30th Propulsion & Emissions Conference 2008 Radisson SAS Hotel, Sweden, Gothenburg on 21-22 May 2008.

A Survey of Concepts for Electric Propulsion in Conventional and Ice Breaking OSVs
Alf Kre dnanes, Dr. Ing., MScEE ABB AS, Business Unit Marine, P.O. Box 94, NO-1375 Billingstad, Norway

ABSTRACT The use of electric propulsion in both open water and ice breaking OSVs has become a marine industry standard for a wide range of applications, and is increasing to new applications and into more geographical regions. The technologies develop continuously, and today there are several approaches to reach the "optimal design" that reduces fuel consumption and environmental footprint, simplifies design and construction with better utilization of the on-board space, and creates a better working environment for the crew. This paper presents the various concepts at the market, and summarizes their technical characteristics and limitations. It is aimed to give yards, designers, and ship owner necessary technical information in order to make a proper selection of system topology within the specifics of a vessel design. Important aspects regarding operational safety and availability of the propulsion and station keeping plant is also discussed. INTRODUCTION Since mid 1990s, OSVs have been equipped with electric propulsion, Fig. 1, where the main propulsors and station keeping thrusters have been driven by variable speed electric motor drives, being supplied from the common ship electric power plant with constant frequency and voltage. Thrusters and propulsors are normally of fixed pitch propeller design (FPP) that reduces the mechanical complexity of the units, and the electric power is normally supplied from fixed speed combustion engines; diesel, gas, or dual fuel. During this period, there has been a continuous development of solutions for the electric power plant for vessels with electric propulsion. The development can be considered as an incremental evolution of concepts, where the building blocks of the electric plant is adapted from the general industry applications, which has a far bigger volume of installations than the marine applications and to a large extent gives the premises for basic technology developments. Even though the suppliers of electric power and propulsion plants utilize building blocks that are based on principally the same fundamental concepts, there is a range of different configurations and preferences in the market. As the technical arguments for the concept appears to be biased and naturally to some extent influenced by a driving force to pursue a sales, it is necessary for ship owners, yards, and designers to be able to evaluate and compare this information to make the decision. As the author represents one of the vendors, this paper should not be considered to be an objective and neutral assessment of competing concepts; however, it is the aim of the author to give a unified description of the most applied concepts. Further, it is the authors objective that decisions shall be made on understanding of the characteristics and resulting effects of the technology, rather than the technology itself. The author has based his information on own and public available information, and hence, actual solutions may deviate from what has been assumed in this paper. It is therefore necessary to supplement the information herein with the detailed information in each project and from each supplier.

1 of 19

Aux gen. Emgcy gen. 440V, 60Hz

99kVA 99kVA

M
230V Bow Thruster

M
Bow Thruster

230V Distribution

M
Thruster

M
Main Propulsor

M
Azimuth Thruster

M
Main Propulsor

M
Thruster

Port Side propulsion and shaft gen.

Stbd Side propulsion and shaft gen.

Fig. 1: Conventional direct mechanical propulsion, and electric propulsion concept for OSV. THE IMPORTANCE OF TECHNOLOGY DEVELOPMENT For many engineers, researchers, and developers, working with technology gives the daily bread and butter. Technology development is both challenging and interesting for those involved, as well as essential for the continuous improvement of the vessels earnings and safety. In evaluation of concepts, the overall performance and characteristics should be assessed rather than the individual component. The advantages of e.g. an efficiency improvement in one component may make no sense if it requires a non-optimal operation of the prime movers in order to function as intended. Utilizing fast changing technologies to obtain some improvement is normal in consumer industry, but in a life cycle assessment of a vessel, it may be painful if that particular technology is obsolete and unavailable after few years. The driving forces for technology development and assessment should be based on the effects and results of the technologies. Generally; the following criterion will be important for the comparison of products, systems, and services, although their weighting and importance may vary over time and between various applications: Cost efficient building and installation Flexibility in design that improves ship utilization High safety for crew High safety for operations Continuous availability to propulsion and station keeping systems Reduced fuel consumption Reduced impact on the external environment, lower emissions Improved working environment for the crew Low maintenance costs Availability to maintenance during the life cycle of the ship Availability to maintenance in the region of operation, often world-wide Spare parts availability Remote and on-board support Minimizing constraints of operations leading to non-optimal performance Reduce negative consequences for other equipment High ice braking and ice management performance for ice breakers

Technology can never become the goal itself, but those who are leading in technology development and capable to utilize this competence in a sustainable and commercially viable context should be better fit to meet todays and futures solutions that meet the users changing demands. Understanding the needs and requirements of the users requires a continuous cooperation in the global maritime network, and continuous improvements. The optimal solution today does not necessarily meet the new requirements in the future.

