Trans IMarE, Vol 105. Part 1. pp 23-52 Cycloconverter drives for ship propulsion * K S S m i t h , BSc(Eng), PhD, AMiEE, *R Y a c a m i n i , BSC, MSC, cEng, MiEE. FIMarE a n d t A C W i l l i a m s o n , BSC, PhD, CEng. MIEE *Department of Engineering. University of Aberdeen and fDepanmeni of Electrical Engineering, UMIST SYNOPSIS The Royal Navy are considering, amongst other options using electrical propulsion, employ ing variable frequency cycloconverters on the next generation of anti-submarine warfare frigates. The ongoing work in the Engineering Department of Aberdeen University is addressing some of the challenges advanced by electrical propulsion. This paper concentrates on the electrical characteristics of cycloconverter drive systems. INTRODUCTION The subject of electrical motor propulsion of ships is one which is being addressed for future use in both the merchant and defence fleets. This paper examines the types of drive units which are being proposed and then sets out to study the system which at the moment looks to be the strongest candidate for use in the next generation of Royal Navy frigates. The power electronic unit considered in most detail is the cycloconverter driving an induction motor. The operating principles of the cycloconverter are explained and typical examples of this converter's use to date are indicated. The advantages offered by this propulsion method are discussed, as is a method of modelling the cycloconverter and elecuical machines using a lime domain simulation. Using these computer models it is possible to predict the performance of the complete marine propulsion system by including the synchronous generators, power electronic convener, and the propulsion motor in the simulation. The computer models are general purpose and are not restricted to any particular drive configuration. Different convener connections and motor arrangements can be readily incorporated into the simulation. This is demonstrated in this paper by examples. ELECTRICAL PROPULSION OPTIONS The possibilities of using electrical propulsion rather tiian gas or steam turbines or diesel sets is not anew idea. During the 1920s and 30s the United States aircraft carriers Lexington and Saratoga had turbo-electric drives and the cruise liners Normandie and Scharnhorst were both electrically driven. However, recent advances in power electronics and elecuical machine design suggest thai the use of electrical propulsion will be a cost-effective alternative to the mechanical systems which are currently most commonly used on vessels. Electrical propulsion has already gained a foothold in the industry as drives for large cruise liners and icebreakers. In the U K , the best known example to date is the much publicised replacement of the Queen Elizabeth II steam turbines with two dc link inverter systems supplying synchronous machines which are directly coupled to the propeller shaft.'-The conveners are of the G E C 'synchdrive' design.'"* The synchdrive is used to provide variable frequency control of the synchronous motors between 0 and 60 Hz. For cruising the K S Smith received his BSc(Eng) and PhD degrees from the University of Aberdeen in 1988 and 1992 respectively. SinceOctot)er 1991 he hasbeen a Lecturer in Engineering at the Department of Engineering, Aberdeen University, with responsibility for the teaching of heavy electrical power engineering. His main fields of interest are the interaction between electrical machines and power electronic converters on closely coupled ac systems such a s offshore oil and gas installations and ships. He has been responsible for harmonic and power system stability measurements on a number of offshore installations in the UK sector of the North S e a . R Yacamini is currently a Senior Lecturer in the Department of Engineering at Aberdeen University. His previous experience includes 10 years a s a design engineer with English Electnc and G E C in the rectifier and high voltage dc transmission fields. During this time he made extensive use of physical simulations for controller design a n d system studies. This industrial period was followed by five years as a lecturer at UMIST where he carried out research, using frequency domain computer programs, into H V D C and reactive compensators. He took up his present post at Aberdeen in 1982 and has been involved in consultancy work for the oil industry for most of this period. The main thrust in his research has been to developtimedomain C A D analysis packages for power electronic applications. H e has published over 50 papers in this and related fields. A C Williamson obtained his B S c at the University of Bristol and then spend 13 years in industry working on the design and development of a wide range of electhcal machines with emphasis, towards the end, on large turbine generators. Since joining UMIST (where he obtained his PhD and is now a Senior Lecturer) he has been involved in the development of various powerelectrical machine combinations. Typical applications have been high s p e e d engine testing dynamometer, high speed, high power vahable speed drives and windturbine driven alternators. variable frequency conveners are not used. The convener accelerates the synchronous machines up to the same frequency as the 60 Hz ship supply and then synchronises its output with the supply. The synchronous motor is then transferred from the convener to the ac system busbars. In Northern waters around Canada, Finland, and the CIS, icebreakers are required to keep waterways ice free during the 23 K S Smith, R Yacamini & A C Williamson winter months. A number of icebreakers have now been built or fitted with electrical proI. » 1 BMW diesel generators pulsion employing cycloconverters. These AfI switchboard Forward switchboard drives give good performance a l the high 600V 50Hz power ratings and low speeds which are required for this application. Twelve pulse cycloconverters have recently been installed 2 . V65MW 2 .9C5kW on the cruise liners Fantasy (US) and Crystal Phase controlled motor generator thyristor rectifiers Harmony (Japan) rated at 2 x 14 M W and 2 x sets 12 M W respectively. This is approaching the power levels required for the propulsion of ifcOV 60Hz ttOVÔOHz the next generation of antisubmarine warfare frigates. Ships supplies Royce Ships supplies G! 2»Rolls The Royal Navy, in considering the deSpey gas turbirws sign of futiffe warships, must take into consideration the required military role of warships in a changing political climate, and note the practices being used by the merchant fleet 2 k V5MW, 750V OC in the design of ships, as these may be aR)lipropulsion motors cable to warship design. Frigates, which will 2 « Fixed pitch propellers form üie backbone of the fleet into the 21st century, have two major roles. They are required to police international waters and Fig 1 : CODLAG propulsion system on the Type 23 ASW frigate maintain free access to trade routes for the merchant fleet. This is essentially a cruising exercise requiring low cost propulsion. The second and major war time activity of the 4 K 20MW generators frigate is in submarine detection. A n imporAft switchboard _Forward_sw|l chboard tant design feature is therefore to reduce the I ( vibration caused by frigate propulsion sysI tems, as this is transferred to the hull of the ship and becomes water bcMiie noise which can be delected by submarines at a consid2x 1MW 2 x U 2MW motor generator erable distance. Frigates should ideally crecycloconverters sets ate as little water borne noise as possible. This reduces the possibility of detection and 2 X 1DMW ttOV 60Hz fctOV 60Hz makes it easier in turn to detect the noise propulsion motors generated by enemy vessels. One possible Ships supplies Ships supplies method of achieving this, which is currently being considered, is to adopt electrical propulsion. This will also reduce running costs Fig 2: General arrangement of an 18 MW electrical propulsion system and, itisbeUeved, would significantly reduce the noise signature of vessels.' The reduction in space required ïot modem electrical machines and advances in power electronic converter technology makes electrical propulsion systems a real possibility. Previous designs of machines and power converters were considered too bulky for naval use.* It is widely accepted that on the QE2, the onboard noise levels were reduced following the installation of the Fig 3: Cycloconverter output waveshape composed of electrical propulsion system. This is on a ship where the segments of the line to line voltage on Input side original steam turbine drives were considered to be quiet. It should be remembered, however, that the converters used on the QE2 only operate during docking and slow cruising, and that audible noise as experienced by the passengers of a cruise liner is different from the water bcäue noise generated by the ships propellers. © 0 Another majcM" advantage which electrical propulsion offers ship designers is the relative freedom with which the prime movers and power electronic converters can be located within the ship structure. The electrical cables linking the generator switchboard, converter, and propulsion motor are more flexible than the mechanical shafting required for gas turbine propulsion systems which require linear alignment. The first of the R N frigates partially to employ electrical propulsion was the Type 23 Anti-Submarine Warfare (ASW) frigate using the combined diesel electrical and gas (CODL AG) 24 © © 0 Fig 4: General arrangement of two Graetz bridges to form a single phase cycloconverter Trans IMarE. Vol 105. Part 1. pp 23-52 0-5 Ratio of Output frequency to Input frequency Fig 5: Chart showing the harmonic frequencies present in the output voltage of a single phase cycloconverter sets have been employed lo isolate the propulsion and ships service busbars. This is unlike offshore practice where the distortion on the low voltage system is tolerated and only filtered out if it is found to be absolutely necessary. The dc motors in the C O D L A G system operate under quiet cruise conditions for towed array sonar operation. The output of the converter CŒitains harmonics at six, and twelve times the supply frequency, superimposed on the mean dc output level. Additional filtering is connected between the converter and the motor to prevent harmonic fluxes in the motor generating noise on the propeller shaft system. When fast acceleration is required the gas turbines are used to provide ' sprint' power. The system load at sea varies from about 1.5 M W at low speeds to about 4.5 M W when motors are at foil power. The motors continue to operate at full power when the gas turbines are in use. The harbour load is 0.4 M W . The Royal Navy is now considering using full electrical iM^opuIsion for the next genera- Fig 6: Primitive model of a three phase induction motor Fig 7: Primitive model of a three phase, salient pole synchronous machine system, illustrated in Fig 1.' Generation is at 600V with two 1.65 M W thyristor rectifiers coupled to the generator busbars. These supply 1.5 M W dc motors which are directly coupled to the propeller shaft system. When the rectifiers are in operation severe commutation notching will occur on the generator busbar, producing waveshapes similar to those observed on offshore drilling rigs.