics required for naval applications, modern machine developments are centring on the use of permanent magnets to provide the excitation for machines which have high power and torque density. This use of permanent magnets arises principally from the new topologies which they are able to engender and brings three benefits: flux concentrating designs can be implemented, the packing density of the active components can be increased and the magnetic (iron) and electric (copper) circuits of the machine can be separated, such that they no longer compete for the same space. Taken together these allow increases in power and torque density of up to eight times that normally thought possible.

In a permanent magnet motor the conventional rotor of an AC machine is replaced by an arrangment of permanent magnets, the orientation and topology of which greatly changes the design and performance of the motor. The applicability and potential of the differing topologies makes permanent magnet machines suitable for both direct and indirect applications for surface ship and submarine, main and auxiliary drive applications.

Permanent magnet radialflux machine The permanent magnet radial flux machine, the most traditional of the novel machine designs, is shown in Fig.3 and incorporates a conventional armature winding with the conductors aligned at right angles to the direction of the machine flux. This design still suffers from the conventional machine torque constraints.

Axial flux machine The axial flux machine, also known as the disc motor, is shown in Fig.4 and incorporates a rotor and two stators, each of which is disc shaped with the axially magnetised permanent magnets located in pockets in the rotor disc. The permanent magnets drive the magnetic flux across the two annular air gaps into the stator core. The current in the stator winding coils interacts with the flux generated by the

magnets producing a tangential force, and the 4 Disc machine

machine torque results from the contribution of all these forces. To maximise the packing density of the active components within the machine the stator armature coil windings have low radius end curves; this produces a high number of poles and is the cause of the axial machine’s disc shape.

Transverse flux machine The transverse flux motor is a novel topology which overcomes many of the constraints of the preceding topologies by implementing, as its name suggests, a transverse flux arrangement which optimises a flux concentrating principle to improve the electromagnetic performance of the machine. The basic arrangement of the machine is in Fig.5 and consists of a circular coil co-axial with the rotor3

5 Basic arrangement of transverse flux

stator winding

I rotor

POWER ENGINEERING JOURNAL FEBRUARY 1998 7

Electric propulsion

6 Flux path in tran~erseflux machine

The stator winding links the flux generated by the permanent magnets by means of a series of stator hoop pole pieces. The flux path is shown in Fig.6. The configuration incorporates stator pole pieces inside and outside the rotor which improves the electromagnetic performance of the machine with useful torque being developed at both the inner and outer surfaces of the rotor.

The most recent design proposed for a 180 rpm 20 MW transverse flux motor is shown in Fig.7 and has been designed to improve both the electromagnetic and mechanical performance of the m a ~ h i n e . ~

Podded drives Recent developments have now produced podded propulsion motor drives that are approaching the requirements for propulsion of a 4000 tonne frigate. Siemens and Shottel have jointly developed a 14MW 150 rpm podded propulsor with a novel topology permanent magnet motor incorporated within the drive unit.5

Power semiconductor devices Power semiconductors have been employed in various motor drives for many years, but the pace of change has now accelerated to previously unimagined levels. Devices such as GTOs are driven by current flowing in the gate circuitry and these have thus been slow to commutate or switch off. The phase of commutation, when the device moves from a state of low voltage and high current to one of zero current and maximum voltage, is one of intrinsic high loss when compared to the two stable states it separates.

Current drive devices being relatively slow to switch have been generating significant heat in this phase, the resulting burden on cooling has made the overall drive bulky and expensive. However, recent developments have created devices driven by gate voltage which has allowed a much quicker speed of commutation and thus reduced losses and consequently

7 Recent proposed design for transverse flux machine

8 POWER ENGINEERING JOURNAL FEBRUARY 1998

Electric propulsion

allowed smaller drive packaging. In addition the frequency of switching has been increased such that pulse width modulated (PWM) frequency conversion is now possible with very little harmonic distortion, either downstream or upstream.

