The practical application of modern
simulation tools throughout the design
and trials of a diesel electric
propulsion system
BACKGROUND
In November 1991, the authors' company was awarded a contract by Ferguson Shipbuilders to design and supply a
full Diesel Electric Propulsion System for a 2000t Lighthouse
and Buoy Tender Vessel. The vessel, MV Pharos, is now in service in Scottish waters with the Northern Lighthouse Board, having been handed over on 6th May 1993.
Based on their experience with naval propulsion and electrical power and control systems, the authors' company produced a system incorporating significant innovation. The use of modern simulation tools was essential in order to achieve the necessary level of confidence in the design,
and to deliver a fully working system within short
timescales.
INTRODUCTION
The design process for naval vessel propulsion systems has frequently used simulation techniques for two decades. The
P H Sallabank and A J Whitehead
Vos per Thorn ycroft Controls, Portsmouth
This paper describes the application of modern simulation tools in the design, setting-to-work, commissioning and operation of major shipboard systems. To illustrate the
tech-niques used, a full Diesel Electrical Propulsion System is used as the case study, tracing the evolution from conceptual design through to service operation, analysis of trials results and final model validation. Past experience with simulation and the efficiency of modern tools
enabled a variety of innovative features to be tested at the design stage, and to be adopted
with confidence in the chosen system configuration. Simulation results and trials results are
presented, with a particular focus on harmonic control, power management performance and transient stability of the system. An example of simulation as a through-life vessel support tool is presented, demonstrating the development, installation and validation of system enhancements in response to specific owner requirements.
Trans IMarE, Vol 107, Part 2, pp 101-117
demand arose from the requirements of automatic remote control, the application of gas turbines and cp propellers with their associated torque/speed interactions, and the prospect of improved manoeuvrability with these configu-rations. In parallel with such marine engineering develop-ments, computing power rose dramatically from the ana-logue machines used in the 1970s, through hybrid, to the digital power of today.
Modern simulation tools have transformed the simula-tion task from a mixed applicasimula-tion engineering/computing technology effort into an almost entirely applications ori-ented task. The computing technology has become virtually transparent.
The target application chosen to illustrate this point is a ship's diesel electric propulsion system. In simulation terms it is a high bandwidth system, with significant time con-stants ranging from a few milliseconds to several minutes. The system elements are highly interactive and the particu-lar project demanded some significant control system devel-opment. Additionally, such systems can be time consuming to commission and an accurate design simulation was seen as potentially reducing tuning and sea trials time.
Authors' biographies
P Sallabank joined Vosper Thomycroft Shipbuilding Division in 1971 as an electrical engineer working on naval vessel power system
design, machinery control and instrumentation. In 1975 he joined the Controls Division team engaged in the development of propulsion system and power system simulation. From 1991 to mid-1994 he was manager of the Marine SystemsDepartment taking responsibility
for commercial marine electrical and control systems, including that for the Northern Lighthouse Board vessel MV Pharos. He has recently
transferred to the Support Projects Division where he is involved in market testing.
A Whitehead graduated from Imperial College, London, in 1985 with a BEng in Electrical Engineering. He continued in employment with his sponsoring organisation, GEC Traction, and was responsible for the electricaldesign, manufacture and commissioning of the British
Railways' Class 91 140 mph locomotive. He moved to Vosper Thomycroft Controls in 1991 as Principal Electrical Engineer and was
responsible for the electrical system design, specification, procurement and commissioning of the diesel electric propulsion system on MV Pharos.
P H Sallabank & A J Whitehead AVK ALTERNATORS ?SO kW IMAM AL TErtNATOR RA kW AUX ALTERNATOR 7S0 WI MAIN ALTERNATOR 750 kW MAIN ALTERNATOR NH kW AU% ALTERNATOR
0-
C-
0-
C-FATATNSWIT6H130A19D-1 .0 VOLTSThis paper outlines both the system studied and the
simulation tools used. Aspects of design where simulation has had the greatest impact are addressed in some detail.
Actual trials results are presented and a comparison with simulation results is discussed. The importance of good trials recording cannot be over-stressed. Without it, the quality and accuracy of simulation models cannot be meas-ured and, most certainly, cannot be improved.
Programme timescales are given and discussed. Whilst not of scientific interest, they do provide an insight into the speed with which complex models can now be built, and also show the benefit to overall project timescales in terms of reduced tuning and trials time.
The experiences of 18 months in-service operation are discussed, providing an example of the application of simu-lation as a through-life support tool to vessel and owner.
CASE STUDY SYSTEM
In order to demonstrate the simulation work done, it is necessary to outline the system to be simulated and the particular areas of design requiring detailed study.
The diesel electric propulsion system to be studied is for a Lighthouse and Buoy Tender Vessel. This is a multi-role all-weather vessel required to carry out both routine and emergency tasks around the Northern UK coast. Typical operations include replenishment, hydrographic survey, buoy handling, towing, wreck location and helicopter op-erations. The requirement is therefore for a high integrity system giving good manoeuvrability, hovering capability, and good fuel economy across several different operational modes and speeds.
Preliminary design work established the basic configura-tion shown in Fig 1.
Key design tasks
The key design tasks that either necessitated simulation or would benefit from simulation were identified as follows:
ANNA HORN/
Tmnsloprwp 800018630 TNYRISTOR TO RYA CONTTiOLLE11 (A.)