2 of 19

VARIABLE SPEED DRIVES FOR ELECTRIC PROPULSION The variable speed drive (VSD) for propulsors and thrusters is one of the most essential components in a power plant for electric propulsion. The VSD consists of: Electric motor, normally asynchronous (induction) motors, but also synchronous motors for the high power range. Other types of motors used in special applications; such as permanent magnet motors and DC motors. Frequency converter, converting the fixed voltage and frequency of the network to a variable voltage and frequency needed to adjust the speed of the electric motor. Optionally line filters or transformers, depending on configuration for reducing the harmonic distortion of currents flowing into the network and voltage adjustment where applicable. A control system, consisting typically of a motor controller and an application controller for the propulsion / thruster control, taking care of the control functions as well as monitoring and protection of the VSD. For the power level needed for OSV propulsion, the Voltage Source Inverter (VSI), Fig. 2, is the dominating topology of frequency converters and used by most suppliers to this market. DC drives, Current Source Inverters (CSI) and Cycloconverters are rarely used and being phased out from new buildings of OSVs. Therefore, this paper only considers the VSI in various configurations. The voltage source inverter consists of a rectifier, a DC link with voltage smoothing capacitors, and an inverter unit as the main components. The DC link may where required be equipped with a breaking chopper to dissipate wind-milling power from the propeller in rapid speed variations or in crash stop conditions of the vessel. The motor controller technology has developed from the scalar U/f control system used since the earliest variable speed AC drives. Field oriented control of AC motors was developed already in the 60s /1/ but did not establish as an industrial standard before in the mid 80s when digital controllers with sufficient capacity and speed became commercially accepted. The introduction of field oriented control significantly improved performance and efficiency of the VSD, however, the control method was still sensitive to the motor parameters and time variations. A new method of vector control that was based on stator flux oriented control and direct torque control of the motor was developed around 1990 /2/, and was first introduced in large scale commercial drives production in the early 90s by ABB under the name of DTC. DTC enables ultra fast control of the torque of the motors, with a robust algorithm that gives high controllability and efficiency with much less sensitivity to variations in motor parameters than the traditional field oriented control. As will be discussed later; the dynamic performance of propulsors and thrusters for OSV is much lower than what the modern VSD may perform, with exception of the need for fast black-out prevention where each fraction of a second is essential.
External interfaces - remote control - automation systems - propulsor and auxiliaries Propulsion Controller Motor Controller Measurement and control signals

Rectifier

DC Link

Inverter

MOTOR

Supply voltage: Fixed AC voltage Fixed frequency

Constant DC voltage

Variable AC voltage and frequency

Fig. 2: The basic modules for a Voltage Source Inverter (VSI).

3 of 19

The key component in the frequency converter is the power semiconductor. Because of the high power, the semiconductors will only operate in the active mode in the transition between being fully off (closed) state and fully on (open) state in order to minimize the power losses. In VSI frequency converters, the components used are either passive components (diodes) or active switching components. Thyristors are only applied in special cases e.g. in configurations where it is a need to control the DC link voltage or for soft charging of the DC link, and in some cases for regeneration of power. Power semiconductors are made for low voltage applications, i.e. system voltages up to 690V between the phases, and for high voltage applications, >1000V system voltage. Depending on configuration, typical voltage levels are 3-3.3kV and 6-6.6kV line-line voltages for high voltage frequency converters. The IGBT (Insulated Gate Bipolar Transistor) is the dominating power semiconductor for low voltage applications and being used by all major suppliers of frequency converters. The low voltage IGBT is normally mounted in compound modules, as shown in Fig. 3(b). Each module may consist of several IGBT switching components, in parallel or separately controlled, together with free-wheeling diodes for reverse currents through the switching elements. For high voltage frequency converters, the GTO (Gate Turn Off) thyristor was for long times used as switching component. In high voltage converters these are used as discrete switching elements with one silicon wafer being installed in press pack (hockey-puck) housing. The GTO is of a robust design, and a highly reliable component itself, although it required a number of auxiliary components to achieve the robustness in operation, and these auxiliary components got quite high stresses from the switching of the GTO. Hence, a further development of the GTO was made by ABB during the 90s where the basic concept of the GTO was improved to reduce the need for auxiliary components, and hence not only increase the overall reliability but also reduce the power losses in the overall system. This new component is called GCT (Gate Controlled Thyristor), and when integrated with its gate control system, IGCT (Integrated GCT). All ABB IGCTs are press-pack devices. They are pressed with a relatively high force onto heat-sinks which also serve as electrical contacts to the power terminals. The IGCT's turn-on/off control unit is an integral element of the component. It only requires an external power supply and its control functions are conveniently accessed through optical fiber connections. The IGCT is optimized for low conduction losses. Its typical turn-on/off switching frequency is in the range of 500 hertz. However, in contrast to the GTO, the upper switching frequency is only limited by operating thermal losses and the system's ability to remove this heat. As the IGBT requires a simpler gate control device, with less power consumption, a large effort has been made to make them available for high voltage. The first IGBTs were of module design, and were not considered reliable enough for demanding propulsion applications. Since IGBTs are made with multiple parallel chips, there is a difficulty - with conventional press-packs - in assuring uniform pressure on all chips; a difficulty which increases with the number of devices in a stack. Today, IGBT in press pack are available from several makers, including ABB /4/, and mainly used in power transmission and static converters, although some makers have eveluated them stable enough to be used in variable speed drives for ship applications.

IGBT Low voltage

IGBT High voltage

GCT/IGCT High voltage

(a)

(b)

(c)

(d)

Fig. 3: (a) The development of power semiconductor switches /3/, (b) low voltage IGBT module, with integrated IGBT switches and freewheeling diodes, (c) press pack high voltage IGBT, and (d) press pack GCT/IGCT.