* ' If the 440V ship service systems were supplied through step-down transformers from the600W busbar, these commutation disturbances would be reflected through the üansformers and appear throughout the low voltage distribution system. This distortion would exceed the levels allowed in naval systems. To overcome this problem rotary converter tion of frigates.'". A 5000t A S W frigate has to achieve a maximum speed of 30 kn using twin shafts. This requires 18 M W per shaft at 200 rev/min. The design of large dc motors is practically limited to around 8 M W , and tandem designs using three or more dc machines are not considered feasible as the size of the overall propulsion motCH" becomes excessive. It wUl therefore be necessary to use ac machines to reach the required power level in conjunction with variable frequency systems such as cycloconverters. Two alternatives are available for the ac propulsion motors: either induction (asynchronous) or synchronous machines. Induction machines are considered to be more rugged and robust, and require less maintenance, as 25 K s Smith. R Yacamini & A C Williamson unlike the synchronous machine there are no electrical terminals to the rotor requiring the use of slip rings. The air gap of an induction motor is generally smaller than that of a synchronous machine which will make it less able to withstand shock. Induction machines are also generally cheaper than synchronous machines. The discussion is therefore ongoing and manufacturers are now looking at the design of induction machines with much larger air gaps, which will give the induction motor the same robusmess as the synchronous machine for this application. The general arrangement of a possible full 18 M W electrical propulsion system is shown in Fig 2. THE CYCLOCONVERTER The cycloconverter is a power electronic circuit which converts an ac input to an ac output at a lower frequency. Unlike many of the converters commonly in use today, such as the 'synchdrive' mentioned above, this is achieved without using an intermediate dc link, ie the cycloconverter is a single stage converter. The cycloconverter is not a new power electronic circuit The principles of this converter are described in one of the earliest texts on power electronics." Considerable development work took place in Germany in the 1930s where cycloconverters were used for unction applications. At this time the principles of grid control, to give a variable output Fig 8: General arrangement of a three phase, six pulse, voltage and frequenc were mastered.'^ The cycloconverter cycloconverter Induction motor drive generally found an)lication where low frequency, high power ac drives are required. Examples include tube mills,'^ '* and railway traction." More recently the converter has (A) been applied successfully to a mine winder,'* the advances in control techniques giving a 1.3k performance comparable with that of a dc I.ZJk drive." UThe basic building block used in the con710struction of the cycloconverter is the six pulse / il ! 3UU — J\ A 1' Graetz bridge. A description of the operation 1 > asoof this converter can be found in standard texts on power electronic circuits.'*" The 0r ' Graetz bridge is normally used either as a hi 1 1 Ml 1 .33U 1 rectifier or as an inverter, converting an ac 1 V " 1' 1 1 li Kl ' \ input to a dc output or a dc input to an ac • Siltl A < 1 1 mi / T pMili / .710 output. This is controlled by changing the 1 il" / firing instants of the thyristors within the .|k ' f' 'i «/ bridge, relative to the three phase supply at W '-Hi; W •1.21k the bridge ac terminals. It is possible to pro-I.Ik duce a low frequency ac output from what V V .|.7)k , would normally be considered as the dc ter1 0 2ku 4in ai» Hkii IWki 12(kii 14JIII lOlin IHIIiii 3IUII Ks) minals of the bridge by continually changing the fuing delay angles of the thyristors. The Fig 9: Simulated currents for the system in Fig 8; solid line Is the output ac voltage waveshape is then cranmotoring condition, dashed is the generating condition posed of segments of the input ac line to line voltage as shown in Fig 3. It is possible with the cycloconverter to control independentiy both the frequency and amplitude of the output voltage. A bridge A switched o f f This can be achieved by blocking the single six pulse bridge cannot supply both positive and negaüiyristOT gate signals to each bridge at the appropriate time. tive half cycles of the output current which necessitates tiie use This is not a trivial problem; determining the correa instant lo of a second bridge. This leads to the basic six pulse single phase transfer load current between bridges has been a m^or problem cycloconverter configuration shown in Fig 4. For this circuit to for cycloconverter designers. operate successfully, it is necessary to allow only one bridge to Three phase cycloconverters are formed by combining conduct at any time. If üie desired load current is positive, tiiree individual single jAase units. A number of three jAase bridge A should be conducting and bridge B switched o f f COTinections are possible which have differing harmonic efNegative load current should be supplied by bridge B , witii fects upon both the input and output sides of the converter. \ V n ' \ iP ! 26 \ . Trans IMarE. Vol 105. Part 1, pp 23-52 modulation process present within the cycloconverter givesriseto con^)onents which are neither integer multiples of the input or the output frequencies. The frequency content of tiie output voltage waveshape of a cycloconverter is therefore extremely complex, containing a wide range of frequencies. These can be conveniently represented using tiie chart shown in Fig 5. The amplitudes of these harmonic frequencies are a fonction of the output voltage ratio (the depth of modulation control signal to the converter) and the load power factor. Each Une in Fig 5 has associated with it a fixed amplitude, regardless of the frequency ratio. When the output frequency of Üie drive is zero, the harmonics present in the output voltage waveshape correspond to integer multiples of the converter pulse number and 2Uii Hkii ICiOin lakii aKkn l(s) tiie supply frequency. For example, a six pulse bridge will contain harmonics of order FIg 10: Cycloconverter output voltage (motoring condition) 6, 12, 18, etc. This corresponds to the left hand part of Fig 5, where the families of harmonics present in tiie cycloconverter originate. When the output frequency of the cycloconverter is increased, the second term in equation (1) becomes non-zero producing frequencies above and below the harmonic parents. This is clearly shown in Fig 5. It should be noted from this chart that at several distinct values of output frequency, the components given by equations ( 1 ) and (2) will be of very low or even zero frequency, ie dc components of voltage. Alternatively tiiey can produce frequencies very close to, or exacüy equal to the desired output frequency. Under tiiese operating conditions, the dc component could possibly saturate the propulsion motor, or the components close to the wanted output frequency may produce a pronounced beating effect of the motor current and voltage waveshapes. It is possible to derive equations which give the frequency content of the current supplied to the 16Un cycloconverter from the ac system. As for the output voltage waveshape this is also found to Fig 11 : Cycloconverter output voltage (generating condition) contain a very wide range of frequencies. When considering using a cycloconverter for a marine electrical propulsion drive a number of questions Twelve pulse cycloconverter circuits can be formed using four, must be addressed, including: What effect w i l l the three phase six pulse bridges. cycloconverter have on the synchronous machines supplying For the six pulse, single phase cycloconverter, supplied the ac system? How severe will die ac voltage distortion be. and from an infinite busbar, the major harmonics present in the what oscillating torques will be induced on the shaft of the output voltage are given by:^° prime mover? Is tiiere a possibility of the oscillating torque exciting a torsional resonance of tiie propulsion motw-propelf, = 6 p f . ± ( 2 n + l ) f „ (1) ler shaft system? Is one cycloconverter configuration better tiian another for a particular appücation and how is this For the twelve pulse converter the harmonics are: assessed? How will the system perform under abnormal conditions such as semiconductor device failure or unbalanced f,= 12pf,±(2n+ l)f„ (2) supply conditions? where f, is the input frequency, f« the output frequency and f^ Üie harmonic frequencies. From these equations it is clear that the output voltage of the cycloconverter contains a large number of harmonic terms. The frequencies given by the relationships in equations ( 1 ) and (2) are not in a mathematical sense true harmonics, as the Questions such as tiiese can be answered by developing computer models to simulate complete electrical propulsion systems and assess the relative advantages and disadvantages of different converter configurations. For tiiis application C A D software has been used, which allows different cycloconverter topologies to be readily assembled and simulated. 27 K S Smith. R Yacamini & A C Williamson POWER ELECTRONIC SIMULATION There is a wide range of computer simulation packages available on the software market which can be ^ p l i e d to the analysis of power electronic and other similar systems. In developing a computer model of a complete marine propulsion system it is necessary to describe electrical machines and power electronic converters, to be able to change the arrangement of the components, implement different control strategies, and also smdy the effects of abnormal opo^tion on the p o f ormance of the converter. One such package is the Saba* simulator, which is marketed as a mixed analogue and digital simulation package and has been found particularly useful for this type of application. The simulation process using this package can be divided into three distinct stages, involving the input of the system data, performing the system analysis, and finally processing and viewing the results of the analysis (post-processing). The input to the simulator consists of a file describing the system to be analysed. This netUst file can contain references to otiier netlists and components which allows full hierarchical systems to be convenientiy analysed. This feature is particularly useful in developing models of large complex systems, where each subcomponent can be tested individually before being inccMporated into the larger system. When the nethst describing the system is complete, the simulator is invoked and the netlist information read and checked for errcx's. If no errors are found the simulation can proceed. The numerical integration techniques used by this simulator include first and second csder methods such as Backward-EulerandTr^zoidal methods. The post-processor allows the results of the analysis to be viewed on the computer screen. This takes the form of waveshapes which is convenient as itmaintainsanengineaing 'feel' forthesystem being analysed. Frequency domain spectra can also be di^layed if required. IN r.| 6uk ..1 4 Ilk - ;uk - -lOk -4Uk - .40k — .•Ok - Ô 3kii (Û11 Util lUkii lÀn l-Âu ItiÙiii ilùii Win is the generating condition TUMn Us) Fig 13: Currents drawn by Individual cycioconvsrters The unique feature of tiie simulator, which makes it particularly useful for the simulation of marine cycloconverter prcpulsion systems, is the ability to define the characteristics of new components. This is achieved using the Mast modelling language.'' Mast is very similar to the programming languages C and Fortran and is used to describe the matiiematical relationships governing the operational characteristics of the desired component. The characteristics of any system can then be programmed, removing the restrictions present in the type of simulators which onlyrepresentelectrical networks. Mechanical quantities can be directly represented in their own units and not by electrical analogues. The connection points to the templates which describe components can be either electrical (x mechanical nodes. Although the simulator is marketed as an electrical simulate^- this facility allows it to be used as a mechanical simulator, and where electrical machines are used as an electro-mechanical simulator. 28 4I11 The programming capabilities of Mast have therefore been used to develop models which describe the operation of induction and synchronous machines, as well as the controller present in the cycloconverter. Electrical machines have been modelled by programming the voltage balance equations of the phase models of the machines in combination with the relationships describing the dynamics of the rotor system. In the case of the induction machine, the representation employed in the simulations is shown in Fig 6.