C8----*---i'
/ Mr Phaso-ShNIng Tau...former 880V THYRISTOR 133.0 ER Ph.* SPRINg 7.NNINme. 880v. gV INYRISTOR 1 38 VA CONTROLLER V OWFig 1 Power generation single-line diagram
Ship/propulsion performance
steady state powering and vessel performance; manoeuvring (for different ship operating modes and various system configurations);
failure modes. Electrical power system
quality of power supplies; load flow analysis; fault level;
transient response; harmonic penetration;
supply system/load interaction; protection co-ordination; failure modes;
operating efficiency and power factor. Control systems
power management; shaft speed control; failure modes.
Simulation output requirements
The required output of these simulation based design tasks
is as follows:
(1,)
Ship/propulsion performance
steady state
Against a base of ship's speed in various hull states and system configurations, the following are required:
ISM AWOL MOTOR
BOW THRUSTER
HEY MAIN SWITCHBOARD
AND GROUP STARTER BOARD
(11) 950 I.W DC MOTOR PROPULSION
NO NW DC MOTOR PROPULSION
MAW SWITCHBOARD ANO GROUP STARTER BOARD
T. 4 5 36. 7 8 .9.. H157
shaft/motor speed; shaft/motor torque;
motor input power; converter output power;
converter input kVA, kW, kVAR, power factor;
efficiency.
Ship/propulsion performance manoeuvring
Plots of ship, machinery and electrical system parameters for a range of ship manoeuvres are required. These manoeu-vres include both classical crash stop and slam accelerations, as well as slow speed manoeuvring, hovering, and unusual or abnormal configurations such as single shaft operation.
The output plots provide information enabling control development to improve manoeuvrability, ensure machin-ery remains within safe limits, and prove security of power
supply, particularly under various failure modes.
Electrical power system
The quality of power supplies determination to provide a common specification to all suppliers of electrical equipment on the vessel.
The load flow under different operating modes and configurations to determine voltage drops, and to vali-date cable ratings; also to identify Power Management System (PMS) switching thresholds and governor and Automatic Voltage Regulator (AVR) droop settings. Making/breaking fault currents at different points in the distribution network to enable correct and economic selection of circuit breakers and protection devices. The transient response of system to motor starts, trans-former magnetisation and faults such as generator trips. Resultant determination of machine reactances, gover-nor and AVR gains and control loop stability to ensure compliance with specified performance and Lloyd's Register requirements.
Harmonic penetration, resulting from thyristor con-verter harmonic currents under different configura-tions and loads, to determine machine reactances and
identify prospective power system resonances. Supply system/load interaction to establish any load control requirements to prevent propulsion demand overloading connected generator capacity.
Protection co-ordination to determine circuit breaker short time and long time trip settings and discrimina-tion proved under all valid configuradiscrimina-tions.
System tolerance of failures is particularly important in a diesel electric ship due to the potential for blackout and consequent threat to ship safety. Output in this study area covers the results of control failures, as well as power system failures, including diesel generator trips.
Operating efficiency and power factor under transient and steady state operating conditions to validate gen-erator selection and determine the need, or otherwise, for power factor correction equipment.
Control system
The studies identified above create a database and knowl-edge of inherent system performance characteristics from which the intended control strategy can be both tested and developed in detail.
PMS studies enable the control logic, switching thresh-olds, generator loading/unloading rates and tuning margins to be established and proved throughout the range of operating modes.
Shaft speed control encompasses all the logic associated with bridge and machinery control room lever control of propulsion. Motor control loop (current, speed and power) settings are established and the interactive pro-tection system executed by the motor control and PMS is demonstrated.
Demonstration of satisfactory failure modes is on the full system, to prove acceptable performance of control in response to machinery failure as well as control failure.
SIMULATION REQUIREMENT
Having specified the simulation tasks and required outputs, the simulation tools and techniques can be established. An additional requirementwas imposed on the simulation tech-nique decision process that being a requirement for a structured approach that would allow each equipment model
to be run and tested in isolation from the system. This
requirement enables model validation work to take place throughout the project life-cycle, from basic equipment de-sign to final acceptance trials.
Whilst this object oriented approach may seem obvious, it can be practically difficult to implement. For example, a distributed control system philosophy may be adopted but the downstream design process can only resolve the opti-mum location for some functions. If failure mode analysis is to be conducted properly using simulation, this approach needs to be strictly followed. The by-product of this ap-proach is that models of individual equipments enable the 'building block' approach when building a system model of a different configuration.
THE SIMULATION TOOLS
Past experience in building single, whole system models of complex propulsion systems has indicated a number of problems. These models are time-consuming to build, inef-ficient to run in terms of run time, and only one study can be conducted at any one time. For this project, which had a
Trans 1MarE, Vol 107, Part 2, pp 101-117
-3..7
8. 9l 2. 3-P H Sallabank & A J Whitehead
short timescale, the power system and propulsion system needed to be studied in parallel.
Two major models were therefore defined: a power system model;
a propulsion system model.
Due to the interactive nature of the overall system, the propulsion system model would incorporate a limited power system model and the control models.
Power system model
Interactive Power System Analysis (IPSA)1 was an
auto-matic choice for this model, as it provided the full range of electrical analysis facilities required for the project. This
analysis package had been used successfully on other projects,
including a number for the MOD(N) whose preferred elec-trical analysis platform is IPSA.
The IPSA package is fully validated and used extensively in the power generation and offshore industries, as well as by a number of major UK electrical consultancy companies. Validation by others, in addition to that carried out in-house, provides a high level of confidence in results and designs obtained through the application of IPSA.
Propulsion system model
MATLAB-SIMULINK2 was used as the base analysis pack-age for this model since it provided the necessary facilities for an efficient and structured model build, with high level instructions and self-documentation. It provided a suitable environment for control system development, this being particularly important given that an innovative approach to
some aspects of control - in particular the PMS - was
planned.