4 of 19

Fig. 5 shows the most common configurations for low voltage VSI frequency converters. The rectifier may be of different types, depending on the requirements for each installations and makers preference; 6-pulse diode rectifier is the simplest design, with a full bridge passive rectifier or several in parallel if necessary to achieve the desired power level. The AC supply voltage is rectified to form a DC voltage, of approximately 1.35 times supply line voltage at full load; i.e. a 690V supply gives approximately 930V DC link voltage at full load; depending on voltage drop and commutation impedance in the supply. The 6-pulse rectifier does not need any supply transformer unless necessary to adapt the voltage. Hence, size and weight is minimized. However, the harmonic distortion from the line currents is high in the order of 2525% THDi, resulting in a voltage distortion THDu of more than 10%. In order to achieve the limit as specified in IEC and which most classification society now has adapted of 5%, harmonic filtering or clean power supply is necessary. A harmonic filter at the distribution switchboard will reduce the distortion at this voltage level and below, but will hardly have any impact on the distortion at the main switchboard. The main switchboard must hence be designed to tolerate a high level of voltage and current distortion, and be documented accordingly as specified in class rules. 12- and quasi 24-pulse configurations looks similar with two paralleled diode rectifiers; but the quasi-24 pulse (Q24) VSD transformers are made in pairs of two and two with 15 deg phase shift between the two in each pair. Depending on load conditions in the prospected operation profile and system parameters, 12pulse configuration may meet the requirements to voltage distortion by class. For most cases, however, the worst case conditions will lead to a THDu in the range of 6-8%, which is above the limit of most of classification societies without using harmonic filters. The Q24-pulse rectifier utilizes the same rectifier topology as the 12-pulse rectifier, but since the supply voltages are phase shifted through the supply transformers, the resulting distortion at the main switchboard is reduced. At ideal conditions, where the loads of the VSDs in each pair are equal, the harmonic distortion will be equivalent to a 24-pulse configuration. Under other conditions, a partial cancellation of the largest harmonic components will occur, and the harmonic distortion will in most installations be under the 5% THDu limit in any practical operation mode. Certain constraints in operation may be necessary in order to guarantee this; however, the Q24-pulse configuration is regarded to be a cost efficient way to meet class requirements for most OSVs. 24-pulse configurations consists of four paralleled 6-pulse rectifiers, each supplied from phase shifted voltages through one 5-winding transformer, or two paralleled 3-winding transformers. This configuration will normally always give distortion under the 5% limit, without constraints in operation. Normally, the transformer will be larger and more expensive than a 3-winding transformer of equivalent rating, and depending on the rating of the diode rectifiers, also the size and price of this may increase. The 18-pulse configuration utilizes the same concept, but with only three paralleled 6-pulse rectifiers and a 4-winding transformer. The THDu will also here normally be within the 5% limit, however, comparing to the 24-pulse topology, the total prize and size is at the same order due to a complex transformer design. 18-pulse rectifiers are therefore rarely in use today. Active rectifiers with switching elements have for some time been applied in demanding industrial applications, especially where the load characteristics requires regenerative braking to such an extent that it makes it beneficial to use this energy by feeding it back to the network. For propulsors and thrusters, the regenerative energy is normally negligible in a fuel and cost of energy assessment. However, since the rectifier consists of switching devices, the current can be shaped similarly to the motor current and with much lower distortion than the currents of a diode rectifier. Even though the classical harmonic filters then can be avoided without use of drive transformers, one should note that there are ample of harmonic voltages from the switching, with high frequencies and with a high level of electromagnetic noise that must be filtered with high frequency (HF) filters. There is limited experience by use of active rectifiers in weak electric power systems as found on vessels, and in complex systems with many drives and a range of operating conditions, it is challenging to perform a complete system analysis of any modes and configurations in order to detect and avoid possible resonance effects from the numerous combination of paralleled HF filters. As the switching elements are more costly than diode rectifiers, the cost of the frequency converter increases, and its losses will also be higher; reducing the benefit of avoiding losses in the drive transformer.

5 of 19

The DC link, Fig. 5, consists of a DC capacitor, in order to smooth the DC link voltage to reduce the voltage ripple from the rectifier to an acceptable level for the output stage of the frequency converter, and to filter the high frequency distortion from the switching elements in the inverter and avoid that these are injected to the supply network. If regenerative power is expected from the load, e.g. in crash stop conditions of shaft propellers, the power will be absorbed by the DC link causing voltage rise of the capacitors unless the power can be fed into the supply network by an active or regenerative rectifier, or dissipated in a load resistor bank. For thrusters and Azimuthing propulsors, regenerative voltage is seldom of a concern, as one may restrict these loads to only operate in motoring mode. In shaft line propulsion, or in Azimuthing propulsors where crash stop or equivalent to crash stop maneuvering is made by reversal of the propeller RPM, there are certain conditions where the propeller wind-milling effect may create reverse power; see Fig 4. The inverter module is normally a full bridge IGBT inverter in low voltage VSIs, Fig 5, with 6 IGBT switching elements each with an anti-parallel freewheeling diode for reversing the currents through the switches. The switching elements in each of the three legs of the inverter must operate in inverse mode, where one IGBT always is controlled off to avoid short circuiting the DC link. The objective of controlling the IGBTs is to feed a voltage vector to the stator winding that forces the currents and flux in the machine towards the targeted amplitude and phase angle. Depending on the control scheme, there are different ways to achieve this, such as field oriented control with PWM or hysteresis control, and direct torque control. The characteristics of these methods will be discussed later. High voltage frequency converters, e.g. for 3-3.3kV system voltage, utilizes the same principles as the low voltage VSI frequency converters. Because of the higher system voltage, and maximum voltage availability of power semiconductors, the high voltage VSI frequency converter normally consists of series connected semiconductors, as shown in Fig. 6 for the rectifier and inverter. In order to reduce the voltage over each component, the converters will normally be equipped with a third voltage level, at the middle of the negative and positive DC bus voltage, the so called neutral point. Clamping diodes ensures that each of the switching elements never will be exposed to higher than half of the DC link voltage, and therefore this configuration is called three level, neutral point clamped (NPC) inverter topology. By increasing the number of series connected switching elements, the number of switching combinations also increases and with a specific maximum switching frequency of the components, the harmonic distortion in the load currents and consequentially the torque ripple will be reduced. Some suppliers use this concept to design multilevel inverter topologies, where the ripple current can be further reduced and lower voltage switching elements may be used. The penalty is then the higher number of components in the system, reducing its availability and reliability.
Torque
Quadrant II Braking Speed<0, Torque >0 Quadrant I Motoring Speed>0, Torque >0