Gate turn off thyristor Until recently, and perhaps still, the majority of high power static converters have used gate turn off (GTO) thyristors as the semiconductor technology. These devices, unlike their thyristor counterparts, can be switched off without the need for forced commutation circuitry. GTOs have therefore allowed a reduction in volume and an increase in switching frequencies. Commercially available GTO devices have voltage and current ratings of the order of 4 kV and 3.5 kA, respectively.

Although GTO devices are an improvement over thyristors they are current driven devices and require a current pulse to be applied to the gate to ensure that the device changes state. In the particular case of switching off, the device requires a negative current pulse of approximately 20% (compared to 100% in a thyristor) of the current in the GTO to be

collector 1 J

gate J emitter

applied to the gate, albeit for an extremely short duration which requires complex and bulky control circuitry.

8 IGBT equivalent circuit

Insulated gate bipolar transistor In recent years the thrust of development has been towards improving the power density of static converters and new devices have emerged to compete against the GTO. The front runner of these new devices is the insulated gate bipolar transistor, or IGBT. The name derives from the fact that the gate is insulated from the

POWER ENGINEERING JOURNAL FEBRUARY 1998

HMS Nor folk

9

Electric propulsion

anode

g a t : d l 7 c cathode

9 MCT equivalent circuit

+

DC in

I U - I

10 Resonant pole converter

I I

1 1 Emitter switched MCT equivalent circuit

10

device itself (in this case a transistor rather than a thyristor) and in terms of its external characteristics the device becomes driven by the voltage at its gate This insulation of the gate of a traditional bipolar transistor is achieved by drivlng the gate through a MOSFET (metal oxide silicon field effect transistor) whose high gate-to-device impedance is achieved by incorporating a layer of silicon oxide between the gate and the demce itself The recent development of IGBTs has allowed these double device packages (MOSFET and bipolar power transistor) to be implemented monolithically on one slice of silicon

IGBTs are commercially available up to a voltage and current rating of 1600V and 800A, although higher current ratings are available at lower voltage ratings These are obviously well short of the GTO ratings but the devices have other advantages and are still rapidly evolving Their major advantage is that they are voltage driven and hence the control circuits are much simpler, with a commensurate reduction in weight and volume The IGBT can also be switched at frequencies greater than ten times that of the GTO, in the order of 20 kHz This has the effect of reducing the amount of filtering required and gives a further reduction in overall weight and volume In addition the module packaging incorporates an insulated base plate which simplifies the cooling arrangement and can allow the use of non- demineralised water An equivalent circuit diagram for an IGBT is given in Fig 8

An important point is the presence of the parasitic transistor which is connected in thyristor configuration with the main transistor This is an inevitable consequence of the manufacturing process and accounts for the main failure mode of the IGBT If the main thyristor is driven too firmly into the on state the parasitic transistor can clamp the device into conduction as if it were a thyristor, under these circumstances the MOSFET is incapable of switching the device into the off state

The status of the IGBT as a rapidly maturing technology is now affirmed The latest developments aim to produce a 3 kV rated IGBT with corresponding 1600A current rating The difficulties of operating IGBTs in series are being overcome and the present view of the limiting power rating of a PWM IGBT converter is already above that required for the electric ship 20MW PWM IGBT converters are being designed and at least one is being built

POWER ENGINEERING JOURNAL FEBRUARY 1998

Electric propulsion

MOS controlled thyristor The MOSFET-controlled thyristor (MCT) has many similar facets to that of the IGBT: both are bipolar semiconductor devices with their gates driven by MOSFET transistors, and both are therefore voltage driven devices capable of very high switching frequencies. As can be seen from the equivalent circuit diagram of an MCT in Fig.9, the topology of the MCT is superficially very similar to that of the IGBT.