MATLAB-SIMUUNK has been successfully applied by the authors' company to a broad cross section of marine simulation activities, including detailed PMS design, ship position control systems and cp propeller pitch control systems. A library of generic models for the primary ele-ments of propulsion systems are maintained, providing a quick model build capability, as well as a proven detailed analysis platform. Future developments of this, and other equivalent packages, will permit automatic generation of application software from the model, and the ability to run simulations in true real-time with wide ranging possibilities for operator training.
Supporting facilities
A number of supporting analysis tools are also used:
Steady state analysis
Spreadsheet analysis of steady state powering in different modes is undertaken. This is particularly useful in propul-sion systems where, for operational reasons, intermediate points on the power/speed curve need to be optimised.
SPEED =_Systcat 1.1 g 0.9 0 0.8 0 0.7 ad 0.9 , 0.8 2 0.7 06 no 0.5 SPEED () , .5,51010 0.69 0 1 10 20 30 40 50 as 70 abbe 100
Fig 2a Steadysystem powerfactor,systempower factor without correction
POWER FACTOR vs SHIP SPEED
1 60
POWER FACTOR vs SHIP SPEED
10 0.99 20 I 30 0.94 OM 40 0.73 50 0.68 _ 0.660.6860[70 _}all 0.76 80 90-] 0.81 100
Fig 2b Steady system power factor; system power factor with correction
Also, given the range of different generator combinations, deck machinery loads and general ship services loads, a spreadsheet is the most efficient method of calculating all possible conditions.
Data testing
Complex, interactive models demand a complex range of data and create a difficult environment for diagnosing data errors, particularly for the less significant data elements. To overcome this, data testing prior to loading into the full model is necessary.
Data format can also present a problem. For example, electrical machine data maybe supplied as design values, or test results or a mix of both.
To resolve these problems a number of spreadsheet mod-els have been developed, which test machine data by creat-ing terminal performance characteristics from design data, and test the sensitivity of the terminal performance to each data element.
Trials recording
Trials recording is an important aspect of simulation, as it
provides the ultimate validation - or otherwise - of the
modelling work done. A Personal Computer (PC) based
0.69 0.56 0.53 0.53 0.55 0.59 0.64I0.70 0.77
2.
10th Ship Induction Motor Start Graph 4; 10 17.11M. r0017. P.O. 1600 Aral., La a. 0.0 (r) = Z. Z tz.1 C.-) 800 I ØB a 1.0 2.0 1.5 4.0 5.0
data logging system was used on this project. This facility had the capability to sample with an adequate bandwidth, and also to generate data files that could subsequently be manipulated on the PC to aid comparison with the simula-tion results.
STUDIES
Steady state analysis
An example of steady state analysis is givenin Fig 2. This
shows the variation of system power factor with ship's speed. The operational requirement of the vesselincluded significant periods to be spent at mid-range speeds. The poor power factor in this range results in generators operatingat
1 per unit current = 1312 Amper es
Trans 1MarE, Vol 107, Part 2, pp 101-117
near maximum kVA, with diesels well below maximum kW.
The effect of power factor correction at constant kVAR can be seen in Fig 2b.
Power system studies
A wide range of studies was conducted, with the aim of not just measuring the performance of the system with typical data values, but of optimising the system performance by modifying machine specification. Examples of studies
con-ducted are given below.
Model building
Electrical machine and equipment models are presented in a menu format. Model building consists of drawing the
2000 1800 1608 1 4 Be 12 , I 0.0 Fig 3a Induction motor data testing; simulated induction motor start
a 2 3 4 5 1
TIME IN SECONDS
P H Sallabank & A JWhitehead'
LOAD FLOW !RESULTS- BUSS AR PU VOLTS ANGLE & LINE kW / kVAR LOADING
aux 240V 0.970 .1 90 40 24 15 Lumped 0973 -168 4 GS8 Stbd 0.073 -1 68 Ftro Pin, 0 973 .190 164 186 Ho PFC Stbd 0 075 .010 10 701 3 0 976 207 0.873 -1.68 124 0 972 42 143 M2 0.975 0 976 0.976 0.02 7E-3 8E-3 Maximo 0.971 -7.59 Al 0.975 002 415 Stbd 007* .166 401 L7! Aurno443 15$ 0 073 168 Buoy Cap 0.974 Fans PT Windt -1."
configuration on screen and loading data - using Per Unit (PU) notation on a specified system base.
Transient response - induction motor start
Figure 3a shows the model prediction of an induction motor current during starting. Figure 3b shows the actual meas-ured machine starting current. Whilst there is a slight differ-ence in shape, the key characteristics of peak value, current decay profile and run-up time show good correlation. This example illustrates the point made earlier: the need to test individual equipment models in order to raise confidence in the overall system model performance.
Steady state load flow
Figure 4 shows the IPSA model of the electrical system with volts, load angle, and load kW and kVAR appended. The kW load of each induction motor represents the absorbed electri-cal power derived from the mechanielectri-cal load. The overall accuracy of this type of load analysis is, therefore, as much dependent upon good mechanical load data, as it is
depend-ent upon machine electrical data. Short circuit analysis
The model provides short circuit data in two forms. It
provides line current for each phase, demonstrating maxi-mum prospective asymmetry (Fig 5a) and provides a kA or kVA value (making or breaking) at any point in the system at a specified time after application of the fault (Fig 5b). This flexibility is important, enabling accurate modelling of dif-ferent devices interrupting the fault at difdif-ferent speeds.