Quadrants I and III Positive Power:

Bollard pull Free sailing Crash stop

Speed (RPM)

Quadrants II and IV Negative Power: Speed

Quadrant III Motoring Speed<0, Torque <0

Quadrant IV Braking Speed>0, Torque <0

Fig. 4: Regenerative power from the propeller may occur when reversal of speed (RPM) is used in crash stop or equivalent to crash stop maneuvering, such as with shaft line propellers. Braking resistors may then be used to dissipate the regenerated energy, which normally is insignificant from an energy consumption perspective in electric propulsion.

6 of 19

Rectifier 6-pulse, passive diode rectifier

DC Link

Inverter

12-pulse, passive diode rectifier

30 electrical degrees phase shifted voltages

DC link with smoothing capacitor

18-pulse, passive diode rectifier

24-pulse, passive diode rectifier

20 electrical degrees phase shifted voltages

Full bridge IGBT inverter

15 electrical degrees phase shifted voltages

DC link with braking chopper and resistor

Active rectifier with IGBTs

Fig. 5: Description of basic modules for typical low voltage, two level VSI frequency converters,

7 of 19

Rectifier 6-pulse, passive diode rectifier

DC Link

Inverter

DC link with smoothing capacitor 12-pulse, passive diode rectifier

30 electrical degrees phase shifted voltages

Neutral point clamped IGCT inverter

24-pulse, passive diode rectifier

15 electrical degrees phase shifted voltages

DC link with braking resistors

Active rectifier with IGCT

R

Fig. 6: Description of basic modules for typical high voltage, three-level neutral point clamped VSI frequency converters.

8 of 19

P (kW) 40000 30000 27000

CSI Cycloconverters
VSI IGCT VSI IGBT
690 1.0 1.5 1.8 2.4 3.3 4.5 6.0 6.9 U (kV)

Motor Power

16000 10000 9000 6000 5000 2000 315

VSI: Voltage Source Inverters with PWM or DTC CSI: Current Source Inverters with Thyristors Cyclo: Direct Converter with Thyristors

Motor Voltage

LV

MV

Fig. 7: ABB drive technology map (indicative check with ABB for details). Fig. 7 shows the drive technology map for ABB. The map is indicative, and for illustration only. For marine applications, the VSI concept is today the preferred solution within the constraints of its ratings and applications. Single inverter, low voltage VSI frequency converters are available up to approximately 4.5-5MW as a practical limit. For higher power levels, two low voltage motor drives can be used in twin or tandem configurations, or the frequency converters should be of high voltage topology. For other industrial applications, or for special customer demands, Cycloconverter and CSI (Current Source Inverter) topologies may be used. However, for the OSV applications, these concepts are very rarely used today since the VSI frequency converters covers any application needs, and offer more technical advantages than the alternative VSD concepts. MOTOR CONTROL METHODS Any motor control method is based on the principle of controlling the motor torque based on a model of the motor. For the OSV applications, normally induction (asynchronous) motors are applied. Induction motors are simple of construction and rugged in design, and are normally used for power levels up to 5-10MW depending on motor RPM. This limit is based on the constraints of motor design and construction, where the air gap of the motor becomes so large that the amount of magnetically stored energy, which is represented by the reactive power of the motor, becomes excessive and impractical for cost efficient use. For higher power levels, normally synchronous motors with wound field windings and brushless excitation are used. For special applications, permanent magnet AC motors, and even DC motors are applied. The power range of most of the OSV propulsors and thrusters is covered by the induction motor. This motor can simplified, and stationary be modeled by the equivalent diagram shown in Fig. 8. This shows a one phase model of the induction motor, with the stator resistance and leakage inductance, rotor resistance and leakage inductance, and the magnetizing inductance for air gap flux. The magnetizing resistor represents the magnetizing losses in the stator and rotor magnetic steel. From this model, it can be derived that the voltages and frequency should vary as shown in Fig. 8b when controlling the speed of the motor. This relation has been known since the origin of the variable speed control, and was utilized in scalar control, Fig. 9, of the induction motor before more modern concepts was developed.

9 of 19

Rs Ls

* L* r Rr / s Lm

Is
Vs

Rm

Ir Im Torque producing current (Rotor current) Re Ir Im Is

T 2 2 Is = Im + I r2 = I m + Ir T N
2

Constant torque region

Field weakening region

Im

Maximum stator current boundary

Maximum torque boundary

Flux () producing current (Magnetization current, Im)

Magnetic flux

Pitching moment limitation

Stator voltage

Stator frequency

RPM

(a) (b) Fig. 8: (a) Simplified equivalent diagram and vector diagram for induction motor, and (b) the corresponding voltage, frequency, flux, and torque characteristics in variable speed drive applications.
DC Link RPM Setpoint + Ramp - limit U