^ The machine consists of three stator windings (sa, sb, and sc) and three rotor windings (ra, rb, and rc). The parameters for tiiis model can be calculated from the fundamental equivalent circuit of the machine. The primitive model of the synchronous machine implemented is shown in Fig 7. In this representation four windings are shown on the rotor of the synchronous machine, two in the direct and two in the quadrature axis." This allows the effects of synchronous Trans IMarE. Vol 105. Pan 1. pp 23-52 of the stator currents to give torque control, can also be implemented. The basic elements of a ship electrical propulsion system have been modelled; these can be used to study the different systems proposed. POSSIBLE CYCLOCONVERTER PROPULSION SYSTEMS lOka llUii Fig 14: Total current drawn from the ac system aWkii t(s) Full electrical propulsion of a 5000t A S W frigate will require two drives each rated at 18 M W . Present Royal Navy frigate experience of electric drives is limited to the 1.6 M W dc drives used in the C O D L A G system. It is unhkely üiat tiie R N will adopt tiie 18 M W drive in the fu^t instance; instead it is felt tiiat a 4 M W drive, combined wifli existing gas turbine technology in an arrangement similar to the C O D L A G system, is likely to be used as a further step towards tiie 18 M W drive. This smaller drive will itself intixxluce a number of new design problems to frigate electrical engineers. These design questions are related to the topology o f the cycloconverter and changes to the power system itself will be required. In order to reach the power levels required for electrical propulsion it will be necessary to increase the voltage of the prime mover generators from 440/660V to 3.3 k V , 6.6 k V o r 11 kV. Switchgearratedattiiesevoltage levels is readily available from manufacturers, and has found extensive use on offshore oil and gas recovery platforms. It has also been suggested that it may be advantageous to raise tiie frequency of generation frcan 60-90 Hz. In designing the 4 M W drive a number of different options have to be considered. What should be the pulse number o f the cycloconverter? Should a tiiree or four wire connection to die propulsion motor be used? Fig 15: Motor current waveshape showing the effect of changing from a Should an induction or a synchronous mafour wire to a three wire connection; dashed is four wire, solid Is three chine be used? Are some transformer conwire nections preferable to others? In attempting to answer these questions a large number of machine saliency to be included in the simulation if necessary. different converter configurations must be considered. The Phase models of electrical machines have been found to be computer modelling of the cycloconverter drive allows die more useful than the more widely used two axis models.^ relative merits of different converter topologies to be considUsing the Saber simulator, it was found that the phase model ered and the performance of the complete electrical propulsion was more robust and gave shorter run times in many cases. In system as a single unit to be analysed. addition using the phase model it is possible to consider the effects of unbalances within the machine and to show how these would affect the overall performance of the cycloconverter SIMULATION RESULTS drive. In the contfoller implemented in tiie work described in tiiis paper an open loop strategy has been employed, based upon die Cycloconverters and motors well known inverse cosine control method. The logic required In order to illusfrate the characteristics of different conto implement this controller, determining tiie thyristor firing verter topologies, and also demonstrate tiie usefulness of time instants as weU as die converter group blocking and deblocking domain simulation as a design tool, waveshapes obtained from signals, has been written into the simulation using the Mast tiie simulation for a number of different cycloconverters are modelling language. Otiier control sfrategies, such as control presented and discussed. 29 K s Smith. R Yacamini & A C Williamson The basic building block used in the design of large cycloconverter drives, as mentioned above, is the six pulse bridge. Two back to back bridges combined give a cycloconverter with a single phase output and six bridges can be used to give a three phase output. Transformers are required at die input to each phase of die cycloconverter to provide electrical isolation between phases, if the motor phases are not isolated. One of the propulsion options being considered is to use a star connected induction motor. A possible configuration f o r a 4 M W six pulse drive is therefore shown in Fig 8. A switch, S W l , is shown in the neutral wire to the motor. For the first set of simulation results this switch is left closed, giving a four wire motor connection. It will be shown later that if S W l is opened to turn the drive into a three wire connection then changes in the waveshapes associated with the drive will occur. The parameters of the system analysed have not been chosen to model an existing or planned cycloconverter installation. The purpose of displaying the waveshapes is to emphasise the possibilities for analysis offered by the simulation. In the cases presented for example, the system supply frequency is 50 Hz, rather than the more normal 60 Hz for a ship, or the 90 Hz suggested for future installations. The supply frequency, and indeed all the other parameters of the system including the machine ratings and reactances, can be conveniently changed to allow the study of other systems. More importantly from the designer's point of view, this allows direct comparisons between systems to be made. The waveshapes obtained from this simulation are shown in Figs 8-14. Two conditions are illustrated corresponding to motoring and generating action in Uie motor with the output frequency of the drive at 15 Hz. Fortiiemotoring condition üie mechanical power supplied by the motor is 2.4 M W whilst for the generating condition üie power is 4.2 M W . The waveshapes for each of tiiese conditions are shown together on the same graph to allow direct comparisons in the time domain to be made between them. Figure 9 shows the stator currents for the two cases studied. There is clearly a phase difference present between these waveshapes. This is required to achieve the reversal of power flow, as no phase difference is present in the motor line to neutral voltage as shown in Figs 10 and 11. This is to be expected as the modulating function to the cycloconverter, which determines the reference signal to the inverse cosine controller has not changed. The change in load is achieved by changing tiie polarity of tiie load torque. It should be noted from close inspection of Figs 10 and 11 that with the motor generating, the bridges witiiin the cycloconverter spend considerably more time in inversion than in rectification mode. This is to be expected since power flow is now from the motor shaft back tiirough tiie converter to üie ac power system. The torque waveshape on üie shaft oftiiepropulsion motor is shown in Fig 13. The reversal in tiie sign of torque required to produce a reversal in power flow is clear. The currents drawn by the individual cycloconverters are shown in Fig 13. Inspection of tiiese waveshapes shows that there are, in fact, a wide range of frequencies present, including frequencies below the supply frequency. The effect of Üie 120 deg phase displacement between each of tiie single phase cycloconverters is to reduce the number of frequencies present in the current waveshape drawn from the supply system. This waveshape is shown in Fig 14. Although tiie harmonic content is reduced tiiere is still a wide range of frequencies present, including non-integer and sub-harmonic terms. The results of passing the lime domain data, present in tiie waveshapes presented above, through a Fourier analysis programme with a resolution of 5 Hz, produced tiie frequencies 30 Rg 16: Primitive model of an induction motor with a double wound stator Fig 17: General arrangement of a six pulse cycloconverter supplying a double wound stator induction motor shown in Tables I and II. for the motoring and generating conditions respectively. The corresponding harmonic spectra are shown in Appendix 1 for the motoring condition. Inspection of Tables I and II shows tiiat the harmonics present in the output voltage waveshape of üie cycloconverter agree with those which would be obtained using equations (1 ) and (2), which are based on a frequency domain calculation. The simulations do have the advantage of showing actual waveshapes which the previous analysis does not show. The waveshapes obtained from the simulation also include üie effects of commutation overlap and the iniercoupling that is present between tiie cyclcoconvertcrs. These effects are ignored in the frequency domain calculation. The wanted output frequency is 15 Hz, and sideband frequencies appear at odd Trans IMarE. Vol 105. Pan 1. pp 23-52 I 3tn I •Uli I 6Uii I Hkn I lUkii I I3in 1 Mn I Idlkn I ItUn 3Xkii ih Fig 18: General arrangement of a six pulse cycloconverter supplying a double wound stator induction motor ^ Jkn i -Uli I 6an ! 8Uii 1 lUkn 1 Okn 1 l*ka 1 IfiOra 1 IWm 2UUn t(s) Fig 19: Voltage across one stator phase of the double wound motor multiples of 15 Hz above and below the supply frequency times the converter pulse number, ie 300 Hz. Sidebands also appear cenffed around600 Hz. For example, sidebands cenu-ed around 300 Hz appear at 285 Hz and 315 Hz, and also at 255 Hz and 345 Hz. Harmonic currents flow in the stator winding of die machine at tiie frequencies present in the output voltage w a v e s h ^ . The 15 Hz wanted component produces the mean dc component of tOTque whilst the harmcMiic currents give rise to harmonic components of torque. The largest harmonic torque is at 300 Hz. This harmonic torque is independent of the output frequency of the cycloconverter drive. The frequency of this component is determined by tiie frequency of tiie supply system and the converter pulse number, and does, in fact. appear at the same frequency at which the sidebands present in the cycloconverter output voltage are centred, ie at the product of six timestiiesupply frequency andtiieccmverter pulse number. This torque is produced by the action of the pairs of sideband current harmonics centred around this frequency. The wanted component of current at 15 Hz produces a flux wave in the air gap of the propulsion motor which is of positive phase sequence. The first pair of sidebands occur at 285 Hz and 315 Hz. The current at 285 Hz produces a flux wave which rotates in the same direction as the 15 Hz flux wave, ie the 285 Hz flux wave is also of positive phase sequence. The flux associated with the component at 315 Hz will rotate intiieopposite direction to that at 285 Hz and is therefore of negative phase sequence. The contribution of these components to the mean value of torque is negligible. They do however combine to produce a harmonic pulsating torque at 300 Hz. A similar argument can be applied to tiie otiier sidebands centred around 300 Hz. A harmonic torque also spears at 600 Hz, which is produced by the sideband current harmonics centred around this frequency. The currents drawn by the individual cycloconverters are very rich in harmonics. Inspection of Tables I and II reveals some large components, eg 20 Hz of 45.46 % and 80 Hz of 25.84 % for the motoring condition. Due to the transformer ccmnections employed and the 120 deg phase shift present at tiie output of each individual cycloccaiverter, many of these harmonics have their amplitude significantiy reduced intiiecurrent drawn from the ac supply system. For example, die COTiponents at 20 Hz and 80 Hz discussed above are reduced to 1.03 % and 0.72 % in die supply current. It should be noted that tiie harmonics generally associated with a six pulse converter, ie 5th, 7tii, U t h , 13tii, etc, are present in both the ac system supply current and the current to the individual cycloconverters. The amplitude of these COTiponents on the ac supply side and on die cycloconverter input are tiie same. No significant attenuation of these components occurs. If the switch S W l in Fig 8 is opened, this has an effect on die current supplied to the propulsion motor. The current waveshapes obtained for the motoring condition studied previously witii the switch closed, and tiiat witii the switch open, are shown togetiier in Fig 15. There is clearly a significant difference between the two w a v e s h ^ s . This effect is due to the output voltage of tiie tiiree phase cycloconverter containing frequencies which are of zero phase sequence. Currents at tiiese frequencies can only flow if tiiere is a neutral connection to die motor. These components are tiierefore present in tiie earlier waveshapes where the switch S W1 is closed but are absent when it is opened. It should be noted, however, tiiat die presence or absence of zero sequence currents in tiie supply to tiie motor does not affect tiie elecfromagnetic torque produced bytiiemachine. This is due to die zero sequence components producing no air gap mmf. 31 K S Smith. R Yacamini & A C Williamson Zero sequence currents can therefore only flow in the stator winding (if a neutral wire is present), and consequentiy the impedance of the induction motor is smallor to components of zero phase sequence as the rotor is not included in the zero sequence equivalent circuit"-^ A n alternative design for the propulsion motor would be to use an induction machine witii a double wound statw winding. The primitive phase model of such a motor is shown in Fig 16, in which the two sets of stator windings are displaced by 30 deg electrical, and each phase winding is supplied by a single phase cycloconverter. The 30 deg phase shift present in the stator windings has the effect of cancelling a large number of the harmonic mmfs present in the air g ^ of the machine, and so reduces the ripple present in I i i I i I ! 