915 Pod 3 0.974 22o10 -I 08 Hdlng 0 963 .163 54 903 5.18 Ml 0.977 HO TR2 0074 0.02 SwIxil 0 975 002 LV 792 0,976 13 -1.69 Capalain Windlass 0.074 0.966 -168 .147 5014 Prnp2 0 974 .148
Fig 4 Simulated system load flow
1c? 0l MG 0.067 1.43 37 20HaalIng 0.970 52 EtR Fans 46 970 .61 GS13 Port 0 979
FAV Pino2 Het Puff, 0.079 0,974 -108 -1,88 0.1 -143 lunaowP1 0973 -1.68 gra?AC -0.57 0'75" -2E-3 Hi PFC Port 43 0.075 -010 0.0 5.0 gTy12400 -1.07
Fault level results from the electrical simulation, using initial design data and, subsequently, as-built data, demonstrated
an error of less than 5% when compared to practically
measured test results.
Transient response generator step load change
Classical diesel generator set step load application and re-moval trials were conducted on the model, during factory
trials and onboard. The results, shown in Fig 6,
provide further good correlations between model and test results.
Transient stability transformer magnetisation
Significant research did not provide an adequate trans-former magnetisation model. The phase shift transtrans-formers rated at 1350 kVA represented a potential problem if, as a result of a failure at sea, they needed to be magnetised from reduced generator capacity. The worst case scenario was considered to be magnetisation from one 750 kW and one 364 kW generators running in parallel.
The transformer was modelled by a variable impedance giving an initial peak magnetisation current of 20 x Full Load Current (FLC) on infinite bars and decaying to steady state over a period of 250 cycles (5s). This represented the worst case available from empirical data.
II
Harmonic control and analysis
There are a number of elements to be considered when looking at the overall system harmonic performance, namely:
321 HO TR1 0.974 0.02 Bow Illru 1 000 867 53 Poll '7;4' FAN Pinot 0 967 35 54 30. 301 24 C...) 57W Prripi 0 989 .1.59
THREE PHASE FAULT ATM Swbd a - WITH MAXIMUM ASYMMETRY IN 'R' PHASE
Fig5a Short circuit simulation; simulated short circuit current waveform
415 3tbr.I 416 Port 43.5 43.5 Buoy Cap 13144np Cm/stain Woldlass T Winch Staering
Trans IMarE, Vol I07, Part 2, pp I01-117
T msec
Port Pro
E0240V
Fig 5b Short circuit simulation; simulated circuit breaker making ratings
Harmonic currents generated by non-linear loads (in 4. The effect of harmonics on significant reactances and this case the thyristor converters), capacitances withinthesystem.
The effect of alternator characteristicson system har- 5. The quality of power supplies at consumer terminals. The full network model shows the Total Harmonic Distor-The effect of the power factor correction units on liar- tion (THD) at different points in the system, without (Fig 7a)
Monies.
and with (Fig 7b) the power factor correction units in circuit. Lumped cop prop Pmpl
FAY Prro2 Hot Porno
THREE PHASE Fault Level ii 'Odor each Busbar Asym. peak, t= 10.0 ms
RED PHASE YELLOW PHASE BLUE PHASE
I kA
2.
PH Sallabank & A J Whitehead
10th SHIP
low) Graph 2: SM POWER (P)
900 800 700 600 SOO-400 300 200 100--10th SHIP K .W.
loco_ Graph SM POWER S I - KVAR
900* 800 700 600 SOO-400 300 200 100 0.0 10.0 0.0 10.0 20.0 30.0 40.0
Fig 6a Step load response; simulated kW response
20.0 10.0
The harmonic current spectrum generated by 12-pulse converters can be, in practice, significantly different to the
theory. Table I illustrates this point. (THDi is the total
harmonic distortion of the current waveform.)
40.0
1 T 1 I I
50.0 60.0 70.0
Fig 6b Step load response: simulated kVAR response
50.0 60.0 70.0
80.0
80.0
These practical variations are caused by phase shift trans-former characteristics and firing circuit design. System volt-age distortion becomes a product of both harmonic currents and the power generation and distribution system design.
10th SHIP
1.10 Graph 1: BUSBAR VOLTAGE - P.O.
1.05
--1.0
0.9
ODo
PF750 + SF364 - RESULTS FILE NO. 15
TIME (SECONDS)
Fig 6d Step load response; actualkWresponse
Harmonic control is required in order to achieve
anacceptable level of voltage distortion at consumer terminals, to avoid resonances, to minimise losses and to remove the need for de-rating equipment.
Alternator stiffness, in the form of sub-transient reactance (Xd"), has the major effect on the voltage distortion caused by harmonic current. The stiffer the supply (lower Xd") the lower the THD, but the higher the short circuit faultlevel.
The model used enables a trade-off study to be con-ducted. The results shown in Fig 7 are based on an Xd" of 10%. This produces fault levels within thecapacity of the
preferred range of circuit breakers and substantially reduces THD, as illustrated in Table II.
The figures in this table are illustrative only and basedon
simple converter/alternator circuits, without the effects ofa
full distribution network, parallel machines and
transform-ers.
Power factor correction units
The justification for power factor correction (ITC) lies in improving fuel economy. The design of these units must
Trans IMarE, Vol 107, Part 2, pp 101-117
0.90
10.0 20.0 30.0 40.0 0.0 60.0 70.0 - 80.0
PHSallabank & A J Whitehead 2(X) 1.05 LU 0
'w4or000mo-,M+0
< I-, .] 0 > 035 0.9recognise the presence of harmonics to avoid resonances, and by careful design the units can act as a harmonic filter as a bonus.