PID controller
Ramp - limit U/f ratio f

f0

10

15

20

Inverter
PWM

M T
Fig. 9: Principle sketch of a scalar controlled VSD with induction motor. The scalar control of the induction motor is based on the stationary motor model, and since the dynamics of the motor is neglected, significant information is disregarded, giving a poor dynamic control of the motor. During the 60s, significant improvements in the motor modeling and control of induction motor was achieved, /X/, and this enabled control methods that were comparable to DC motor control. The principle was to decouple stator current into a magnetizing component, and a rotor current component and control each of these quantities in a synchronous rotating frame, oriented along the vector of the rotor flux. The new control algorithms were called field oriented control, or rotor flux oriented control, Fig. 10. As these control methods required digital controllers for optimal performance, they did not become commercially available until the mid 80s. Still, the field oriented control suffered from two major weaknesses: The motor model is very depending on the rotor time constant, which is time variant and since it cannot be directly measured, it must be continuously estimated in order to obtain satisfactory performance and efficiency of the converter. The field oriented controller requires a closed loop control of the two current components, and the speed and dynamics of the outer loop control is limited by the bandwidth of the two current control loops. During the 80s, another approach to modeling and control was introduced, /2/. Instead of referring the model to the rotor flux vector, the motor variables were now referred to stator flux. Also, instead of controlling the two decoupled current components, advanced algorithms for directly controlling the switching pattern of the inverter of the VSI were developed. This resulted in a very robust algorithm, not sensitive to the highly variable rotor time constant, and with ultra fast dynamic response, only limited by the current rise time of the stator windings. The Direct Torque Control (DTC), Fig. 11, was the first large scale industrial use of this control method, and has been applied by ABB since the beginning of the 90s , made possible only by using dedicated, high speed ASICS in combination with fast signal processors.

10 of 19

DC Link RPM Setpoint + Ramp - limit

PID controller

PID controller e-j

f0

10

15

20

2
PWM

Inverter

Flux controller

PID controller

M T

e j

2 3

Motor model

Parameters

Parameter adaption

Fig. 10: Field oriented, or rotor flux oriented control of induction motors.
DC Link RPM Setpoint + Ramp - limit
Flux ref

PID controller

Torque ref

Direct torque and flux hysteresis control

Inverter Optimal switching logics

Flux controller

Torque act RPM act

Flux act

Motor model

Fig. 11: Direct Torque Control (DTC) of induction motors minimizes at each sampling the deviation between desired and actual flux and torque in the motor. Here shown induction motor, it is also applied in synchronous and permanent magnet motors. For VSD in propulsion and thruster applications, the dynamic performance of the modern frequency converters is much faster than needed for normal speed control. Typically, the speed regulation must be slowed down by use of speed ramps in order to not overload the mechanical systems in the thruster and propulsor power transmission, and to avoid excessive load variations in the prime movers. However, there are certain concerns that needs fast acting and precise control algorithms to be solved; Black-out prevention: In order to avoid black-out by power phase back of the VSD, the load power of the propulsors and thrusters should be reduced within 300-500 ms, Fig. 12, after a fault in the power generation system. This can only be achieved with the fastest dynamic performance in order to avoid load shedding. Out of water effects: When thrusters and propulsors are running at high load in rough weather conditions, air suction and out of water effects lead to sudden loss of load torque with a high risk for tripping by over speed if the dynamics of the controller is not fast enough to avoid runaway. Power efficiency: The minimization of power losses in the motor is based on a precise control of the magnetization and torque producing currents. Errors in the model may give unnecessary high losses in the motor.

11 of 19

REMOTE I/O CABINET

ENGINE SAFETY & SWITCHBOARD/ CONTROL SYSTEM INTERFACE GENERATOR PANEL ENGINE CONTROL - HARDWIRED & SAFETY SYSTEM

REMOTE I/O CABINET

10 sec 1 sec 0.1 sec Risk for Blackout Fixed speed CPP load red DP power limitation PMS load reduction

DIESEL GENERATOR

COMMUNICATION NETWORK

FREQUENCY CONVERTER

MAIN CONTROLLER COMMUNICATION NETWORK

OPERATOR WORKSTATION COMMUNICATION NETWORK

0.01 sec 100% 200% 300%

VSI Drives FPP load red Load x MCR

PROPULSOR OR THRUSTER

Fig. 12: Black-out prevention algorithms requires very fast dynamic response of the VSD controller for thrusters and propulsors. 300-500 ms is normally regarded as maximum response time for the complete load reduction loop including detection, transmission, and load control times. PROPULSION CONTROL METHODS The essence of any motor control is to achieve a good torque control of the motor drive. When torque control is achieved, the motor controller can be regarded as a torque amplifier with a time lag, and the time lag is depending on the applied motor control algorithms, and the motors own electrical dynamics. Then, it is easy to adapt the motor drive to various types of outer control algorithms, and the most normal for variable speed control, is to create a closed loop speed (RPM) control of the propulsor or thruster drives, where the set-point from the remote control system is regarded as an RPM set-point, as shown in Figs. 9-11. However, one could also use other approaches to control the propeller, by interpreting the remote control signal as directly being the torque reference to the torque controller of the VSD, amplified by the motor control algorithm, and even to use the set-point as a power command to the VSD, controlling the torque to keep the shaft power equal to the remote set-point. RPM control, torque control, and power control have all been applied in various applications. Their differences can be observed by the characteristic behaviors in Fig. 13. Since a closed loop RPM control uses the PID controller to minimize the deviation between reference speed and actual speed, load variations at the propeller will instantly result in torque and hence power variations at the shaft. These power variations will immediately be reflected on the power taken from the power plant, and result in frequency variations and possible dynamic overload of the prime mover if the load variations are high. This is primarily a problem of concern for ice breaking OSVs, where load variations are sudden and may be very high, but also for ships operating in high waves the load may vary quite significantly. If using the remote set-point as a torque reference, the load torque variations will not be reflected in the torque produced by the VSD, which is controlled to follow the set-point only. As seen in Fig. 13, this may give large variations in the propeller RPM and will therefore also give power variations at the shaft and hence in the power plant, since the load power is proportional to load torque and RPM.
Torque
Propeller load torque
0,7

Speed (RPM) Control

rp m

Torque Control

Power Control

Max torque limit Max power limit Torque Control Power Control RPM Control

0,6 0,5 0,4 0,3 0,2 0,1 0 0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10 load to rqu e el torque po w er

RPM

(a)

(b)

Fig. 13: Characteristics of RPM (speed) control, torque control, and power control of a propulsor or thruster, (a) in RPM-torque frame; (b) in time frame.