6Un Kin UKkn 13Jm \43a\ 160m I8O111 ZXkn l(s) 3lh 4kii the induced machine electromagnetic torque. This can be illusu-ated by the results obtained Fig 20: Machine torque for double wound motor from the computer simulation ior the system shown in Fig 17, which features such a motor. The current supplied to the motor is shown in Fig 18. This waveshape is not signiflcantiy different from that drawn by the three phase, three winding motor shown in Fig 9. The motor line to neutral voltage is shown in Fig 19. This waveshape is seen to contain more high frequency terms due to the increased number of converters supplied from a common busbar. The effects of commutations taking place in other converters are present from the connection at the point of common coupling. Intarcoupling effects between the six stator windings of the machine are also present The most significant diffCTcnce between the three phase machine and the six phase machine lies in the torque shown in Fig 20. Comparison of this waveshape with Fig 12 clearly shows that the ripple component is reduced. (Note that the dc value between the two cases is different.) The cycloconverter —! 1 1 1 i 1 1 I I 3Jni -till fiUn Bkn Ittkn 12Un 14Cini 160m IHm aXkn t(s) input current shown in Fig 21 is broadly similar to that of Fig 13 as would be expected. Fig 21: Current drawn by an Individual cycloconverter The current drawn from the ac system is shown in Fig 22 which displays less harmonic distortion than die earlier case of F i g 14. content of the torque is reduced, as are the harmonics present These general observations on the time domain waveshapes in tiie current waveshape from the ac system. are confirmed by inspection of tiie results obtained from Fourier analysis shown in Table III. Ships generators The frequencies present in tiie motor voltage and the motor In the earlier sections the effects of the cycloconverter upon current for the double wound machine and the earlier tiiree the propulsion motor, and especially tiie torque produced by phase machine are almost identical in both frequency and the motor, have been considered in detail. Large cycloconverter amplitude. The current drawn from the ac system contains less drives, if applied to ships, will have a significant effect upon tiie hamionics tiian tiie earlier case; the side bands previously synchronous generators which supply the system. The centred around tiie 50 Hz fundamental are significantiy reduced. It should be noted that the harmonics characteristic of cycloconverter propulsion system on a frigate would, in fact, a six pulse drive, ie the 5th and 7th, are not eliminated as they be tiie largest single load on tiie frigates electrical system. A n would be with a twelve pulse converter. They are, however, understanding of the possible detrimental effects of the drive reduced using the double wound stator. on the alternators is therefore required. The characteristics of non-linear loads, such as diode and From the waveshapes and tables of harmcmics presented, it thyristor bridges, on die performance of synchronous mais apparent tiiat the cycloconverter drive with a double wound chines has been studied previously These works show that stator on the propulsion motor offers a number of advantages for wound rotor machines which do not display the effect of over a three phase machine. Most significantiy tiie harmonic 32 Trans IMarE. Vol 105. Pan 1. pp 23-52 akii lakii IHkii 3irkii l(s) Fig 22: Total current drawn from the ac system (N .,1 12.3k . ent system components, are included automatically in tiie time domain simulation. Figure 23 shows tiie torque on tiie shaft of tiie alternator when the cycloconverter is running. This waveshape is seen to be rich in harmonics, which are tabulated m Table IV and also shown graphically in Appendix I. The most significant of these is at 300 Hz (30 %), produced by tiie action of the 5ti3 and 7ti3 harmonic currents, drawn by the cycloconverter, which flow in the alternator stator. It should also be noted tiiat tiiere are a number of lower frequencies present in this waveshape some of which have relatively large amplitudes, forexamplea 15 Hzcomponent of 3.8 % and 30 Hz of 6.8 %. There is clearly a possibiüty of these or other such low frequencies exciting a resonance of the prime mover-altemator rotor system, careful consideration of which will be required. The current flowing in die stator windings is shown in Fig 24. It should be noted that the amphtudes of the frequencies present in this waveshape are not significantiy different from those obtained using tiie earher simpler representation of the supply system by an emf behind a fixed inductance. If Üüie simulation had been perfOTmed with a synchronous machine displaying more significant sahency this would not have been the case. The simulation results given in this section have shown tiiat tiie Saber time domain simulation package has been developed to model a range of cycloconverters and electrical machines, and used to perform system calculations for the complete electrical propulsion system as one unit, not as separate parts. The interaction between the constituent parts of the system is therefore maintained in tiie simulation. akii 2U(kn l(s) CONCLUSIONS Fig 23: Torque on the shaft of an aKemator supplying a three phase, six pulse 4 MW cycloconverter drive sahency, tiie machine can be represented by an emf behind a fixed reactance. For a machine which does display saliency, tiie ccHnmutating reactance is no longer constant and a more ccMnplex machine model is required. The foil phase model discussed in the section on power electmoic simulation is therefore used in the ship propulsion simulations to represent the alternator supplying the cycloconverter drive. T o demonstrate the complete simulation o f the cycloconverter propulsion system, \he synchronous machine model was combined with tiie earher six pulse cycloconverter supplying a three (^ase three winding induction motor. The signals which this simulation provides, in addition, to those obtained eariier, are the electromagnetic torque on tiie shaft of the alternator, the currents flowing in the damper windings and also the terminal voltage of the machine. With tiiis simulation the effects of tiie time varying commutating reactance of tiie alternator and all the intercoupling present between the differ- In tiiis paper, tiie operating principles of the cycloconverter have been discussed. This converter spears at tiie moment to be tiie most hkely candidate for use in the next generation of Royal Navy frigates, should tiie Navy adopt ac elecuical propulsion. It has been demonstrated that it is now possible to analyse the complete electrical propulsion system, including die synchronous generators, cycloconverters and tiie propulsion motors which may be found on future marine installations. As a design tool this computer simulation is extremely useful. The way in which the parameters of one system building block affecttiieperformance of another can be readily calculated. The computer simulation is not restricted to any one particular converter configuration and can tiierefore be used to assess die advantages and disadvantages of different electrical propulsion system arrangement. The computer models have been achieved using tiie Saber simulator, tiius taking advantage of a modem computer aided engineering software package. Of particular importance is tiie Mast modelling language, which can be used to describe die operating characteristics of new components which may not be 33 K S Smith. R Yacamini tSc A C Williamson available in the system library. This has been used particularly in tiiis work to define die characteristics of electrical machines which supply and fed from the power electronic converters. The hierarchical nature of tiie simulation package allows the different components of the system to be developed separately and tiien added into the complete system simulation. This also allows the system topology to be readily changed. The ability to simulate mixed electro-mechanical systems has been emphasised. This ability is being extended at Aberdeen University into the analysis of motor noise generation and the behaviour of shaft systems and propellers. The results obtained from a number of different system studies have been used to illustrate the usefulness of this metiiod for analysing the performance characteristics of different cycloconverter systems. Fig 24: Current flowing in the motor windings of an alternator connected to a 4 MW cycloconverter drive ACKNOWLEDGEMENTS The authors would like to acknowledge the Science and Engineering Research Council for providing funding and M r D Bain. Reprographics Section, Department of Engineering. Aberdeen University, for preparing illustrations. REFERENCES 1. J B Borman. T h e electrical propulsion system of tiie QE2: some aspects of the design and development'. G E C Publication No 3493-353. 2. P Bloom, 'QE2 goes diesel electric'. Modern Power Systems (USA). Vol 6, No 9, pp 19-23 (1986). 3. D Finney, 'The synchdrive - a synchronous motor variable speed drive system', GEC Journal for Industry. Vol 5, No 3, pp 108-114 (October 1981). 4. D Finney. 'Synchdrive converters for high voltage motors', GEC Journal for Industry, Vol 7, No 1, pp 2530 (February 1983). 5. W J Levedhal, "Integrated ship machinery systems re\/isiied\NavalEngitu;ersJournal,pp93-0\ (May 1989) 6. J V JoUiff and D LGreene, 'Advanced integrated electric propulsion: a reality of tiie eighties'. Naval Engineers Journal, pp 232 - 254 (April 1982). 7. E J Greer, 'Electrical power engineering in modem surface warships'. GEC Review, Vol 2, No 3, pp 151157(1986). 8. K S Smitii and R Yacamini, 'Commutation voltage spikes on isolated offshore power systems', Proc 24th Universities Power Engineering Conference, Belfast, pp 417-420 (September 1989). 9. R Yacamini, L Hu and R Fallaize, 'Calculation of commutation spikes and harmonics on offshore platfomis'. lEE Proc, Vol 137, Pt B , No 1 (January 1990). 10. P TNonon and M Murphy, 'Realising the potential - f u l l electric propulsion of surface warships'. RINA International Symposium on the Future of Surface Warships, London (June 1990). 34 11. H Rissik, The Fundamental Theory of Arc Converters, Chapman and Hall Ltd (1939). 12. R Feinberg, 'Frequency changing using mercury arc mutators', y / f f , pp 531-543 (1939). 13. E Blauenstein. 'The first gearless tube mill', BrownBoveri Review, Vol 3, pp 96-105 (1970). 14. J Langer. 'Static frequency changer supply system for synchronous motors driving tube mills'. Brown-Boveri Review, Vol 3, pp 112-119 (1970). 15. W Faust, 'Static frequency changers for 16 2/3 c/s railway networks' Brown-Boveri Review, pp 519-525 (August 1964). 16. D G Taylor, 'Squirrel cage induction motor cycloconverter drive at Wearmouth Colliery ' A ' pit friction winder. Part 1. Development of drive system for winder application', Afm/>2^ Technology,pp 4-9 Oanuary 1988). 17. D M Cross, 'Squirrel cage induction motor cyclo-convertcr drive at Wearmouth Colliery ' A ' pit friction winder. Part 2, Drive control and regulating system'. Mining Technology, pp 11-15 (January 1988). 18. C W Lander. Power Electronics, McGraw-Hill (1987). 19. J G Kassakain, M F Schlecht and G C Verghese, Principles of Power Electronics, Addison-Wesley (1991). 