The design target was 10% THD without filters. Since theoretical harmonic currents from thyristor converters are normally greater than the practical values, the model predic-tion of 8% TI-ID worst case was expected, and proved to be pessimistic. Worst case onboard measurements without PFCs connected are <6%, and with PFC are <4% TED.
The harmonic penetration analysis provided by the model recognises both discrete components within the filter design and other reactive and capacitive components in the supply network. Resonances will therefore be identified at a compo-nent level by the model.
Figure 7 illustrates the attenuation of harmonics down to individual consumer terminals. These results, combined with transient stability studies and load flow studies,
pro-PF750 + SF364 - RESULTS FILE NO. 15
50
TIME (SECONDS)
Fig 61 Step load response; actual voltage response
TIME (SECONDS)
Fig 6e Step load response; actual kVAR response
PF750 + SF364 RESULTS FILE NO.15
vide the full results necessary to predict the worst case quality of power supplies.
Protection co-ordination
Resulting from the chosen system fault level, the protection co-ordination (discrimination) through the distribution net-work can be designed and tested.
Once specified in the electrical simulation model, protec-tion device tripping characteristics are operaprotec-tional in all transient stability studies, providing easy identification of any inadvertent or unwanted tripping operations.
Overall propulsion system simulation
The overall propulsion system simulation was implemented
50 105
'1
-Total Root Steal Proof CC Prop, AC Prop Sib Aux 240V 8.802 Aux 2400 141
Lumped FInn Pop
14, 6155
are Sum value as % of the fundamental voltage Dow Thor () 608 Semi 8.827 load.? 8 829
8 8 8
8 781 10091017 P. pp.. FAN Pmpl FAN P0gs2 RN Pump OslO 8827 8.760 8 028 0 1170Fig 7a Harmonic analysis; harmonic levels without filters
Harmonic PenetrationResults-- Harmonicdistortion factor(voltage) Total ROCA ,C2iiare Sum value as % of the fundamental&be Prp voltage
Bow Tiny 0 AS M3 M2 7256 7 293 7 298 6 112 BT AC Fans 8.781 11V TR1 10480 1-1.8.1411 DV 3011 7.316 Loa02 8.158 F/W Prml Fans 6.120 Prnpl 6.127 Steering 8.811 Steering 6 147 T Winch 8820 CD MSwhola 7.334 T WInch 6 100 Buoy Can 8 826 1.1,crms61 Buoy Cap 6.180
on MATLAB-SIMULINK, which allows a structured and layered approach to model building. Figure8 provides an illustration of different layers of the model, as they appear on the screen. These are supported by data tables at the lowest level in which equations, coefficients and constants are defined.
Fig 8a effectively defines the scope of the model. The propulsion control system is contained within block 1, and
Al 8811? 10.536 MI
0
10.412 M Swbd b 10 5311 Capslam YAndUns 8 828 8744 SAN Prop2 8.820 NV 102 7.342 LV TR2 415 Pon 6207 6 160 B.H.to Car1'1°4" SAN Pmp2 6 159Fig 7b Harmonic analysis; harmonic levels with filters
MG 9795 8.808 6 137 C) %Von Env SwIld 8.784 0513 Pori 0 828
8 8
F',,v1672 0:1P5r0
Pon Psp Lurnp004 8.6170
Pat Pro 0 CJ Prop2 AC Prop PO 1-11-0 PFC Pori Ep 400820 Prop2 AC Prop int kr6-9 PFC Punthe power generation and power management systems in
block
Manoeuvre simulation
In order to compare trials results with simulation results, the original model has been programmed to reproduce lever movement sequences recorded during trials.
Trans IltlarE, Vol 107, Part 2, pp 101-117 Harmonic Penetration Results - Harmonic distortion factor(voltage)
A2 Ms Mt 10.405 10 458 It 485 M Svrtscl 10.517 Propl DC Propl AC BT AC 10 520 Prop Sib HI PFC 51/x1 4/5 Sthol 415 Port 8 828 628 VP Cons Heating 8.740 Al7.356 MI 7 269 M Swbol b 7 356 LV Tim 415 Stbd 8.207 6 100 Ens:1240V 6.128 onizia'n 0. c`8,7209"66 0/6 Fans 6 105 11.145 GDO Port 6.159 Lumped4 6 152
7-PFCPH Sallabank & A !Whitehead
Slow speed manoeuvres
Slow speed manoeuvring was a particular area of study in order to ensure that maximum use of installed capacity was made. The overall fit between simulation and trials results is very good - especially bearing in mind the difficulty often encountered in modelling transient propeller performance and hull/propeller interaction under these conditions.
Crash stop manoeuvres
Figure 9 shows simulated (9a) and trials (9b) crash stop manoeuvres.
At t=2970 in Fig 9b, a significantperturbation in both shaft
speed and motor volts can be seen. (Similar small effects can be seen at t=2920-2950 and at t=3080-3100.) These are shaft speed changes caused by hull /propeller interaction in the slow speed transient regions.
The crash stop overall stopping time and ahead reach measured during sea trials were within 5% of the simulation predictions.
PROGRAMME
The programme for this project is significant for a number of reasons. The timescale of 18 months from contract to ship acceptance was not generous, given the development work planned. Additionally, this was a commercial contract, with commercial consequences attached to late delivery and un-satisfactory performance.
In addition to the 18 month period, two months of pre-liminary design work in support of the tender was carried out. This included a power system model first-pass, using estimated load data to confirm generator sizing, fault levels and harmonic penetration.