12 of 19

The speed is not constant in Power Mode. At this point the ship is turning and due to increased propeller load, the speed is decreasing. The torque increases inverse proportionally to speed in order to keep constant power; Power = Torque x (2 x x RPM/60).

Fig. 14: Logging from sea trial, showing that the load power is constant when operating in power control mode, even though the load torque may change significantly by effects from waves and course changes. Power control is based on a control algorithm where the remote set-point is regarded to be the commanded power (kW) set-point to the propeller, and the VSD controller will then control the load torque to produce a power to the motor shaft which equals this set-point. Any load variations will be counteracted by keeping the motor shaft power constant, as seen in Fig. 13, and the load power of the prime movers will therefore be minimized. The power control works well within the constraints and limitations of the components in the VSD, as shown in Fig. 14; which is a real logged characteristics of a propulsion drive system in heavy seas and during course changes where the load varies significantly without causing load power variations of the prime movers. Power control is found advantageous at high propeller loads in most ship applications, such as free sailing and transit modes. RPM control gives some faster dynamic performance, and is normally used in operation modes with lower load on the propellers, and at high dynamic requirements, such as maneuvering mode and station keeping. For ice braking vessels, power control is a necessity, since the loads of the propeller may vary excessively and very fast, e.g. when the propeller hits ice blocks. Torque control is rarely used today, as speed (RPM) and power control are better alternatives for the various operational modes. Research has shown that the use of power control may also give significant achievements in power stability of the network and reduction in load variations for the prime movers in station keeping conditions. Until today, there has not been reported any successful use of such control schemes in DP systems, and the slower response of the VSD drive in power control mode may be one explanation to this. Note that it is necessary to include additional control and monitoring functions to achieve safe and reliable control of the propulsors and thrusters in all operating modes, however, it is beyond the scope of this paper to describe all these in details.

13 of 19

SYSTEM CONFIGURATIONS As previously shown, the basic topologies for the VSD are relatively similar among the various suppliers to the application of electrical propulsion. From a ship application perspective, one of the main technical differences are related to how these products are put together in a system configuration for electric power generation, distribution, and propulsion / station keeping. Several system configurations are applied, of which the most common ones are shown in Fig. 15. For each main configuration, there may be several variants for optimization to the actual requirements for each vessel. The main challenge in system design is to meet the class requirements and ship specific requirements at a minimum total cost including equipment and installation costs, and with a best possible life cycle economy. Each vessel may have its own specific requirements, e.g. whether space is a scarce resource or not in the design. Propulsion transformers are large and heavy equipment, and the 6-pulse, active rectifier, and the Q12-pulse with phase shifted main voltages are examples on transformer-less solutions, but not without other penalties. The 6-pulse and Q12-pulse solutions normally will require some kind of harmonic filter installations, unless significant restrictions and constraints of operations shall be applied, which may cause deterioration of the fuel economy of the prime movers and limitations of operational windows for the vessel. The Q12-pulse solution in particular depends on a complex main switchboard with two feeders to each frequency converter for balanced loads that is a necessity for maximum performance. Each feeder will carry a 6-pulse current that to some extent will enter the respective generators on the switchboard, or flow through the two primary sides of the distribution transformers, and give additional losses that to some extent will counteract the benefits of avoiding the losses in the drives transformer. The active rectifiers increase the number of active components in the installation, and the complexity of the installation as each of the rectifiers requires a HF harmonic filter that introduces resonance modes of the installations that should not be excited by the switching frequency of the rectifier. Also, the size and costs of the frequency converter itself will increase, as well as the power losses in the rectifier, counteracting at least partly the benefits of the transformer-less design. Hence, there exists no one ideal design for all vessels. The different solutions have different characteristics, and only when considering the requirements and limitations for a vessel design, the best solution can be applied. Note, that in the configurations of Fig. 15, there are no thrusters with change-over feeder from the two switchboards or switchboard sections. Change-over supply is often used in OSVs for retractable Azimuthing thrusters in order to be able to provide power to this even with a fatal failure of one of the main switchboards. Fig. 16 shows typical solutions for such change-over arrangements for different frequency converter topologies. The use of change-over circuits is an issue for discussion with class. Although class rules allows for changeover supply, class does not necessarily allows thrusters with change-over to constitute a part of the station keeping capability after a single failure, if the power source must be automatically or manually reconnected after the failure. For the configurations in Fig. 16 (b) and (c) this is necessary to avoid paralleling of the bus tie or transfer feeder via the thruster cables. The configurations in Fig. 16 (a) and (d) could meet the requirement for continuous availability provided that each branch of the rectifier parts is rated appropriately. Dual feeding through diode rectifiers normally will lead to a 2x100% rating of the rectifiers, as the rectifier loads are given mainly by the total load and the difference of the supply voltages. Even a small difference in voltage supply is a source for large unbalance in the rectifier loads. Also the dual feeding of active rectifiers must be rated 2x100% if full load capability shall be obtained after a worst case fatal fault of one switchboard section. In addition to the rating, there are also other complicating effects for such designs, as also all auxiliary and control voltages must be tolerant and continuously available after a fault on any of the main switchboard, and likewise for the critical signal interfaces to the power management system and other automation systems. This fault tolerance may be achieved, but it is not necessarily straight forward and must be discussed with, and approved by class in each case. However, if the result is that e.g. a retractable thruster can be considered as available for the DP capability after a single fault, it may be worth the efforts.