However, now the presence of the second bipolar transistor is very much part of the design and is a fundamental element of the devices operation. The two MOSFETs work together: one switches the device on as a normal thyristor, the other activates when the device is to be switched off and short circuits the upper of the three junctions present in a thyristor. In effect this turns the device from a four-layer thyristor to a three-layer transistor and thus allows the device to commutate off by normal transistor action. The original MCT was a p-channel device and unfortunately not capable of hard switching when the MCT had to interrupt both current and voltage, limiting the first generation of MCTs to about 600V and 65A. However, the following developments are underway which might affect the suitability of MCTs for use in power circuits:

Resonant pole converters: Resonant pole converters are beginning to be developed; a single phase limb of such a converter is shown in Fig.10. The principle is that the auxiliary switch, when fired, creates a current and voltage transient that allows both the auxiliary and main switch to commutate under soft switching. The main switch commutates under zero volts but high current; the auxiliary switch commutates under high voltage but zero current. This soft switching operation could allow the use of p-channel MCTs to be used at much higher ratings than they currently can achieve.

Emitter switched MCTs: Emitter switched MCTs, for which an equivalent circuit diagram is given in Fig. 11, have been developed and use a MOSFET in the main emitter circuit to switch the device off. Although these devices are robust and capable of hard switching they suffer from the high conduction losses which arise from the ‘off‘ MOSFET being permanently in the main power circuit. 0 h’-channel MCTs: New topology n-channel MCTs have been produced which are easier to manufacture and exhibit significant ability to

perform under hard switching regimes. HMS /nvincib/e Although not fully developed or marketed yet these devices will represent a real alternative to the IGBT for high power use. A robust MCT with a rating of above 300A and lOOOV when hard switching without snubber circuits should confidently be expected to be marketed within the next year.

Variable speed drives Having assessed the design implications of both available and developing semiconductor devices the drive technologies that are suitable for variable speed marine drive applications6 are briefly reviewed.

Sy nchroconverter The synchroconverter is essentially a two-stage conversion drive for a synchronous motor

POWER ENGINEERING JOURNAL FEBRUARY 1998 1 1

Electric -womdsion

500-1 000 V DC 6 6 - 1 1 kV AC

12 Electric ship consisting of first-stage rectification with a architecture current fed second stage from an intermediate

line reactor. Control of the rectifier thyristor firing angles governs motor voltage and the magnitude of the link current which ultimately determines motor speed. The inverter thyristors are fired to ensure correct phase sequencing of the link current to the motor and are commutated by motor generated EMFs. In this way the motor is synchronised to the shaft and the output inverter bridge is dependent on the synchronous motor back EMF for its commutation voltages; thus it is a load commutated inverter. The drive readily achieves four quadrant operation for reversal and regeneration by appropriate phasing of rectifier and inverter bridges. HMS Splendid

Cy cloconvevtev As previously mentioned the cycloconverter uses the basic fully controlled bridge rectifier to generate a variable voltage output which by correct phasing of the firing angle of the devices can be made to approximate to a sinusoidal variation. Since each bridge needs to accept bidirectional current two fully controlled bridges are required in each output phase, connected in anti-parallel. Therefore, for the simplest three-phase arrangement a minimum of six bridges is required, each with at least six devices. Thus the cycloconverter has many more devices than an equivalent PWM inverter or synchroconverter, in addition the control techniques necessary to ensure safe transfer of current between the two bridges in each arm across the current zero (which will not in general align with the voltage zero) tend t o r complicate the converter and also introduce more harmonic distortion than might at first be expected.

Pulse width modulated converter The pulse width modulated converter is, like the synchroconverter, a two-stage conversion drive but the fundamental method of speed control is different. The inverter is self commutating in that forced commutation devices such as GTO thyristors are used which therefore removes the dependence on the synchronous motor load of a synchroconverter. The semiconductor devices are controlled using PWM techniques which can drive the motor with near sinusoidal voltages; however, this requires devices capable of much higher switching frequencies than either a synchroconverter or cycloconverter. The resonant pole converter is a type of PWM converter.

The. pace of development of both semiconductor devices and the converter topologies that they enable are so high that any prediction about the type of converter to be used in an IFEP ship is uncertain. However, it is thought likely that an IGBT based PWM converter will at least be a strong contender if not the self evident first choice.

