SCIENCE
`
`Computer modelling of the automotive
`energy requirements for internal
`combustion engine and battery
`electric-powered vehicles
`
`J.R. Bumby, B.Sc, Ph.D., C.Eng., M.I.E.E., P.H. Clarke, B.Sc. Ph.D., C.Eng.,
`F.lnst.E., and I. Forster, B.Sc.
`
`Indexing terms: Measurement and measuring, Instrumentation and measuring science, Computer simulation
`
`Abstract: In the paper the road vehicle simulation package Janus, developed in the Engineering Department at
`Durham University, is described. Janus is a flexible simulation package that allows internal combustion engine
`vehicles, electric vehicles and hybrid vehicles to be simulated, and their performance and energy consumption
`evaluated over standard driving cycles. The simulation techniques used in these programs are described and the
`simulation program shown to produce results comparable with experimental data.
`
`= battery state of charge
`= time
`= ambient temperature, °C
`= airgap torque, N m
`= engine compression braking torque, N m
`= tractive effort at the road wheels, N
`= vehicle velocity, m/s
`= armature voltage, V
`= battery voltage, V
`= head (or tail wind) velocity m/s
`= vehicle mass, kg
`= modified vehicle accelerative mass, kg
`= time interval
`= chopper efficiency
`= battery charge efficiency
`= gearbox partial efficiency
`= battery discharge time at constant current
`11, n
`= battery discharge time at constant power
`density PDi, h
`= air density, kg/m3
`= hill severity, degrees
`= field flux, Wb
`
`soc
`
`tT T
`
`a
`TCB
`TE
`
`Vv
`
`avb
`
`ww A
`
`t
`
`1c
`
`= maximum
`= armature
`= field
`= discharge rate
`= 5 hour discharge rate
`= step number
`= per unit
`
`Suffixes
`
`5kp
`
`u
`
`NB cc denotes engine cubic capacity, 1 cc = 1 cm3
`
`1
`
`Introduction
`
`In 1976 the Department of Industry commissioned The
`International Research and Development Co. Ltd., New-
`castle upon Tyne, to produce a worldwide survey of hybrid
`electric vehicles [1-3]. This report, prepared by A.J.
`Mitcham and J.R. Bumby, described numerous types of
`hybrid vehicles and their operating philosophies. It was
`evident that hybrid vehicles, and hybrid electric vehicles in
`particular, could be designed to meet a number of different
`
`265
`
`List of symbols
`= vehicle acceleration, m/s2
`= vehicle projected frontal area, m2
`= brake mean effective pressure, kPa
`= engine cubic capacity, cm3
`= aerodynamic drag coefficient
`= discharged ampere-hours. A h
`= coefficient of rolling resistance
`= engine compression ratio
`C3 = motor loss constants
`= distance travelled, km
`= EMF, V
`= battery energy density, kJ/kg
`= battery open-circuit terminal voltage when
`fully charged, V
`= battery terminal voltage, V
`= cold engine fuel flow factor
`= acceleration due to gravity, 9.81 m/s2
`= armature current, A
`= battery current, A
`= field current
`= battery discharge current
`= wheel inertia, kg m2
`= DC motor EMF constant
`= DC motor torque constant
`= polarisation resistance, Cl
`= rotational speed, rev/min
`= Peukert index
`= DC machine armature conduction loss, W
`= DC machine brush loss, W
`= DC machine core loss, W
`= DC machine field resistance loss, W
`= input power, W
`= DC machine mechanical loss, W
`= DC machine stray loss, W
`= battery power density, W/kg
`= gearbox losses, W
`= battery ampere-hour capacity, A h
`= wheel rolling radius, m
`= ohmic resistance, ft
`
`Knn P
`
`aL
`PbL
`PcL
`PfL
`P{
`PmL
`PsL
`PDi
`PGBL
`
`Qr
`
`D
`R
`
`Paper 4O48A (SI), first received 1st November 1984 and in revised form lsl May
`1985
`The authors are with the Department of Engineering, University of Durham,
`Science Laboratories, South Road, Durham DH1 3LE, United Kingdom
`
`IEE PROCEEDINGS, Vol. 132, Pi. A, No. 5, SEPTEMBER 1985
`
`Page 1 of 15
`
`aAB
`
`MEP
`cc
`
`cdCD
`crcRc u c 2 ,<
`
`dE E
`
`D
`Es
`
`E,
`fr
`
`9I
`
`ahI
`
`fhKk
`
`,
`k2
`
`FORD 1905
`
`

`
`operating objectives. However, as to which was the best
`design to use to meet a particular operating objective was
`much less obvious. Consequently, the report recommended
`that initially the performance potential of different hybrid
`vehicle designs should be examined using computer simu-
`lation.