During the preliminary design and tender phase the decision was taken that the overall timescale was acceptable,
provided that simulation was extensively used during de-sign. Past experience had indicated that the simulation tools available to the project would enable rapid model build and provide the necessary confidence level needed for key de-sign decisions early in the dede-sign process.
The short setting-to-work and sea trials programme is attributed largely to the extensive use of simulation through-out the contract period.
SERVICE EXPERIENCE
In the 18 month period since vessel hand-over, a practical involvement has been maintained with the Northern
Light-house Board (NLB) and MV Pharos, both in respect of
fulfill-ing hardware guarantee obligations and, subsequently, with assisting NLB in satisfying classification society periodic survey requirements. Throughout this period it has proved possible to validate simulation predictions over an extended timescale and, indeed, to apply simulation to assessing and implementing an owner's request for enhanced vessel per-formance.
Table I 12-pulse converter harmonic currents
Table II THDand fault levels at different Xd"
In broad terms, no fundamental defects with the vessel's electrical and control system design have materialised, pro-viding further practical validation of the simulation activi-ties undertaken and their benefits. Key elements worth noting are:
Energisation of the main propulsion and bow thrust phase-shift transformers has been successfully carried
out from a single 750 kW main generator, with no
tripping of protection devices.
Protective device settings remain, as originally pro-posed by the IPSA simulation. There has been no re-ported spurious tripping or cascade failure of protec-tion devices.
There is no evidence of any harmonic problems or poor quality of electrical power supplies.
The NLB specific power management system is funda-mentally unaltered from that proposed by the original MATLAB-SIMULINK propulsion simulation, and con-tinues to satisfy NLB's operational requirements. A considerably higher than predicted harbour load neces-sitated adjustments to some PMS thresholds, to enable this load to be supported by a single 365 kW auxiliary generator, whilst maintaining adequate safety margins. Simulation was used to determine the revised thresh-olds and ensured a fast and trouble free
implementa-tion.
One particular request received from NLB, was to inves-tigate the feasibility of enhancing the vessel's stopping per-formance above that originally accepted during trials. The MATLAB-SIMULINK propulsion model provided a vali-dated platform for investigating this request and, in particu-lar, for assessing the benefits or otherwise of changing
Harmonic number %offundamental Theoretical IEEE (Refs 3 and 4) Manufacturers Measured 5 0 2.6 3.36 1.2 7 0 1.6 1.68 0.7 11 9.1 4.5 9.08 4.4 13 7.7 2.9 7.68 3.4 17 0 0.2 0.84 0.4 19 0 0.1 0.55 0 23 4.3 0.9 4.33 0.6 25 4.0 0.8 3.98 0.4 THDi 13.3 6.3 13.8 5.8 Xd" 15% 10% THD 15% 11% (12-pulse) Fault level 50 66 (Making kA)
3
Shalt Speed 111 Demand SneedError TSPOOd pp Error 2.2.2.1 Advance Co-effiecent CalculalronFigaa Propulsion system simulation: top level simulation
PI
PI wall anh.windop
Fig 8b Propulsion system simulation; motor controller simulation
2.2.2.2 J to Kt' Calculation
currentDernand
propulsion motor regeneration limits and speed/current ramp rates. Figure 10a shows a simulation of the vessel's stopping
performance as accepted during trials and Fig 10b a simulation
of the enhanced performance which could be achieved. Table III shows a comparison of vessel performance post-modification with that predicted, and that for pre-modifica-tion for a crash stop manoeuvre from maximum ahead speed.
The modification was implemented with minimal dis-ruption to vessel operation and precisely in accordance with the parameters derived by simulation. Key elements of the modification were an increase in the propulsion motor
re-generation level and a slowing down of the rate at which
3 Ship Dynamic Model ire PIO Current Controller 103.41 urtrt.o3 Convert To Knots Fl
Trans 1MarE, Vol 107, Part 2, pp 101-117
Cued Error Ora
F 'ea Angle Lends
logo 11 _op Vohnoo ShroSceed Out Drstance Out2 SiMTsne 003 11. EirrngAngle 1
Fig Sc, Propulsion system simulation; propeller thrust and torque simulation
shaft reversal occurs, thus decreasing the amount of propel-ler cavitation and increasing 'bite'.
CONCLUSIONS
The Diesel Electric Propulsion System outlined benefited significantly from the use of modern simulation tools.
The level of innovation in terms of harmonic control, power factor correction and load control could not have safely been achieved within the timescale, without theuse of
these tools and the engineering expertise to apply them.
°IC
Convert Armature Mu (F7/ VOMage Actual IVA/ Currant All Kq V 22.2.3 Jo Kq Propellor-Torque 00118 Ou117 Torque Cake aeonSpend 00 115
Po
`'111101 I(a) Of F ON9.651 Propellor Advance KI Thrust KM) Culls 111. 1(u) (F1 111C9,65) Propeller Torque rP011 Motor/Shah and m.o.:dors Levet Demand Shaft Conversion Speed (Fe) 21 1.1 Re "emPHSallabank & A J Whitehead (Lao 1500 501 .300 .1010 -1500 AMPS 011.10 iSpOONDS)
Fig 9a Crash stop;simulatedcrashstop
VOLTS
The close correlation between simulation and trials re-sults is attributable to good data and good modelling. How-ever, a simulation oriented approach to system design from
day one creates the environment and communications nec-essary to achieve such results.
The key aspects of this exercise can be summarised as
follows:
The quick build of a preliminary model enables funda-mental design decisions to be made early and also pro-vides key specification data for long lead equipment. An object oriented approach to model building, coupled with the design engineer hands-on facility offered by modern simulation tools, creates a greater system per-formance awareness in the design team.