20. B R Pelly. Thyristor Phase Controlled Converters and Cycloconverters, Wiley, New York (1971 ). 21. MAST reference manual (Ver 3.01), Analogy Ltd. Beaverton, Oregon, U S A (1990). 22. A K De Sarkar and G J Berg, 'Digital simulation of tiiree phase induction rc\oiors\IEEEPAS, Vol PAS 89, No 6, pp 1031-1037 (July 1970). 23. P C Krause, Analysis of Electric Machinery, McGrawHill, New York (1986). 24. B Adkins and R G Harley, The General Theory of Alternating Current Machines, Chapman and Hall (1975). 25. A C Williamson, 'The effects of system harmonics upon machines'. International Conference on Harmonics in Power Systems, UMIST, Manchester, England (1981). 26. G C Jain, 'The effect of voltage waveshape on tiie performance of a three phase induction motor', IEEE Trans IMarE. Vol 105. Pan 1. pp 23-52 27. 28. 29. 30. Trans Power Apparatus arui Systems, Vol PAS-84, pp 561-566. W J Bon wiek and V H Jones, 'Performance of a synchronous generatOT witia a bridge rectifier', Proc lEE, V o l 133, Pt C, No 6, pp 1338-1342 (September 1972). W J Bon wiek and V H Jones, 'Rectifier loaded synchronous generatOTS with damper windings', Proc lEE, Vol 120, No 6, pp 659-666 (June 1973). W J Bonwick 'Voltage waveform distortion in synchronous generators witii rectifier loading', Proc lEE, Vol 127, No 1, pp 13-19 (January 1980). S Moriyasu and C Uenosono, ' A n analysis of tiie charactOTstics of a synchronous machine connected to a dc link', Archiv für Electrotechnic, Vol 69, pp 111-120 (1986). APPENDIX 1 FREQUENCY DOMAIN SPECTRA The following section contains tiie frequency domain spectra (Figs 25-32) for some of the waveshapes presented in the section cm simulation results. The spectra have been presented to supplement die tables of harmonics (Tables I-IV) referenced in the text APPENDIX 2 SIMULATION VERIFICATION At the time of writing this paper tiie information available in tiie public domain regarding tiie types of cycloconverter systems and the parameters for the drives proposed for warship propulsion is extremely limited. As no cycloconverter propulsion for frigates currentiy exists, verification of the computer model against a real system is not possible. As the simulator has a modular approach to developing submodels of the various components which comprise the cycloconverter drive, it has been possible to test each of these individually and compare the performance of tiie submodel with the results of tests published previously for these items. This approach has been used to verify the accuracy of the induction and synchronous machine submodels with both sinusoidal and distorted busbar waveshapes. Assistance of a technical nature on the modelling of modem cycloconverters was received from M r Derek Taylor of C E G E L E C , Rugby. Comparisons in botii tiie time and frequency domains, between site measured and simulated waveshapes for the cycloconverter output voltage and currents, gave confidence in the ability of the simulation to model correctiy a cycloconverter. The assistance of M r Taylor in this exercise is gratefully acknowledged. 35 K s Smith. R Yacamini & A C Williamson Table I: Fourier analysis for motoring condition T a b l e No 1 255 0 .00 .00 10.73 20.40 .00 260 0 2.84 3.20 .00 .00 .00 265 0 .00 .00 .00 .00 .00 270 0 .00 .00 .00 .00 .00 Motor 275 0 .00 .00 .00 ,88 .00 Current Current Current Voltage Torque 280 0 .62 8.86 .00 .00 .00 Motoring Freq 0.0 36 Syscem .56 Cyclo Condition Motor Motor .86 .00 .00 285 0 .00 .00 6.76 29.66 .00 100.00 290 0 .00 2.53 .00 .00 .00 5.0 .00 .00 1.09 .00 2.28 295 0 .00 .00 .00 .51 .00 10.0 1.45 6.58 .00 .00 3.34 300 0 .00 .00 .00 .00 8.29 15.0 .00 .00 100.00 100.00 2.91 305 0 .00 .00 .00 .00 .00 20.0 1.03 45.46 .73 .00 3.28 310 0 .00 4 .48 .00 .00 .00 25.0 .88 .00 1.25 .00 2.09 315 0 .00 .00 4.37 20.73 .00 30.0 2.13 1.21 .64 .00 1.92 320 0 .73 4.61 .00 .00 .00 35.0 1.54 .00 1.17 .91 .82 325 0 .00 .00 .00 .00 .00 40.0 1.99 3.49 .55 .00 1.33 330 0 .00 .00 .00 .00 .00 45.0 1.31 .00 8.14 3.25 .00 335 0 .00 .00 .00 .00 .00 50.0 100.00 100.00 .53 .00 .60 340 0 1.21 1.39 .00 .00 .00 55.0 1.08 .52 .89 .54 .00 345 0 .00 .00 2.07 5.50 .00 60.0 1.12 .95 .00 .00 .00 350 0 5.10 5.22 .00 .00 .00 65.0 1.37 .00 .74 .00 .00 355 0 .00 .00 .00 .00 .00 70.0 2.69 2.93 .00 .00 .00 360 0 .00 .00 .00 .00 .00 75.0 1.14 .00 2.80 3.19 .00 365 0 .00 .00 .00 .00 .00 80.0 .72 25.86 .00 .00 .00 370 0 .00 .79 .00 .00 .00 85.0 .93 .55 .00 .67 .00 375 0 .00 .00 .00 .00 .00 90.0 .53 .66 .00 .00 .99 380 0 .00 3.67 .00 .00 .00 95.0 .69 .55 .00 .00 .00 385 0 .00 .00 .00 1.05 .00 100.0 .72 .90 .00 .00 .00 390 0 .00 .00 .00 .00 .00 105.0 .00 .52 2.15 3.10 .00 395 0 .00 .00 .00 .00 .00 110.0 .72 8.68 .00 .00 .00 400 0 .00 .67 .00 .00 .00 115.0 .00 .00 .00 .64 .00 405 0 .00 .00 .00 1.B2 .00 120.0 .00 .00 .00 .00 .00 410 0 .00 .90 .00 .00 .00 125.0 .56 .00 .00 .79 .00 415 0 .00 .00 .00 .50 .00 130.0 .68 1.90 .00 .00 .00 420 0 .00 .00 .00 .00 .00 135.0 .57 .00 2.58 2.76 .00 425 0 .00 .00 .00 .52 .00 140.0 2.76 3.27 .00 .00 .00 430 0 .00 1.70 .00 .00 .00 145.0 .00 .00 .00 .64 .00 435 0 .00 .00 .00 2.14 .00 150.0 .00 .00 .00 .00 .00 440 0 .72 .86 .00 .00 .00 155.0 .00 .00 .00 .55 .00 445 0 .00 .00 .00 .00 .00 160.0 1.83 2.31 .00 .00 .00 450 0 .00 .00 .00 .00 .00 165.0 .00 .00 1.41 2.73 .00 455 0 .00 .00 .00 .00 .00 170.0 .78 2.04 .00 .00 .00 460 0 3.22 3.24 .00 .00 .00 175.0 .00 .00 .00 .00 .00 465 0 .00 .00 1.04 3.77 .00 180.0 .00 .00 .00 .00 .00 470 0 .00 .70 .00 .00 .00 185.0 .00 .00 .00 .00 .00 475 0 .00 .00 .00 .76 .00 190.0 .52 11.31 .00 .00 .00 480 0 .00 .00 .00 .00 .00 195.0 .00 .00 1.70 4 .44 .00 485 0 .00 .00 .00 .68 .00 200.0 .00 1.91 .00 .00 .00 490 0 .00 4.83 .00 .00 .00 205.0 .00 .00 .00 .00 .00 495 0 .00 .00 .88 6.71 .00 210.0 .00 .00 .00 .00 2.53 SOD 0 .00 .80 .00 .00 .00 215.0 .00 .00 .00 .60 .00 505 0 .00 .00 .00 .00 .00 220.0 .00 23.71 .00 .00 .00 510 0 .00 .00 .00 .00 1.56 225.0 .00 .00 3.69 12.33 .00 515 0 .00 .00 .00 .60 .00 230.0 .56 1 .54 .00 .00 .00 520 0 .00 4.46 .00 .00 .00 235.0 .00 .00 .00 .00 .00 525 0 .00 .00 1.64 13.45 .00 240.0 .64 .00 .00 .00 .00 530 0 .00 1.12 .00 .00 .00 245.0 .00 .00 .00 .00 .00 535 0 .00 .00 .00 .00 .00 250.0 14.30 14.80 .00 .00 .00 540 0 .00 .00 .00 .00 .00 Trans IMarE. Vol 105. Part 1. pp 23-52 Table I: Fourier analysis for motoring condition (cont) 545.0 .00 .00 .00 .88 .00 835.0 .00 .00 .00 .72 550.0 4.20 4 .27 .00 .00 .00 840.0 . 00 .00 .00 .00 .00 555.0 .00 .00 1.38 6.18 .00 845.0 .00 .00 .00 .00 .00 560.0 1.76 1.86 .00 .00 .00 850.0 1.61 1.67 .00 .00 .00 565.0 .00 .00 .00 .67 .00 855.0 .00 .00 .57 3.61 .00 570.0 .00 .00 .00 .00 .00 860.0 1.69 1.69 .00 .00 .00 575.0 .00 .00 .00 .55 .00 865.0 .00 .00 .00 .00 .00 580.0 .00 2.92 .00 .00 .00 870.0 .00 .00 .00 .00 .00 585.0 .00 .00 .54 5.21 .00 875.0 .00 .00 .00 .92 .00 590.0 .00 2.07 .00 .00 .00 880.0 .00 1 .80 .00 .00 .00 595.0 .00 .00 .00 .00 .00 885.0 .00 .00 .00 .95 .00 .00 600.0 .00 .00 .00 .00 1.15 890.0 .00 .92 .00 .00 .00 605.0 .00 .00 .00 .62 .00 895.0 .00 .00 .00 .00 .00 .00 610.0 .00 2.11 .00 .00 .00 900.0 .00 .00 .00 .00 615.0 .00 .00 .55 5.10 .00 905.0 .00 .00 .00 .81 .00 620.0 .00 2.56 .00 .00 .00 910.0 .00 1.24 .00 .00 .00 625.0 .00 .00 .00 .00 .00 915.0 .00 .00 .00 4.03 .00 630.0 .00 .00 .00 .00 .00 920.0 .00 1.36 .00 .00 .00 .00 635.0 .00 .00 .00 .52 .00 925.0 .00 .00 .00 .96 640.0 .96 .85 .00 .00 .00 930.0 .00 .00 .00 .00 .00 645.0 .00 .00 .85 4.60 .00 935.0 .00 .00 .00 .00 .00 650.0 2.34 2.48 .00 .00 .00 940.0 .85 1.63 .00 .00 .00 655.0 .00 .00 .00 .65 .00 945.0 .00 .00 .00 4.04 .00 660.0 .00 .00 .00 .00 .00 950.0 1.03 1.08 .00 .00 .00 665.0 .00 .00 .00 .58 .00 955.0 .00 .00 .00 .87 .00 670.0 .77 1 .00 .00 .00 .00 960.0 .00 .00 .00 .00 .00 675.0 .00 .00 .51 4.65 .00 965.0 .00 .00 .00 .00 .00 .00 680.0 .00 1.34 .00 .00 .00 970.0 1.03 .61 .00 .00 685.0 .00 .00 .00 1.12 .00 975.0 .00 .00 .00 2.10 .00 690.0 .00 .00 .00 .00 .00 980.0 .00 1.37 .00 .00 .00 695.0 .00 .00 .00 .00 .00 985.0 .00 .00 .00 .00 .00 700.0 .00 1.86 .00 .00 .00 990.0 .00 .00 .00 .00 .00 705.0 .00 .00 .00 4 .47 .00 995.0 .00 .00 .00 .00 .00 710.0 .00 1 .34 .00 .00 .00 1000.0 .00 1.72 .00 .00 .00 715.0 .00 .00 .00 .00 .00 .00 720.0 .00 .00 .00 .00 725.0 .00 .00 .00 .00 .00 730 .0 .00 2.07 .00 .00 .00 735.0 .00 .00 .54 5.94 .00 .00 740.0 .74 .91 .00 .00 745.0 .DO .00 .00 .00 .00 750.0 .00 .00 .00 .00 .00 755.0 .00 .00 .00 1 .56 .00 760.0 2.15 2.17 .00 .00 .00 765.0 .00 .00 .78 4.33 .00 770.0 .00 .00 .00 .00 .00 775.0 .00 .00 .00 1.11 .00 780.0 .00 .00 .00 .00 .00 785.0 .00 .00 .00 .84 .00 .00 790.0 .00 1 .80 .00 .00 795.0 .00 .00 .00 3.96 .00 800.0 .00 .95 .00 .00 .00 805.0 .00 .00 .00 .00 .00 810.0 .00 .00 .00 .00 .51 815.0 .00 .00 .00 .00 .00 820.0 .00 2.04 .00 .00 .00 825.0 .00 .00 .00 2.27 .00 830.0 .00 1.13 .00 .00 .00 Table prepared using a Fourier analysis with a resolution o f 5 H z . 37 K s Smith. R Yacamini & A C Williamson Table II: Fourier analysis for generating condition Table 38 2 255 0 .00 .00 8.87 18.36 .00 260 0 1.83 2.19 .00 .00 .00 265 0 .55 1.41 .00 .00 .00 270 0 .00 .00 .00 .95 .00 Motor 275 0 .00 .00 .00 .00 .00 Current Current Current Voltage Torque 280 0 .00 7.71 .00 .00 .00 285 0 .00 .00 2.77 12.45 .00 .00 Generating Freq No system Cyclo Condition Motor Motor .0 .00 00 6.12 .00 100.00 290 0 .00 3.31 .00 .00 5.0 1.36 4 81 2.85 .00 1.58 295 0 .00 .50 .00 .00 .00 10.0 .00 6 28 2.62 .00 1.53 300 0 .00 .00 .00 .79 3.95 15.0 .00 00 100.00 100.00 3.16 305 0 .00 .75 .00 1 .14 .00 20.0 2.00 51 61 1.88 .00 1.21 310 0 .00 4.30 .00 .00 .00 25.0 .56 2 55 1.54 .00 .97 315 0 .00 .00 2.65 13.31 .00 30.0 .65 1 52 5.28 2.10 1.55 320 0 .00 6.09 .00 .00 .00 35.0 2.98 8 84 .89 .82 .58 325 0 .00 .00 .00 .00 .00 40.0 3.66 3 57 .71 .00 .00 330 0 .00 .00 .00 .58 .00 45.0 .92 2 13 .99 .69 .83 335 0 .00 .57 .00 .00 .00 50.0 100.00 100 00 .78 .00 .00 340 0 1.25 1.47 .00 .00 .00 55.0 1.60 3 70 .78 .57 .00 345 0 .00 .00 2.22 6.46 .00 60.0 .73 1 98 3 .90 2.49 .54 350 0 4.28 4.38 .00 .00 .00 65.0 1.53 4 80 .66 .00 .00 355 0 .00 .00 .00 .68 .00 70.0 1.60 00 .54 .00 .00 360 0 .00 .00 .00 .00 .00 75.0 .00 1 13 2.35 1.80 .00 365 0 .00 .60 .00 .00 .00 80.0 .76 22 52 .00 .00 .00 370 0 .00 1.25 .00 .00 .00 85.0 .82 2 82 .00 .00 .00 375 0 .00 .00 .76 4.29 .00 90.0 .00 51 3.03 2.83 1.46 380 0 .00 2.40 .00 .00 .00 95.0 1.02 3 75 .00 .50 .00 385 0 .00 .50 .00 .57 .00 100.0 .00 1 29 .00 .00 .00 390 0 .00 .00 .00 .00 .52 105.0 .00 72 .82 1.02 .00 395 0 .00 .00 .00 .00 .00 110.0 .00 8 35 .00 .00 .00 400 0 .00 .98 .00 .00 .00 115.0 .59 2 24 .00 .00 .00 405 0 .00 .00 .00 3.54 .00 120.0 .00 00 2.24 3 .05 .00 410 0 .00 1.06 .00 .00 .00 125.0 .86 2 81 .00 .62 .00 415 0 .00 .78 .00 .00 .00 130.0 .74 2 22 .00 .00 .00 420 0 .00 .00 .00 .00 .00 135.0 .00 00 1.13 1.42 .00 425 0 .00 .56 .00 .00 .00 140.0 3 .04 2 12 .00 .00 .00 430 0 .00 3 .17 .00 .00 .00 145.0 .00 1 72 .00 .00 .00 435 0 .00 .00 .84 4.03 .00 150.0 .00 00 1 .60 3 .05 .00 440 0 .00 .68 .00 .00 .00 155.0 .63 2 21 .00 .00 .00 445 0 .00 .86 .00 .51 .00 160.0 4.08 3 88 .00 .00 .00 450 0 .00 .00 .00 .98 .00 165.0 .00 00 1.65 1.47 .00 455 0 .00 .53 .00 .68 .00 170.0 .93 1 51 .00 .00 .00 460 0 4.75 4.92 .00 .00 .00 175.0 .68 1 91 .00 .00 .00 465 0 .00 .00 1.55 7.17 .00 180.0 .00 00 1.11 2.73 .00 470 0 .64 1.10 .00 .00 .00 185.0 .52 1 70 .00 .00 .00 475 0 .00 .00 .00 .71 .00 190.0 .00 13 42 .00 .00 .00 480 0 .00 .00 .00 1.81 .00 195.0 .00 60 1.73 4.48 .00 485 0 .00 .54 .00 .00 .00 200.0 .00 1 69 .00 .00 .00 490 0 .00 4.56 .00 .00 .00 205.0 .84 2 31 .00 .63 .00 495 0 .00 .00 .86 6.84 .00 210.0 .00 68 .84 2.50 2.36 500 0 .00 .00 .00 .00 .00 215.0 .00 77 .00 .00 .00 505 0 .00 .00 .00 .00 .00 510 0 .00 .00 .00 220.0 .65 22 89 .00 .00 .00 1.83 1.15 225.0 .00 75 3.80 14.90 .00 515 0 .00 .00 .00 .52 .00 230.0 .57 2 69 .00 .00 .00 520 0 .00 3.47 .00 .00 .00 235.0 .00 98 .00 .00 .00 525 0 .00 .00 .55 5.75 .00 240.0 .00 62 .59 1.67 .00 530 0 .00 2.43 .00 .00 .00 245.0 .00 74 .00 .00 .00 535 0 .00 .00 .00 .00 .00 250.0 11.25 11 15 .00 .00 .00 540 0 .00 .00 .00 1.87 .00 Trans IMarE. Vol 105. Part 1. pp 23-52 Table II: Fourier analysis for generating condition (cont) 545.0 .00 .00 .00 .51 .00 835 0 .00 .00 .00 .00 .00 550.0 2.97 2.76 .00 .00 .00 840 0 . 00 .00 .0.". .68 .00 555.0 .00 .00 .77 3.55 .00 845 0 .00 .00 .00 .82 .00 560.0 3.03 2.86 .00 .00 .00 850 0 1.25 1.10 .00 .00 .00 565.0 .00 .00 .00 .62 .00 855 0 .00 .00 .00 .00 .00 570.0 . .00 .00 1.76 .00 860 0 1.08 1.04 .00 .00 .00 575.0 .00 .00 .00 .56 .00 865 0 .00 .00 . 