14 of 19

G 690V

FC

FC

FC

FC

440V

440V

6-pulse: No drive transformers Harmonic filters needed to get THD<5% Weight: Low Footprint: Low Operational constraints: Medium Total efficiency: Approx: 90-91% 12- and quasi 24-pulse 3-winding transformers, phase shift for Q24 Harmonic filters for 12-pulse, not Q24 Weight: High Footprint: High Operational constraints: Low/medium Total efficiency: Approx: 90% Quasi 12-pulse with phase shifted mains voltages /5/: No drive transformers, oversized distribution transformers for power transfer Weight: Medium Footprint: Medium Operational constraints: High Total efficiency: Approx: 90% included harmonic losses in generators and distribution transformer 24-pulse: 5-winding transformers (or 2 x 3-winding) No harmonic filters Weight: High Footprint: High Operational constraints: Low Total efficiency: Approx: 90% Active rectifiers: No drive transformers High frequency input filters for harmonics Weight: Low Footprint: Medium Operational constraints: Low / Medium Total efficiency: Approx: 90-91% 690V: Main switchboard voltage 440V: Main distribution voltage G: Generator M: Motor (Propulsors and thrusters) FC: Frequency Converter AR: Active Rectifier DC/AC: Inverter

G 690V

440V

FC

FC

FC

FC

440V

M
G 690V

M
G

M
G

M
G 690V

440V

440V

G 690V

440V

FC

FC

FC

FC

440V

G 690V

AR DC/AC

AR DC/AC

AR DC/AC

AR DC/AC

440V

440V

Glossary:

Fig. 15: Alternative system configurations with main characteristics. 690V Main SWBD voltage is shown, high voltage, e.g. 6.6kV is used when generator capacity typically exceeds about 10MW.

15 of 19

AR INV

AR

INV

FC

FC

M
6-pulse 12- / Q24-pulse

M
24-pulse Active rectifier DP 3 Change-over

(a)

(b)

(c)

(d)

(e)

Fig. 16: (a)-(d) Alternative for thrusters with change-over from two switchboard sections. Note, that in DP3 applications, the change-over circuits must be equipped with isolating switches also at the load end of the feeding cables (e). HARMONIC FILTERNING AND CLEAN POWER SUPPLIES Frequency converters are inherently non-linear components due to the switching characteristics of the rectifier components, meaning that they do not draw sinusoidal currents from the network, even though being fed by sinusoidal voltages. The non-sinusoidal currents into the rectifier consist of a fundamental voltage, and a series of harmonic components with a wide content of frequencies which depends on the rectifier type and system configuration. For the type of converters that have the highest level of distortion in the currents, typically those with 6-pulse, 12-pulse, and Q12-pulse rectifiers, the level of harmonic distortion in the currents may lead to voltage distortion above the class limits. Most class societies now have adapted the IEC 60092-101 requirement, such as ABS, American Bureau of Shipping:

When the limit of the applicable regulation will be exceeded with the decided frequency converter and system configuration also after optimizing the design of the generators and transformers in the plant, there are still several ways to manage the harmonic distortion level. A harmonic filter can be applied. There are two main different types for harmonic filters; passive LC filters (alternatively damped LCR), Fig. 17 and active filters, Fig. 18. For ship applications, passive filters are more commonly applied, due to their lower costs; especially since they can be used at lower voltage levels in the distribution system to filter the voltage distortion not necessarily in the complete installation, but for the sensitive equipment only. Active filters are, similarly to the active rectifier in frequency converters, a topology with switching elements like IGBT for low voltage systems. However, while the active rectifier is designed to draw sinusoidal currents from the network, the active filters measures the distorted load currents, and cancels the harmonic components of the VSD by injecting harmonic current with 180 deg phase shift to the power plant. Both methods are efficient filters; it is normally a matter of economy to select which solution to apply.

16 of 19

One stage passive LC filter; un-damped

Z()

Two stage passive LC filter; un-damped

Z()

Lg

Lg

Aggregated Generator / motor model

First order undamped LC filter

Aggregated Generator / motor model

First order 11th harm undamped LC filter

First order 5th harm undamped LC filter

Frequency scan:
Z
Filter

Frequency scan:
Z
Filter

Generator 2 . . f . Lg

Generator 2 . . f . Lg Resulting

Resulting

11 13

23 25

f / f1

11 13

23 25

f / f1

Fig. 17: Passive LC filters, left: one stage for filtering of one (5th) harmonic frequency; right: a two stage filter, here for filtering 5th and 11th harmonic components. Introducing capacitive components in an inherent inductive network, as done with passive filters, must be done with care as it introduces not only the desired series resonance to achieve low impedance current paths for the harmonic currents of concern, but also parallel resonances that may cause excessive distortion globally or locally in the installation if they are excited by harmonic currents. Therefore, when including harmonic passive filters, an extensive system analysis must be made to ensure that such parallel resonances that always will exist, are not excited. Active filters are less vulnerable to such, however, as for the active rectifiers, they also need HF harmonic filters in their feeders to avoid high frequency harmonics and common mode noise to distort the network.
ig i1
11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87

i2

ig = i1+i2

i1
Nonlinear load

i2
Active filter

Fig. 18: An active rectifier can measure the non-linear current and compensate instantaneous its harmonic distortion. Some filters need one or more periods to calculate the harmonic components that shall be compensated.