Planning from the outset and proper trials recording in order both to validate models and improve the database are essential.
RESULTS FILE NO.34
SHAFT SPEED LEVER DEMAND IA.A1.9.-64060AArve. 200 3110 2141 100 .700
Keeping the model 'live' throughout the project life-cycle and controlling data updates reduces the risk of unexpected performance during trials or subsequent in-service operation.
Providing clear data requirements to equipment suppli-ers prior to contract substantially improves both the quality, timescale and cost control of the exercise. Maintaining up-to-date validated simulation models throughout the operational life of a vessel provides the owner or operator with a service which permits simple and realistic assessment of any proposed enhancements or changes to vessel performance.
Regarding the use of simulation in the design of diesel electric propulsion systems, a number of observations can be made. There is substantial scope for further innovation and optimisation in these systems. Unlike mechanical systems, the majority of the parameters within electrical systems are HOW AMPS 500 0 -500 .1000 16011. VOLTS SI1AFT SPEED LEVI:ROSMAN() 350 400 500 650 600 2600 27130 2800 2900 3100.1 3100 TIME (SECONDS)
Fig 9b Crashstop; actual crash stop
Initial vessel speed (kn) Shaft speed at start of regeneration (rev/min) Time from lever movement
to shaft reversal (s) Time from lever movement to dead in water (s) 150 100 -50 -100 0
Table Ill Enhanced vessel stopping performance
Ahead Reach
100 200 300 400 500
Time (Seconds)
Fig 10a Northern Lights marine results: crash stop pre-modification
Trials Simulation - Trials -pre-modification predicted post-modification
100 200 300 400 500
Time (Seconds)
Ahead Reach
Fig 10b Northern Lights marine results; crash stop post-modification
invisible, and only receive scrutiny if there is a problem. Simulation provides visibility. Visibility aids understand-ing and analysis, the essential understand-ingredients for improvement.
REFERENCES
Interactive Power System Analysis (IPSA), Copyright U MIST (1993). MATLAB-SIMULINK, Copyright The Math Works Inc (1991). IEEE Guide for Harmonic Control and Reactive Compensation for Reactive Power Converters, IEEE Standard 519.
C K Duffey and R P Stratford, Update of Harmonic Standard 519, IEEE Transaction, Vol 25, No 6 (November/December 1989).
Trans Mart, Vol 107, Part 2, pp 101-117
14.2 14.3 14.4 120 120 120 160 90 97 210 155 150 H600 800 600 700
PH Sal/thank & A J Whitehead
Discussion
Lt. R D Hayhoe RN '(Naval Support Command, Bath) I
would like to congratulate the authors for giving a very interesting and informative paper. There are a number of issues on which I would welcome the authors' expert com-mentary.
1. Would the authors expand on the potential for the real time implementation of their simulation technique as:
a design tool for investigating plant behaviour (in-corporating power and propulsion systems simul-taneously) and defining associated control system requirements;
an aid to human factors studies and ergonomic design of machinery control rooms;
ic a combination with suitable hardware models in order to build cost effective and versatile
proce-dural trainers for the coaching of ships' marine engineering staff.
2. Do the authors perceive their style of plant modelling causing a philosophical shift in the marine engineering industry's attitude towards sea trials, and any signifi-cant changes in the manner in which such trials, are conducted?
3. The authors stated that detailed operational perform-ance data required for simulations were requested from manufacturers before any contracts had been placed.
Since the selection of equipments was greatly influ-enced by simulated plant performance, how much reticence did the project team encounter from pro-spective machinery suppliers in disclosing all nec-essary data?
Were simulated machinery performances within the overall system model fed back to the associated manufacturer (for the mutual benefit of supplier and customer)?
4. What quality assurance criteria were applied when requesting, and in scrutinising, the source data sets supplied for simulations?
How were the boundary conditions for discrete data, sets initially defined, and subsequently refined, for over-, all propulsion system simulation?
P H Sallabank and A J Whitehead (Vosper 'Thornycroft Controls, Portsmouth)
I. Modern computer hardware enables real time, or faster,, analysis of complex marine power and propulsion
sys-tems.
a. The authors' company have successfully applied this technology to defining power and propulsion control system requirements, and to investigating/ optimising the interaction between items of electri-cal, control and propulsion plant.
'b. Whilst simulation has not, to date, been applied to ergonomic design studies for machinery control rooms, the technology available allows display pa-rameters and screen page layouts for VDU displays to be assessed and developed with confidence durr
ing the design phase of a project.
The facility of undertaking real time simulation 'leads logically into the provision of training facili-ties supported by the same simulation. Whilst the MV Pharos simulation was not used for this pur-pose, the technology is available to achieve this and as indeed being applied by the authors' company in
the field of ship position control systems.
A philosophical change in attitude towards sea trials is not currently foreseen. The availability of accurate simulations enables a reduction in sea trial duration through the ability to undertake a considerably greater, degree of validation during the design phase of a project.. Accurate simulation enables 'tuning' to be undertaken during design, not by what could historically be called a 'trial and error' approach during sea trials; this has
been validated by our experiences on MV Pharos. Surprisingly little reticence was encountered from machinery suppliers concerning provision of simu-lation data. This is attributed to the fact that data requirements were nominated to suppliers prior to hardware contract placement and did not form part of separate contracts; data was generally supplied at no additional cost. Our experience indicates that obtaining simulation data as a discrete item, even post-hardware delivery, is normally achievable, but at a cost.