00 .00 .00 580.0 .00 2.74 .00 .00 .00 870 0 .00 .00 .00 .64 .00 585.0 .00 .00 .00 .80 .00 875 0 .00 .00 .00 .00 .00 590.0 .00 2.27 .00 .00 .00 880 0 .00 1.80 .00 .00 .00 595.0 .00 .00 .00 .00 .00 885 0 .00 .00 .00 .00 .00 600.0 .00 .00 .00 1.60 .00 890 0 .00 1.03 .00 .00 .00 605.0 .00 .00 .00 .00 .00 895 0 .00 .00 .00 .00 .00 610.0 .00 2.07 .00 .00 .00 900 0 .00 .00 .00 .71 .00 615.0 .00 .00 .00 4.10 .00 90S c .00 .00 .00 .61 .00 620.0 .00 1.88 .00 .00 .00 910 0 .00 .58 .00 .00 .00 625.0 .00 .00 .00 .00 .00 915 0 .cc .00 .00 1.00 .00 630.0 .00 .00 .00 .99 .00 920 0 .00 .92 .00 .00 .00 635.0 .00 .00 .00 .75 .00 925 0 .00 .00 .00 .00 .00 640.0 .98 .00 .00 .00 .00 930 0 .00 .00 .00 .92 .00 645.0 .00 .00 .56 2.17 .00 935 0 .00 .00 .00 .00 .00 650.0 1.82 1.88 .00 .00 .00 940 0 .72 2.19 .00 .00 .00 655.0 .00 .00 .00 .00 .00 945 0 .00 .00 .00 3.41 .00 660.0 .00 .00 .00 .00 .00 950 0 .98 .78 .00 .00 .00 665.0 .00 .00 .00 .00 .00 9S5 0 .00 .00 .00 .60 .00 670.0 1.45 1.88 .00 .00 .00 960 0 . i .00 .00 1.04 .00 675.0 .00 .00 .00 3.01 .00 965 0 .00 .00 .00 .00 .00 680.0 .00 1.37 .00 .00 .00 970 0 1.39 .83 .00 .00 .00 685.0 .00 .00 .00 .89 .00 975 0 .00 .00 .00 2.10 .00 690.0 .00 .00 .00 .00 .00 980 0 .00 1.25 .00 .00 .00 695.0 .00 .00 .00 .00 .00 985 0 .00 .00 .00 .00 .00 700.0 .00 2.40 .00 .00 .00 990 0 .00 .00 .00 .00 .00 705.0 .00 .00 .00 3.24 .00 995 0 .00 .00 .00 .00 .00 710.0 .00 1.34 .00 .00 .00 1000 0 .00 1.49 .00 .00 .00 715.0 .00 .00 .00 .00 .00 720.0 .00 .00 .00 .81 .00 725.0 .00 .00 .00 .72 .00 730.0 .00 2.75 .00 .00 .00 735.0 .00 .00 .00 4.81 .00 740.0 .00 1.01 .00 .00 .00 745.0 .00 .00 .00 .51 .00 750.0 .0 .00 .00 .55 .00 755.0 .00 .00 .00 .00 .00 760 .0 1.39 1.40 .00 .00 .00 765.0 .00 .00 .66 4.59 .00 770.0 1.15 1.33 .00 .00 .00 775.0 .00 .00 .00 .00 .00 .00 .00 .76 .00 .00 .00 .52 .00 780.0 785.0 . G .1 790.0 .00 1.72 .00 .00 .00 795.0 .00 .00 .00 1.97 .00 800.0 .00 1.50 .00 .00 .00 805.0 .00 .00 .00 .B ! .00 810.0 .00 .00 .00 .69 .00 815.0 .00 .00 .00 .G0 .00 820.0 .oc 1.17 .00 .00 .00 825.0 .00 .00 .00 2.61 .00 830.0 .00 1.56 .00 .00 .00 Table prepared using a Fourier analysis with a resolution o f 5 H z . 39 K S Smith. R Yacamini & A C Williamson Table III: Fourier analysis for double wound motor Table No 3 255.0 .00 .00 5.41 23.25 .00 260.0 .00 1.92 .00 .00 .00 265.0 .00 .00 .00 .86 .00 270.0 .00 .00 .00 .00 .00 Motor 275.0 .00 .00 .00 .00 .00 Current Current Current Voltage Torque 280.0 .00 3.13 .00 .00 .00 285.0 .00 .00 1.74 24.79 .00 Double Freq 40 System Wound M o t o r Cyclo Motor Motor .0 .00 .00 .00 .00 IOC .00 290.0 .00 5.87 .00 .00 .00 5.0 .00 .00 .00 .00 .00 295.0 .00 .00 .00 .70 .00 10.0 .98 5.18 .00 .00 .00 300.0 .00 .00 .00 .00 .00 15.0 .00 .00 100.00 100.00 .00 305.0 .00 .00 .00 .99 .00 20.0 .00 58.62 .00 .00 .00 310.0 .00 1.85 .00 .00 .00 25.0 .00 .00 .00 .00 .00 315.0 .00 .00 1.47 24.27 .00 30.0 .00 .00 .00 .00 .00 320.0 .00 8.21 .00 .00 .00 35.0 .00 .00 .00 .00 .00 325.0 .00 .00 .00 1.85 .00 40.0 .00 .79 .00 .00 .00 330.0 .00 .00 .00 .00 .00 45.0 .00 .00 6.03 5.06 .00 335.0 .00 .00 .00 .76 .00 50.0 100.00 100.00 .00 .00 .00 340.0 .00 1.02 .00 .00 .00 55.0 .00 .00 .00 1.40 .00 345.0 .00 .00 1.80 10.55 .00 60.0 .00 .00 .00 .00 .00 350.0 2.41 2.65 .00 .00 .00 65.0 .00 .00 .00 .00 .00 355.0 .00 .00 .00 .00 .00 70.0 .71 .52 .00 .00 .00 360.0 .00 .00 .00 .00 .00 75.0 .00 .00 4.06 5.05 .CO 365.0 .00 .00 .00 .88 .00 80.0 .00 23.92 .00 .00 .00 370.0 .00 .58 .00 .00 .00 85.0 .00 .00 .00 1 .34 .00 375.0 .00 .00 1.02 7.06 .00 90.0 .00 .00 .00 .00 .00 380.0 .00 2.51 .00 .00 .00 95.0 .00 .00 .00 .00 .00 385.0 .00 .00 .00 .86 .00 100.0 .00 .00 .00 .00 .00 390.0 .00 .00 .00 .00 .00 105.0 .00 .00 2.72 5.69 .00 395.0 .00 .00 .00 1.56 .00 110.0 .83 8.01 .00 .00 1)0 400.0 .00 .50 .00 .00 .00 115.0 .00 .00 .00 2.16 .00 405.0 .00 .00 .54 4.87 .00 120.0 .00 .00 .00 .00 .00 410.0 .00 1.10 .00 .00 .00 125.0 .00 .00 .00 .94 00 415.0 .00 .00 .00 .00 .00 130.0 .54 .65 .00 .00 OC 420.0 .00 .00 .00 .00 .00 135.0 .00 .00 1.35 3.52 00 425.0 .00 .00 .00 1.15 .00 140.0 .00 3.35 .00 .00 00 430.0 .00 1.83 .00 .00 .00 145.0 .00 .00 .00 .00 00 435.0 .00 .00 .00 3.64 .00 150.0 .00 .00 .00 .00 00 440.0 .00 .00 .00 .00 .00 155.0 .00 .00 .00 1.50 00 445.0 .00 .00 .00 .83 .00 160.0 .00 2.21 .00 .00 00 450.0 .00 .00 .00 .00 .00 165.0 .00 .00 .98 2.91 00 455.0 .00 .00 .00 1.30 .00 170.0 .72 1.19 .00 .00 00 460.0 .00 4.00 .00 .00 .00 175.0 .00 .00 .00 .00 00 465.0 .00 .00 .54 4.41 .00 180.0 .00 .00 .00 .00 00 470.0 .00 .00 .00 .00 .00 185.0 .00 .00 .00 1.27 00 475.0 .00 .00 .00 .63 .00 190.0 .00 9.78 .00 .00 00 480.0 .00 .00 .00 .00 .00 195.0 .00 .00 1 .00 4.10 00 485.0 .00 .00 .00 .70 .00 200.0 .00 .00 .00 .00 00 490.0 .00 5.89 .00 .00 .00 205.0 .00 .00 .00 1.33 00 495.0 .00 .00 .63 6.91 .00 210.0 .00 .00 .00 .00 00 500.0 .00 .64 .00 .00 .00 215.0 .00 .00 .00 1.03 00 505.0 .00 .00 .00 .64 .00 220.0 .00 23.58 .00 .00 00 510.0 .00 .00 .00 .00 .00 225.0 .00 .00 2.13 8.84 00 515.0 .00 .00 .00 .71 .00 230.0 .00 .00 .00 .00 00 520.0 .00 1.84 .00 .00 .00 235.0 .00 .00 .00 .00 00 525.0 .00 .00 .74 6.13 .00 240.0 .00 .00 .00 .00 00 530.0 .00 1.56 .00 .00 .00 245.0 .00 .00 .00 1.07 00 535.0 .00 .00 .00 .00 .00 250.0 6.88 7.65 .00 .00 00 540.0 .00 .00 .00 .00 .00 Trans IMarE. Vol 105, Pan 1. pp 23-52 Table III: Fourier analysis for double wound motor (cont) 545.0 .00 .00 .00 .76 .00 835 .0 .00 .00 .00 1.13 .00 550.0 1.48 1.62 .00 .00 .00 840 .0 .00 .00 .00 .00 .00 555.0 .00 .00 .00 3.05 ,00 845 .0 .00 .00 .00 .63 .00 560.0 .00 3.04 .00 .00 .00 850 .0 .55 .56 .00 .00 .00 .00 565.0 .00 .00 .00 .00 .00 855 .0 .00 .00 .00 1.47 570.0 .00 .00 .00 .00 .00 860 .0 .00 .96 .00 .00 .00 575.0 .00 .00 .00 .83 .00 865 .0 .00 .00 .00 .00 .00 580.0 .00 .89 .00 .00 .00 870 .0 .00 .00 .00 .00 .00 585.0 .00 .00 .00 2.71 .00 875 .0 .00 .00 .00 .00 .00 590.0 .00 3.18 .00 .00 .00 880 .0 .00 .00 .00 .00 .00 595.0 .00 .00 .00 1.11 .00 885 0 .00 .00 .00 2.59 .00 600.0 .00 .00 .00 .00 .00 890 0 .00 .73 .00 .00 .00 605.0 .00 .00 .00 .63 .00 895 0 .00 .00 .00 .74 .00 610.0 .00 .00 .00 .00 .00 900 0 .00 .00 .00 .00 .00 615.0 .00 .00 .00 7.58 .00 905.0 .00 .00 .00 .59 .00 620.0 .00 1.43 .00 .00 .00 910 0 .00 .00 .00 .00 .00 625.0 .00 .00 .00 .67 .00 915 0 .00 .00 .00 3.25 .00 630.0 .00 .00 .00 .00 .00 920 0 .00 .00 .00 .00 .06 635.0 .00 .00 .00 1.05 .00 925 0 .00 .00 .00 1.17 .00 640.0 .00 .00 .00 .00 .00 930 0 .00 .00 .00 .00 .00 645.0 .00 .00 .00 2.50 .00 935 0 .00 .00 .00 1.45 .00 650.0 .84 .88 .00 .00 .00 940 0 .00 .51 .00 .00 .00 655.0 .00 .00 .00 .00 .00 945 0 .00 .00 .00 1.39 .00 660.0 .00 .00 .00 .00 .00 950 0 .00 .00 .00 .00 .00 665.0 .00 .00 .00 1.20 .00 955 0 .00 .00 .00 .70 .00 670.0 .00 .53 .00 .00 .00 960 0 .00 .00 .00 .00 .00 675.0 .00 .00 .00 .00 .00 965 0 .00 .00 .00 1.03 .00 680.0 .00 .75 .00 .00 .00 970. 0 .66 .00 .00 .00 .00 685.0 .00 .00 .00 .73 .00 975. 0 .00 .00 .00 1.25 .00 690.0 .00 .00 .00 .00 .00 980. 0 .00 .51 .00 .00 .00 695.0 .00 .00 .00 1 .19 .00 985. 0 .00 .00 .00 .65 .00 700.0 .00 1.14 .00 .00 .00 990. 0 .00 .00 .00 .00 .00 705.0 .00 .00 .00 1 .40 .00 995. 0 .00 .00 .00 1.46 .00 710.0 .00 .00 .00 .00 .00 715.0 .00 .00 .00 .00 .00 720.0 .00 .00 .00 .00 .00 725.0 .00 .00 .00 .53 .00 Table prepared using a Fourier analysis with a resolution of 5 H z . 730.0 .00 1.91 .00 .00 735.0 .00 .00 .00 6.16 .00 740.0 .00 .00 .00 .00 .00 745.0 .00 .00 .00 .66 .00 750.0 .00 .00 .00 .00 .00 .00 755.0 .00 .00 .00 1.12 .00 760.0 .00 1.72 .00 .00 .00 765.0 .00 .00 .00 6.20 .00 770.0 .67 .72 .00 .00 .00 .00 775.0 .00 .00 .00 .58 780.0 .00 .00 .00 .00 .00 785.0 .00 .00 .00 .64 .00 790.0 .00 .76 .00 .00 .00 795.0 .00 .00 .00 7.86 .00 .00 800.0 .00 .96 .00 .00 805.0 .00 .00 .00 1.35 .00 810.0 .00 .00 .00 .00 .00 .00 815.0 .00 .00 .00 .00 820.0 .00 .66 .00 .56 .00 825.0 .00 .00 .00 2.75 .00 830.0 .00 1.48 .00 .00 .00 41 K S Smith. R Yacamini & A C Williamson Table IV: Fourier analysis for simulation including alternator Table No 4 ulation including alternator Freq 5.40 9.68 255 .0 .00 .00 260.0 1.13 1.89 265.0 .00 .00 .77 270 .0 .00 .00 2.36 .62 .94 1 .00 System System Machine 275.0 .00 .00 .84 Voltage Current Torque 280.0 .00 .63 1.30 285.0 .00 .00 1.33 .c .00 .00 100.00 290 .0 .00 .00 .84 5. 0 .00 .00 2.28 295 .0 .00 .00 .00 10 . ii .00 1.77 3.16 300 .0 .00 .00 30.64 15 . 0 .00 .00 3.82 305 .0 .00 .00 .00 20 .00 3.19 2.98 310 .0 .00 .00 .72 25 . 0 .00 .00 1.43 315 .0 .00 .00 .87 30 .0 .00 1.06 6.84 320 .0 .00 .74 .66 35 . 0 .00 1.75 .97 325 .0 .00 .00 .00 40 . c .00 1.54 1.49 330 .0 .00 .00 1.55 45 . 0 .00 1.00 1.55 335 .0 .00 .00 .00 100.00 100.00 1.09 340 .0 .94 1.18 .00 55 . c .00 .90 .55 345 .0 .00 .00 .00 60 . 0 .00 1.32 2.96 350 .0 2.92 3.90 .00 c5 . c .00 1 .45 .00 355 .0 .00 .00 .00 70 . 0 .00 1.50 .90 360 .0 .00 .00 1.07 75 .0 .00 .00 1.01 365 .0 .00 .00 .00 80 . c .00 1.54 .69 370 ,0 .00 .51 .00 85 .00 .00 .00 375 ,0 .00 .00 .00 90 .0 .00 .52 6.96 380 ,0 .00 .00 .00 95 .c .00 .57 .00 385 .0 .00 .00 .00 100 .0 .00 .00 .62 390 ,0 .00 .00 3 .74 50 . - 42 250 .0 105 .0 .00 .00 .70 395 .0 .00 .00 .00 110 .0 .00 .81 .54 400 .0 .00 .00 .00 115 . 0 .00 .00 .00 405 ,0 .00 .00 .00 120 .0 .00 .00 1.77 410 .0 .00 .00 .00 125 .0 .00 .00 .00 415 ,0 .00 .00 .00 1 tc.0 .00 .69 .00 420 ,0 .00 .00 .68 135 .0 .00 .00 .55 425 .0 .00 .00 .00 140. .0 .73 2.45 .00 430 ,0 .00 .00 .00 145, .0 .00 .00 .00 435 .0 .00 .00 .00 I5i; ,0 .00 .00 .94 440 ,0 .00 .50 .00 155 .0 .00 .00 .00 445 ,0 .00 .00 .00 160 ,0 .00 .97 .00 450 ,0 .00 .00 .00 165 .0 .00 .00 .00 455 .0 .00 .00 .00 170 .0 .00 .00 .00 460 .0 2.56 2.56 .OC 175 . 0 .00 .00 .00 465 .0 .00 .00 .00 180 .0 .00 .00 1.39 470 ,0 .00 .00 .00 185 . c .00 .00 .00 475 ,0 .00 .00 .00 190 ,0 .00 .00 .00 480 ,0 .00 .00 .87 195 .0 .00 .00 .00 485 .0 .00 .00 .00 200 .0 .00 .00 .00 490 ,0 .00 .00 .00 205 . Ü .00 .00 .00 495 .0 .00 .00 .00 210 .'! .00 .00 3 .80 500 .0 .00 .00 .00 215 .00 .00 .00 505 .0 .00 .00 .00 220 .0 .00 .58 .00 510 .0 .00 .00 10.72 225 .0 .00 .00 .00 515 .0 .00 .00 .00 230 .0 .00 .56 .00 520 ,0 .57 .00 .00 235 .0 .00 .00 .00 525 .0 .00 .00 .00 240 . Ü .00 .00 .78 530 ,0 .00 .00 .00 245 . 0 .00 .00 .00 535 .0 .00 .00 .00 Trans IMarE. Vol 105. Pan l.pp 23-52 Table IV: Fourier analysis for simulation including alternator (cont) 540.0 .00 .00 1.15 830.0 .00 .00 .00 545.0 .00 .00 .00 835.0 .00 .00 .00 550.0 2.46 1.97 .00 840.0 .00 .00 .58 555.0 .00 .OC .56 845.0 .01) .00 .00 560.0 2.09 1.77 .52 850.0 1.14 .61 .00 565.0 .00 .00 .00 855.0 .00 .uu .00 570.0 .00 . o: 1.63 860.0 2.21 1.18 .00 575.0 . 00 .00 .00 865.0 .00 .00 .00 580.0 .00 .00 .00 870.0 .00 .00 .67 585.0 .00 .00 .00 875.0 .00 .00 .00 590.0 .00 .00 .00 880.0 .00 .00 .00 595.0 .00 .00 .00 885.0 . C') .00 .00 600.0 .00 .00 6.99 890.0 .00 .00 .00 605.0 .00 .00 .00 895.0 .00 .00 .00 610.0 .00 .00 .00 900.0 .00 .00 2.44 615.0 .00 .00 .00 905.0 .00 . 0 (] .00 620.0 .00 .00 .00 910.0 .00 .00 .00 625.0 .00 .00 .00 915.0 .00 .00 .00 630.0 .00 .00 .00 920.0 .00 .00 .00 635.0 .00 .00 .00 925.0 .00 .00 .00 640.0 .61 .00 .00 930.0 .00 .00 .00 645.0 .00 .00 .00 935.0 .00 .00 .00 650.0 1.96 1.41 .00 940.0 .00 .00 .00 655.0 .00 .00 .00 945.0 .00 .00 .00 660.0 .00 .00 .00 950.0 1.13 .55 .00 665.0 .00 .00 .00 955.0 .00 .00 .00 670.0 .88 .61 .00 960.0 .00 .00 .00 675.0 .00 .00 .00 965.0 .00 .00 .00 680.0 .00 .00 .00 970.0 1.79 .82 .00 685.0 .00 .00 .00 975.0 .00 .00 .00 690.0 .00 .00 1 .36 980.0 .00 .00 .00 695.0 .00 .00 . 00 985.0 .00 .00 ,00 700.0 .00 .00 .00 990.0 .00 .00 .59 705.0 .00 .00 .00 995.0 .00 .00 ,00 710.0 .00 .00 .00 715.0 .00 .00 .00 720.0 .00 .00 2.41 725.0 .00 .00 .00 730.0 .1)0 .00 .00 735.0 .00 .00 .00 740.0 .00 .00 .00 745.0 .00 .00 .00 750.0 . n Q .00 .00 755.0 . 'J : .00 .00 760.0 2.66 1.58 .00 765.0 .Ù0 .00 .00 770.0 .75 .00 .00 775.0 .01) .00 .00 780.0 .00 . CO .52 .00 785.0 .C0 .CO 790.0 .00 .00 .00 795 .0 .00 . 00 .00 800.0 .00 .00 .00 .00 .00 810.0 .00 .00 5.95 815.0 .00 .00 .00 820.0 . i: .00 .00 825.0 .00 .00 .00 805.0 Table prepared using a Fourier analysis with a resolution of 5 H z . 43 K S Smith. R Yacamini A C Williamson MAG(%) 20- 18-^ Reference: Table 1 Signai: ac system current 16-^ 14-^ 12 J 10^ 4-^ 2-^ 200 400 600 800 i Ik • 111., 111. ..111., 1.2k 1.4k JUl 1.6k IIIII I i l II 1.8k II II iLiL 2k f(Hz) Fig 25: Frequency spectra of harmonics present in the ac system current for motoring condition (see Fig 14 and Table I) MAG(%) 55- 50-^ Reference: Tabic 1 Signal: cycloconverter input current 45-4 40—^ 35-J 30-J 25-^ 20-^ 15^ 10_J 5-4 idiJi 200 400 1 I N I il ll II li 1.1[ ll II li II II il |i II I. IIIIII \lli 11 11 II .1II i| Il II u II .. II [I . .1II • • l.