89

17 of 19

Dirty side
M G

Clean side

Rectifier (AC to DC)

Inverter (DC to AC)

Rectifier (AC to DC)

Inverter (DC to AC)

Fig. 19: Clean power supplies; from top: Rotating motor-generator set; Static frequency converter; Static frequency converter with battery back-up (UPS) Earlier, it was quite common to install clean power supply, where the distorted network was decoupled from the supply to sensitive equipment by use of either rotating or static converters, as shown in Fig. 19. With exception of the UPS supply, which is class requirements for a range of navigation and control equipment, the clean power supplies are less used today when class rules normally requires the general power distribution system to fulfill stringent requirements to harmonic distortion, unlike what was common practice only a few years ago. POWER PLANT CONTROL, ACTIVE AND REACTIVE LOAD SHARING Vessels with electric propulsion will normally have a relatively advanced power management system (PMS) to ensure optimal operation and reliable access to propulsion power. For DP vessels with class 2 or 3 notations, a PMS is also required by class, and should then also be redundant in accordance with the redundancy requirements of the class notation. There are basically two types of PMS configurations applied to OSVs; the mirror image configuration, where the PMS configuration reflects the power system design with one control system responsible to interface and control each of the redundancy segregated parts of the power system; and the duty-standby system where one PMS control system controls the complete plant, with a second control system in hot stand-by to take over after failure of the duty controls. Both configurations fulfil the class requirements, as long as functionality and fault integrity is met. As a general comment, the more standard off the shelf systems tend to be of the dutystandby design, with less and simpler functionalities than the more advanced, engineered solutions. Related to the design of the electric power plant, load sharing is an important function. Both active power (kW) loads and reactive power (kVAr) loads shall be equally shared between paralleled generator sets, according to class. The active power load sharing is either passive by droop control (Fig. 20) or active isochronous, by communication between the governor controllers to enable a regulation towards the average load of the generator sets. Passive load sharing has the benefit of not having a common communication line, with less risk for common failures in the system compared to the isochronous load sharing. However, the frequency adjustment is slower, with larger frequency variations for load changes. Especially for OSVs, where the propulsion load may vary much in DP operations and in heavy sea conditions, this may cause problems for synchronization of new generators to the system and increase the risk for excessive frequency variations and entering into undesired load reduction conditions. Therefore, isochronous load sharing is often used in OSVs, however, special considerations on the fault tolerance to avoid single point of failure is then required.

18 of 19

Frequency / RPM
Reference RPM +2.5% Nominal RPM

Line-voltage
Engine 1 Equal droop Engine 2
Reference Voltage +2.5% Nominal Voltage

Generator 1 Equal droop Generator 2

-2.5%
Paralleling - Equal fuel supply - Active load sharing

-2.5%
Paralleling - Equal magnetization - Reactive load sharing

100%

100%

Active Power Load

Reactive Power Load

(a) (b) Fig 20: Droop control of (a) governor and (b) automatic voltage regulator, showing load sharing of two paralleled generator sets. The same principles apply also for the sharing of reactive power among the paralleled generators; however, droop control mode, Fig. 20 (b), does normally not give voltage variations that are large enough to create problems for synchronization and protection of the plant. Use of cross compensation, and active equalization of reactive loads are available technologies in voltage regulators, but only exceptionally used in ship applications since such solutions requires common signal interfacing and hence inherently increase the risk for single point failures without otherwise contributing to a more safe and reliable system.

CONCLUDING REMARKS In design of the electric plant for OSVs with electric propulsion, it may appear that there are large variations in the basic technologies being offered by the different makers. However, this is not necessarily the case. Most vendors utilize similar basic components and solutions, even though the composition of systems and preferences in design may differ to some extent often depending on the makers available technologies. In order to understand and assess the technical argumentation for the different solutions, it is the authors opinion that rather than preferring a specific technology above another, the important issues of the effects of the technology on safety, performance, and life cycle economy should be considered. This paper has presented the most commonly applied solutions in electric propulsion with the objective to give the decision makers background information to better understand the concepts and to make the most beneficial selection for the specific vessels requirements. REFERENCES 1.F. Blaschke, "Das Verfahren der Feldorientierung zur Regelung der Asynchronmaschine" [The Field Orientation Method for Controlling an Asynchronous Machine], Siemens Forschungs und Entwicklungsberichten [Siemens research and development reports] 1972, pp. 184 et seq.*. 2. Depenbrock, M., Direct self control for high dynamics performance of inverter feed a.c. machines. ETZ Arch. v7 i7. 211-218. 3. Presentation: http://www.spec.ncsu.edu/selected%20presentations/Electric-power%202005-Talk.pdf 4. Simon Eicher, Munaf Rahimo, Evgeny Tsyplakov, Daniel Schneider, Arnost Kopta, Ulrich Schlapbach, Eric Carroll, 4.5kV Press Pack IGBT Designed for Ruggedness and Reliability; ABB Switzerland Ltd, Semiconductors, CH-5600 Lenzburg, Switzerland, IAS 2004, Seattle October 3-7. 5.Product information: http://www.akerkvaerner.com/Internet/IndustriesAndServices/OilAndGas/PowerandAutomation/AKPASElectri cPropulsion.htm

19 of 19