Simulated machinery performances, both from the overall system model and from individual equip-ment validation models, were fed back to the asso-ciated manufacturers; this was an integral part of
model validation and system development/
optimisation.
Quality assurance for simulation data, in our experil-ence, relates to the accuracy of the data supplied by equipment manufacturers. Our normal practice is to apply tolerances contained in British, European or In-ternational Standards, as relevant to individual items of equipment. The accuracy of simulation output is more heavily dependent on certain key parameters than oth-ers; we have undertaken an extensive amount of sensi-tivity analysis on simulation data and, based on this, in some circumstances we will specify considerably tighter; but realistic and practically achievable, tolerances. At all stages of the simulation, from proposal to vessel handover, the most realistic data available was obtained from manufacturers. For some manufacturers, achiev-ing equipment performance in accordance with simula-tion data was a contractual requirement. At all stages a
5.
the project, equipment performance was checked against overall propulsion simulation predictions, thus main-taining the simulation as a 'desk top prototype' through-out the project life cycle.
Dr G Armstrong (Three Quays Marine Services Ltd) May
1, firstly, thank the authors for their paper and the presenta-tion which, for me, is very much preaching to the converted: I am strongly in favour of propulsion simulations being done at the design stage. Sea trials can be busy enough and anything which can be done to reduce the uncertainty is, of course, extremely helpful. Time spent on sea trials is an expensive way of fine tuning the control system.
Could the authors give more detail on the simulation
timescale in relation to the design period? Was ship
data and the model available in time to influence the design, and if it was possible to improve the performance of the ship after delivery, could this not have been done at the design stage?
Secondly, Fig 7 of the paper is based on a value for Xd" of
10%, which, I believe, is a conservative value for subtransient
reactance. Was this value specified as a result of the
simula-tion?
P H Sallabank and A J Whitehead (Vosper Thornycroft Controls, Portsmouth)
Simulation formed an integral part of the MV Pharos project from the proposal stage, through detailed de-sign, factory acceptance, basin trials, sea trials and in-service operation. Data was available from manufactur-ers in time to influence design. In some circumstances, early simulation enabled us to specify equipment pa-ra meters to manufacturers as part of performance
crite-ria.
The ship performance improvement undertaken after delivery relates to vessel stopping performance. MV Pharos utilises a regenerative braking system during stopping, exporting power from the fp propellers into the electrical power system. As-delivered, the vessel satisfied the owner's performance requirements, whilst maintaining a conservative approach to regeneration and stressing of equipment. With the benefit of the
experience of a number of months of in-service operation,
an improvement in stopping performance was consid-ered feasible without infringing safety margins. This was possible due to a higher than expected hotel load on the
vessel and greater experience of the propeller's four
quad-rant transient performance characteristics.
Hull/propeller interaction in the transient operating regime is notoriously difficult and unreliable in terms of model performance. A conservative approach is there-fore necessary for ship and machinery safety reasons. The result is often a larger than necessary safety margin measured at sea, with the scope to improve manoeuvra-bility without infringing prudent safety margins. Alternator sub-transient reactance values were speci-fied to the alternator manufacturer as a direct result of simulation. Sub-transient reactance values directly af-fect electrical system fault levels and harmonic levels; a
Trans IMarE, Vol 107, Part 2, pp 101-117
'trade-off' study involving technical performance and equipment costs was undertaken, leading to choice of optimised sub-transient reactance values.
J Walker (BMT Defence Services Ltd) I noted that the accuracy of modelling of electrical machines has achieved
high standards and, indeed, the presentation showed a
difference of approximately 1% in one case between mod-elled (from design data) and actual (from trials results) response. I also noted Mr Sallabank's comments with re-spect to propeller transient characteristics where the lack of
information at the design stage had been overcome by
experienced judgements, but that these problems were not as critical as propeller diameter. With these comments in mind, I would wish to determine which aspects of modelling are most critical in a simulation exercise and which aspects are still considered to be the greatest unknowns or risks at this time.
P H Sallabank and A J Whitehead (Vosper Thornycroft Controls, Portsmouth) Modelling of the electrical genera-tion and propulsion system contains little risk and few, if any, unknowns. Equipment and control system models are well established and validated, the principal risks relating to data quality and experience of the simulation operator. One area which does contain risk is the modelling of magnetising inrush currents into a transformer - significant research on our part has not uncovered a suitable model.
For the overall propulsion and vessel performance simu-lation, the areas of greatest unknown are transient four quadrant propeller modelling, cavitation and complex wa-ter flows around the swa-tern of the vessel.
G T Gerrard (Laing Oil & Gas) Thank you for a very absorbing and useful paper.
My questions are as follows:
It would be interesting to know the reasons for selection of dc propulsion motors rather than ac motors with variable frequency drives.
Your comments on transformer magnetisation with re-duced generation capacity, particularly the long decay period, were surprising. I would appreciate knowing over how much of this period the effect produced a current significantly over full load..
PI H Sallabank and A J Whitehead (Vosper Thornycroft Controls)
The use of dc rather than ac propulsion motors was specified by the owner and their consultants. MV Pharos utilises two 1 MW propulsion motors and, at the time of the initial design, dc motors and controllers offered considerable cost savings over equivalent ac technol-ogy. Issues of motor size and weight were not crucial to the design of Pharos, nor is motor maintenance, in par-ticular brush wear, given the vessel's operating profile. Transformer magnetisation produced a primary cur-rent significantly in excess of full load curcur-rent for a period of approx 0.5s with abnormal reduced genera-tion capacity.
2.
1.