^l. • I I 1 . . .11 600 800 Ik 1.2k 1.4k 1.6k 1.8k 2k 2 f(Hz) Fig 26: Frequency spectra of harmonics present in the cycloconverter input current (see Fig 13 and Table I) 44 Trans IMarE. Vol 105. Pan 1. pp 23-52 MAG(*) 60 - 55 Reference: Table 1 Signal: motor stator voltage 50-1 45 — 40- 35- 30- 25- 20- 1.6k 1.8k 2k f(H7,) Fig 27: Frequency spectra of harmonies present in the motor stator voltage (see Fig 10 and Table I) MAG(^Ç-) 40- Reference: Table 1 Signal: motor stator current 35-^ 30-^ 20-J 10-J 5-^ -i—„J< à i h 200 400 600 800 Ik .2k 1.4k ].6k 1.8k 2k r(Mz) Fig 28: Frequency spectra of harmonics present in the motor stator current (see Fig 9 and Table I) 45 K S Smith. R Yacamini & AC Williamson MAG(%) 30- 27.5 - Reference: Sicnal: 25- 22.5 Table 1 motor electromagnetic torque - 20- 17.5 - 15 - 1 2.5 — 10- 7.5 - 5- 2.5 - 0 J-X 200 600 Ik 800 1.2k 1.4k 1.6k 2k 1,8k f(Hz) Fig 29: Frequency spectra of harmonics present in the motor electromagnetic torque (see Fig 12 and Table I) MAG(%) 20- 18 J Reference: Table 4 Signal: ac system voltage 16 . J 14-^ 10-4 8-^ 2-^ A 200 400 600 800 Ik 1,2k 1,4k 1.6k 1.8k 2k 2.2k 2.: Fig 30: Frequency spectra of harmonics present In the ac system voltage (see Table IV) 46 f(H7.) Trans IMarE. Vol 105. Part 1. pp 23-52 MAG(%) 20- Reference: Table 4 Sienal: ac system current 18-4 16 14 12-^ 10 J 6-^ 4-4 2-^ À. A 200 400 600 -* M . . Ai t - A- - g T .4k 800 1 1.6k • •• 1 2k 1.8k f(Hz) Fig 31 : Frequency spectra of harmonics present In the ac system/alternator atator windings (see Fig 24 and Table IV) MAG(%) 50- 45 J Reference: Table 4 Signal: alternator torque 40-4 35 J 30 J 25-J 20-^ 15 10-^ 5-^ V ^ W i 0 MAG(%) : f ( M z ) 200 400 600 800 h. II i Ï 1.2k 1.4k V 1.6k • 1.8k 2k f ( H z ) (.3)1 o r Fig 32: Frequency spectra of harmonics present In the electromagnetic torque of the alternator (see Fig 23 and Table IV) 47 K S Smith. R Yacamini &AC Williamson Discussion C d r N A Haines (Royal Naval Engineering College, Manadon) 1. The authors have mentioned in their paper that their Saber simulations have been able to model both the cycloconverters and the generators and motors in the proposed propulsion system. With specü"ic regard to the motors and generators, I wonder if they could explain more fully the type of ac machine models used, ie d-q models, phase models or equivalent circuit models. In addition, could they please explain how they obtained representative ac machine parameters for their chosen models, and how sensitive the accuracies of die simulations are to diese parameter values? 2. In their paper, the authors have shown in Fig 17 two six pulse cycloconverters feeding a double wound stator induction motor. 1 wonder if the authors could comment on the advantages they see in employing this type of converter/motor arrangement in preference to a single 12pulse converter with a conventional singly-wound stator induction motor. • K S Smith, • R Yacamini and t A C Williamson (•University of Aberdeen and t U M I S T ) The authors thank Cdr Haines for his question which addresses the problems associated widi modell ing electrical machines and obtaining parameters for the machines. Three methods are available to model the electrical machines in the proposed cycloconverter propulsion system. These are two-axis (or d-q) models, phase models, and equivalent circuit models. All three methods have previously been used by the research group at Aberdeen University. Experience has shown diat die phase model is the most convenient forihis type of simulation. The phase model allows the effects of zero-sequence currents and voltages to be correctly calculated as well as allowing unbalances within the electrical machine to be analysed. Equivalent circuit models give the machine impedances at different frequencies. As such they belong to the realm of frequency rather than time domain analysis. The simulation is based upon the voltage balance and rotor dynamic equations for the induction machine and synchronous machine primitive models shown in Figs 6 and 7 of the paper These equations are solved in the time domain using a time stepping integration technique. For the induction machine, this gives a six by six impedance matrix and for the synchronous machine a seven by seven matrix. For the double wound induction machine (see Fig 16 of the paper) a nine by nine matrix is obtained. Some of the inductances in the matrix are time varying, due to the rotation of die rotor, which requires that the impedance matrix is recalculated at each time step. The use of a two-axis machine representation would considerably reduce the size of the matrices and remove the necessity to recalculate the impedance matrix at each time step. However in order to link the machine model with the cycloconverter model a phase to two-axis and two-axis to phase conversion would be required in the simulation. It has been found üiat implementing this conversion within Mast, the programming language used by Saber produced computer run times which exceed those of the full phase model. For this reason phase models are our preferred method of representing machines. The machine parameters used in the simulations are based upon the standard parameters available from manufacturers. For the induction motor the single phase equivalent circuit 48 with the rotor parameters referred to the stator is used. From this the mutual inductances between stator phases, rotor phases, and stator and rotor phases required for the voltage balance equations can be determined. For the synchronous machine, the standard direct axis and quadrature axis reactances and time constants are used to determine the self inductances and mutual inductances for the windings of the primitive synchronous machine model shown in Fig 7 of the paper The accuracy of the parameters available for elecuical machines is generally poor, and the authors acknowledge that a simulation is only as accurate as die parameters used. For this reason is it necessary when performing case studies, to run the simulations using a range of parameters within the tolerances set by the manufacturers? In a computer simulation, exact agreement between measured and simulated responses is unlikely to be obtained. The value of computer simulation does not lie in determining the exact response of a system. Its usefulness lies in the ability to study the performance of a system around a particular operating point, and to assess die effects of changes upon die system performance. The principal advantage of the double wound winding over the single wound winding for die induction motor stator when supplied from six pulse cycloconverters, is that the 5th and 7th harmonic fluxes produced by the abc and uvw windings cancel each other out, removing the characteristic 6th harmonic from the rotor electromagnetic torque. The 6th harmonic torque would be present in a single wound motor supplied from a six pulse cycloconverter A single 12 pulse cycloconverter with a conventional single wound stator induction motor would also be free from the 6th harmonic torque. Although this suggests that there is no particular advantage to be gained from the use of the double wound machine, from an operational point of view some advantages do exist. For example, it would be possible to carry out maintenance on the converters which supply the uvw windings, whilst the motor continues to operate (on reduced power), on the abc windings. C d r C G Hodge (Royal Navy, Foxhill, Bath). 1 presume that the digital simulation of the electrical system used fixed values of inductance. The graphs of harmonic content of the voltage and current waveforms produced by the simulation could be used to calculate the harmonic impedance of the machine. If this is done. I believe it would show a fundamental impedance üiat could be related to the synchronous impedance of the machine; and harmonic impedances that could be related to a u^sient impedance and their harmonic number Can the auüiors explain how simulation based on a fixed value of inductance can produce results which exhibit variable inductance levels? * K S Smith, *R Yacamini and t A C Williamson (*University of Aberdeen and t U M I S T ) Cdr Hodge's question is closely related to that of Cdr Haines. As suggested, we have subsequently calculated the harmonic impedance of the synchronous machine using the current and voltage spectra obtained from the time domain simulation. For this particular exercise the machine's direct axis parameters were set to Xj = 1.5 pu, Xd- " 0.3 pu, and Xj. = 0.1 pu, on an S base of 6 M V A , at a rated voltage of 6.6 k V . Using the Fourier post-processor gave the following amplitudes (peak values) for the 6k + 1 hamionics present in the alternator voltage and current waveshapes when supplying a six pulse cycloconverter (the Trans IMarE. Vol 105. Part 1. pp 23-52 voltage is line to line). From this die harmonic impedance at each harmonic number can be Harmonic determined (see Table I). n From these impedance calculations it would appear that the synchronous reac1 5 tance of the machine is 11.49 ohms, and that 7 the impedance to the harmonics is approxi11 mately given by 1.3n, where 1.3 is the 13 average value of die impedances in the last 17 column of the table (ignoring die funda19 mental) and n is the appropriate harmonic number Convening these to per-unit gives a synchronous reactance of 11.48/7.26 = 1.58 pu, and a harmonic impedance of 1.3n/7.26 = 0.18n pu. Considering the parameters which were inidally input to the simulation, it would appear that the time domain simulation predicts that the impedance to the fundamental corresponds to die direct axis synchronous reactance (Xj = 1.5 pu), and that die impedance to harmonics corresponds to (x^. + \g)l2. This is in agreement with the suggestions made by Cdr Hodge. Table I shows that the time domain simulation can display the phenomenon described as 'variable inductance levels'. We do not agree with this description. The table shows the variation of machine reactance with frequency (the resistive component is ignored). These reactances have been obtained by converting die current and voltage waveshapes produced by the time domain simulation into the frequency domain, and then calculating the impedance in the frequency domain. In time domain analysis, reactance has no meaning. The voltage balance equations are expressed in terms of resistances, inductance, and instantaneous rate of change of current. It should be noted, however, that some of the mutual inductances in the synchronous machine model vary with rotor angular position. These time varying inductances can therefore be described as 'variable inductances'. The term 'direct axis reactance' is an unfortunate one. This expresses the machine impedance at a particular rated frequency; perhaps when dealing with time domain analysis the term 'direct axis inductance' would be more appropriate. M Murphy ( C E G E L E C Projects Ltd, Rugby) Thank you once again to die authors for their very interesting paper following closely on the heels of an earlier contribution at the Institute. I have two questions: 1. The ripples shown in the ac motor currents (Fig 9 of the paper) represent those achieved in die first generation designs and later motors have much lower ripples. However, the machine torque shows a cyclic ripple (Fig 12 of the paper) which is not observed on a real system. Can the authors explain this? 2. How do the ac harmonics and resultant ac voltage (supply side) compare wiüi a six pulse dc drive? Has any work been done to explore the optimum relationships for motor pole number, motor frequency and ac supply frequency which may optimise the quality of ac supply waveforms? • K S Smith, •R Yacamini and t A C Williamson (•University of Aberdeen and tUMIST) The auüiors Üiank Mr Murphy for his kind comments and also for setting us two challenging questions. We understand from our discussions with drive manufacturers, diat it is now possible to design induction motors for cycloconverter applications which have leakage inductances which are larger than additional designs. Limiting die starting current for direct on-line starting is not a problem with cycloconverter drives. This increased leakage inductance will Table I: Harmonic impedance calculations Voltage (V) Current (A) Impedance Z (ohms) 9027.27 509.94 266.22 206.76 175.34 112.74 114.20 785.22 75.83 30.57 15.52 11.05 4.74 4.26 11.49 6.72 8.71 13.25 15.86 23.78 26.81 Z/n (ohms) 11.49 1.34 1.24 1.20 1.22 1.39 1.41 limit the amplitude of the motor current ripple. In the simulation results presented in the paper, the motor parameters chosen for the simulation are based on 'first' generationmotor designs. The motor torque shown in Fig 12 of the paper is actually the air gap electromagnetic torque. The torque delivered to the load through the motor shaft coupling differs from this due to die inertia of the rotor die ripple in the torque producing a small speed variation. The air gap elecu-omagnetic torque cannot be measured directly and we assume that the torque waveshape observed by M r Murphy is in fact measured at the motor coupling. This is illustrated using the simulation shown in Figs 1 a and 1 b overleaf. The electromagnetic torque and load torque are shown along with the small variations in the rotor speed. Another feature of real systems which is not present in die simulation is the closed loop speed controller. Modem drives use a vector controller to control the torque produced by the motor, by controlling the motor line currents. We would suggest that the torque control loop will have sufficient bandwidth to damp low frequency ripple. We are currently investigating this effect. In answer to the second question, for a traditional six pulse dc drive, the current harmonics injected into the ac system correspond to die characteristic 5di, 7di, l l ü i . 