The electric warship With the advent of high power density motors and drives, as described above, IFEP is likely to be the next major propulsion change for frigate sizes warships. However, there is likely to be an impact on the remainder of the electric warship. There are various scenarios of which

12 POWER ENGINEERING JOURNAL FEBRUARY 1998

Electric propulsion

this article outlines one Fig.12 shows a conceptual electric power

system, consisting of two galvanically separate systems. Using a high-voltage AC propulsion system reduces the currents to be supplied to the propulsion motor converters. The DC ship service distribution system is battery backed. The novel heart of the system is the two reversible inverter rectifiers which will link the two systems. With these, any engine or the battery can supply power to propulsion and ship services. If a submarine battery is used to support the power system, it can provide the action load and propel the ship at up to 12 knots for half an hour. The 6- 10 MW cruise engine will support the maximum activity load and provide sufficient propulsive power for about 80% of the operating profile. As such, this will be the most used engine. If the complex cycle boost engines are not speed constrained one should be able to support the anchor load indefinitely without reducing reliability or increasing the maintenance load.

Turning now to the distribution system, Fig.13 shows the two sides of the DC ring main passing through one zone of the ship. Within that zone there will be a power source, either a prime mover or a battery section, and supplies required by various consumers will be generated by zone power supply units (ZPSU). Each ZPSU will be supplied from both sides of the ring main and will output the required power supply to consumers. Essential consumers will receive duplicate supplies and non-essential consumers will receive a single output.

The ZPSUs are visualised as an assembly of identical, compact, intelligent, programmable power electronic inverter modules. Each module will be capable of generating one of a variety of outputs and can be configured by software if moved between locations and duties. The design will enable the inverter module to generate a wide variety of the special requirements currently generated by weapon equipment power supply units.

Conclusion This article has presented the concept of the electric warship. Although much remains to be done, nothing appears likely to challenge significantly the view that the electric warship will not only be cheaper to own, simpler to operate, easier to maintain, and safer to use, but more importantly, that it will be much more

- bulkhead ---+

-

500-1000 V DC ring main

bulkhead \r

J- I

.... 5...

I 500-1 000 V DC ring main

effective at war, possessing hitherto 13 Electric ship power unimagined resilience to action damage. It is, in short, an opportunity which the authors do not believe the Royal Navy can afford to miss and thus is an essential strategy that requires activity to commence in the near future.

distribution system

This article is based on a paper presented at the Electric Machines and Drives Conference held at the University of Cambridge on the 1st-3rd September 1997

References 1 STAPERMSMA, D.: ‘The importance of (e)mission

profiles for navel ships’, INEC94, Cost Effective Maritime Defence, Paper 6 , Figure 2, IMarE, September 1994

2 MOD(PE)Report, D/SSUFP412/503/01/03 dated 22nd July 1994. Report of FEP Phase 1 Studies (unpublished)

3MITCHAM, A. J., and DULLAGE, B.: ‘A novel permanent magnet propulsion motor for future warships’, INEC94, Cost Effective Maritime Defence, IMarE, September 1994.

4 MITCHAM, A. J.: ‘Motors and drives for surface ship propulsion: comparison of technologies’, Marine Electrical Propulsion Conference, IMarE, October 1995.

5ANDERSON, P., and GRAGEN, U,: ‘New type of permanent field machines for diesel electric propulsion systems’, All Electric Ship Conference, SEE, Carre De Sciences, Paris, France, March 1997.

6 KALLAH, A. S., and MURPHY, M.: ‘Electric propulsion systems-selecting the right system’, Marine Electrical Propulsion Conference, IMarE, October 1995

OBritish Crown Copyright 1998hlOD: Published with the permission of the Controller of Her Britannic Majesty’s Stationery Office.

David Parker is Senior Electrical Engineer, Directorate of Marine Engineering, Ministry of Defence, Bath, UK. He is an 1EE Fellow and a Fellow of the IMarE. Chris Hodge is Head of the Power Systems Group, Directorate of Marine Engineering. He is a Fellow of the IMarE.

\

POWER ENGINEERING JOURNAL FEBRUARY 1998 13