13di, etc harmonics. The amplitude of the current depends upon the load characterisücs and the speed of the dc machine (which determines die back emf). Speed conu-ol is achieved by varying the converter firing angle which controls the voltage applied to the armature of die machine (assuming constant field excitation). The frequencies present in the line current do not change with converter firing angle (assuming the dc current has no ripple component, and that the commutation overlap angle is small) nor does their amplitude relative to die fundamental component. An earlier paper illustrates the waveshapes obtained from site measurements and the Saber simulation for dc drives (see Ref 1 overleaO. With a six pulse cycloconverter drive, the current harmonics injected into the ac system include the same characteristics harmonics and in addiuon modulation product frequencies. The frequency of the side-bands is a function of the converter output frequency, and it is diese components which giveriseto die presence of frequencies which are neither harmonics or subharmonics of die fundamental frequency. The convener input transformer connections also influence the frequencies present in the ac line current. The complete family of frequencies present in the line current can therefore be divided into two groups: those due to the modulation process, and those which are circuit dependent. These harmonic frequencies are discussed at length by Pelly (see Ref 2 overleaf). For a cycloconverter with a balanced three phase output, the first sideband terms to appear in die family of frequencies is die fundamental supply component modulated by six times die output frequency. The amplitude of this component is usually less than 4% of die fundamental. The conuibution made to the 49 K S Smith. R Yacamini & A C Williamson Electromagnetic torque 500 Load torque 420 J 0 \ I \ I L 10 Time (s) x10-2 15 20 Fig la: Variation of torque with time 450 C 5 Shaft speed 440 X «-1 ^ TD 430 S 420 Q. (/î ro 410 400 10 Time(s) xlO"^ Fig 1b: Variation of shaft speed with time 50 15 20 Trans IMarE. Vol 105, Pan 1. pp 23-52 distortion of the line current is limited, and does not therefore present any appreciable subharmonic load to die input system, if this system can be regarded as infinite. The distortion of the input current wave is determined substanually by the circuitdependent harmonic components. With circuits of sufficiently high pulse number it is possible to obtain current waveshapes with very low distortion. It is claimed by Pelly that for three phase cycloconverters supplied from large ac systems, the distortion of the input current waveform is generally less than that of the input current waveform of a rectifier, with the same pulse number supplying a dc output. The resulting ac voltage distortion for a cycloconverter drive is therefore less than for the dc drive. It should be noted, however, üiat subharmonic frequencies will be present in the ac line current, which will produce low frequency torques on the rotor of a ship's alternator. The work carried out at Aberdeen University has not considered in detail the optimum relationships for motor pole pair number, motor frequency, or ac supply frequency. However it is interesting to consider here the possible advantages offered by increasing the ac supply and motor frequencies. Traditional cycloconverter designs would limit the output frequency to around one third of the input frequency. In die case of a 60 Hz system, diis would give a maximum cycloconverter output frequency of 20 Hz. In order to produce the required 200 rev/ min shaft speed, a motor with six pole pairs would be required. If the ac system frequency is increased to 90 Hz, the upper operating frequency of the cycloconverter becomes 30 Hz. This would require a motor widi twelve pole pairs. At this higher operating frequency, the impedance of the motor (assuming the same inductances) to the characteristic cycloconverter harmonics is increased, reducing the current ripple and improving die electomagnetic torque waveshape. If die maximum output frequency is further increased, beyond the traditional limit to 60 Hz, a motor widi 18 pole pairs is required. At this higher operating frequency the advantage of higher machine imp)edance is further enhanced, again reducing the harmonics present in the motor line current. Raising the ac system supply frequency from 60 to 90 Hz, also will increase the impedance of the alternators, decrease die amplitude of the supply current harmonics and so improve the ac system voltage. An optimisation study would dierefore appear attractive. References 1. R Yacamini. Lihua Hu and ID Stewart. 'Electric drives on ships and oil platfomis". Trans IMarE. Vol 104, Part 4 (1992). 2. B R Pelly, Phase Controlled Converters and Cycloconverter, Wiley. New York (1971). Professor J O Flower (Department of Engineering, University of Warwick) This has been a most interesting presentation, on a fascinating topic, by the audiors, and diere are a number of questions I should like to ask. It is unclear to me, bodi from reading die paper and from listening to the lecture, as to the interplay between the Sabre and die Mast software used in the simulations. Please may I have some further explanations? Further, I should be most interested to know why die authors make special mention of die BackwardEuler and the Trapezoidal integration routines for this application. Is it merely that, for this integration problem, first and second order methods are accurate enough and, thereby, the integration process can be speeded up to some considerable extent? During the simulations I wonder how long the process took to settle down to give, for example, the steady state waveforms such as those shown in Fig 14 of the paper Since it is stated, in the paper that there are frequencies present which are neither harmonics nor subharmonics of the fundamental frequency, then we have an aperiodic waveform; is this a problem? The ingredients of these problems are those which can lead to chaotic behaviour; have the audiors noted any evidence of this in their results? 1 was a little surprised to see that there is interest in power supply frequencies up to 90 Hz. I wonder why this is seen as advantageous. These higher frequencies will certainly lead to higher losses, although, to compensate, it should be possible to raise die power-to-weight ratio of the machines. This presumably might be desirable in die confined space of a frigate. Would the authors comment on this, please? Just a comment to finish with. M r Yacamini told us about his ambition for including propeller and ship dynamics in future work; a laudable ambition and 1 look forward to learning of diis work in due course. However, we have also heard this evening of the difficulties of obtaining electrical machine parameters for simulation purposes. I fear diese difficulties will be nought compared with the problem of obtaining data on marine propellers under transient conditions. • K S Smith, •R Yacamini and t A C Williamson (•University of Aberdeen and tUMIST) The authors diank Professor Flower for his comments. The software used for the analysis is called Saber. This name is applied to the complete simulation system. Mast is a programming language which is used to supply information about die characteristics of components which constitute the system. A compilation process converts die Mast programme into code which die Saber executable file can interpret. This is then used to set up the maüices which describe the system to be anal ysed. When tlii s stage i s completed, die analysis of the system can proceed. The integration algorithms available within the Saber simulator are restricted to the first and second order Backward-Euler and Trapezoidal methods. The software vendors have advised us that higher order methods are generally not suitable when analysing complex systems, and so they are not made available in Saber, which is essentially a general purpose equation solver. The question of settling down time is an interesting one. Approximately Is of simulation time is required to ensure diat die transient associated with the initial point of the simulation has decayed. As the cycloconverter produces frequencies which are neitlier harmonics or subharmonics of the supply frequency, a very long period may be required before the waveshapes repeat themselves. If, however, the input and output frequencies do remain constant, then a definite period associated with the lowest frequency generated can be identified. For example, if the ac supply system frequency is 50 Hz and the converter output frequency is 15 Hz, then the lowest frequency present in the converter waveshapes will be at 5 Hz. If the questioner would not allow the term 'steady-state' to be applied until the lowest frequency present has repealed itself in the waveshape, then perhaps we should use die term 'quasisteady-staie' to describe die waveshapes presented in die paper If the computer simulation was allowed to run for sufficient time, 'steady-state' waveshapes would be obtained. The waveshapes produced by the simulation therefore have a define period, and are not aperiodic. We have seen no evidence of chaotic behaviour in our simulations. The possible advantages of using power supply frequencies up to 90 Hz, have been discussed in response to M r Murphy's question. High frequency generation at 400 Hz is common on aircraft, and it has been considered for use on isolated offshore oil and gas installations (see Ref 1 overieaf).This paper shows that the biggest advantage in terms of reduced weight occurs 51 K S Smith, R Yacamini Je A C Williamson when the operating frequency is raised from 50 Hz (or 60 Hz) to 200 Hz. This gives a weight saving of 30%. Beyond this frequency diere is a diminishing return, a frequency of 400 Hz producing a saving of 32%. To the audiors ' knowledge no work has been carried out to investigate the suitability of generation at 200 Hz for frigate applications. We thank Professor Flower for his comments on our 'laudable ambition '. We feel however diat the si mulations of the size and complexity that we are presently carrying out would not have been possible even two or three years ago. Recent improvements in software design and the capability of small 52 computers have made this possible. We also feel that recent advances in data acquisition systems and information technology allow us to dream about simulating the transient behaviour of complete ship propulsion systems. After all, that is what is expected of us who live in ivory towers. Reference 1. R Yacamini. D A Hitchens and J C de Oliveira. 'Weight reduction in offshore electrical power modules by running the system at higher frequencies'. Proc 5th International Symposium on Offshore Engineering. Vol 5. pp 739-752. Federal University of Rio de Janeiro, Brazil (September 1985).