`this paper
`recommendation,
`this
`Stimulated by
`describes a general road vehicle simulation program for
`evaluating the performance and energy efficiency of inter-
`nal combustion engined vehicles, battery electric vehicles
`or hybrid electric vehicles. This simulation package was
`conceived as a user-friendly interactive program, capable
`of evaluating the performance of a vehicle that is charac-
`teristic of a particular vehicle group, e.g. small car, medium
`car, light commercial van etc. The- program was given the
`name Janus. However, during the development of Janus, it
`became evident that additional software options must be
`included to allow the user to define precisely the vehicle to
`be simulated, thereby extending the program to specific
`vehicle studies. Indeed, it is this option that is used in
`program verification.
`Since the publication of the Mitcham and Bumby
`report, the use of computer simulation has played an
`increasingly important role in evaluating the potential of
`new vehicle technologies, and in the design of specific vehi-
`cles [4—6]. A number of simulations have been written spe-
`cifically for one type of vehicle [7-10], while others, such
`as ELVEC [11-12] and HEAVY [13-14], are general road
`vehicle simulation packages. The majority of
`this
`published work has been in, and from, the USA, and this is
`particularly true of the two general packages ELVEC and
`HEAVY, both of which are powerful design tools. One of
`these, HEAVY, has been mounted on the SERC comput-
`ing network, but, at the present time, cannot be run inter-
`actively. In its conception, Janus was designed to be an
`interactive simulation package directly applicable to Euro-
`pean vehicles.
`In this paper, the characteristics of Janus are described
`in detail and used to demonstrate how a particular vehicle
`can be quickly and easily assembled from a standard sub-
`routine
`library.
`Initially
`the
`fundamental
`equations
`describing the vehicle dynamics are reviewed. In the final
`part of the paper, the use of Janus in simulating the per-
`formance of both electric and internal combustion engined
`road vehicles is demonstrated. In the case of the electric
`vehicle, this is achieved by simulating the General Electric
`ETV-1 electric car [15]; while when considering the inter-
`nal combustion engined vehicle its usefulness in examining
`new vehicle concepts is demonstrated by investigating the
`influence of 'fuel off at idle'.
`
`Vehicle dynamics
`2
`To provide the necessary propulsion power, any vehicle
`drive train must be able to provide sufficient tractive effort
`at the road wheels to overcome aerodynamic drag, rolling
`resistance and hill gradient effects, while still providing the
`necessary vehicle acceleration. Consequently, at any parti-
`cular velocity and acceleration, the net tractive effort
`required at the road wheels can be expressed as the alge-
`braic sum of these components, i.e.
`
`TE= Td+ Tr+Tg+ TflN
`where the respective components of tractive effort are
`Td = l/2PCdA(V + KJ 2 N
`Tr
`=CrWgN
`
`(1)
`
`(2)
`
`(3)
`
`Tg = Wg sin <f> N
`
`(4)
`
`•
`
`(5)
`
`Ta=WaN
`and
`Cd = drag coefficient
`A = vehicle frontal area, m2
`Cr = coefficient of rolling resistance
`W = vehicle mass, kg
`<j> = hill severity and percentage grade = 100 tan <p, %
`V = vehicle velocity, m/s
`Vw = head wind velocity, m/s
`a = vehicle linear acceleration, m/s2
`p - density of air = 1.226 kg/m3 (at 15°C
`and 105 Pa (1 bar) ambient conditions)
`
`In eqn. 3 the coefficient of rolling resistance is dependent
`on the type of tyre used, the tyre pressure and the vehicle
`speed. However, it is difficult
`to quantify reliably for
`vehicle simulation studies, and a number of authors have
`used a quadratic, velocity-dependent equation for rolling
`resistance [7, 9, 11, 16], while others assume the coefficient
`of rolling resistance to be a constant [17]. In general, the
`variation in the coefficient of rolling resistance with veloc-
`ity is small up to about 90 km/h, with the major changes
`occurring at speeds significantly greater than this. Indeed,
`in tests on the General Electric ETV-1 electric car the
`energy required
`to overcome rolling resistance was
`observed to decrease with speed [15], possibly due to
`increased tyre heating at higher speeds. To allow for simu-
`lation flexibility the coefficient of rolling resistance used in
`Janus contains both a constant and a velocity-dependent
`term, both of which can be defined by the user.
`The accelerative tractive effort in eqn. 5 relates solely to
`the linear acceleration of the vehicle and takes no account
`of the rotational inertia of the road wheels and engine. The
`inertia of the road wheels can increase the effective vehicle
`weight by about 2% during acceleration, and is included
`in eqn. 5 by using an effective accelerative weight W given
`by
`
`W' = W + -f kg
`rD
`
`(6)
`
`where
`Iw = inertia of the road wheels, kg m2
`rD = rolling radius of the wheel, m
`The engine inertia is not included directly in the tractive
`effort but as a local energy demand within the engine algo-
`rithm itself. It is interesting to note that, if referred to the
`wheels through the overall gear ratio, the engine inertia
`can increase the effective accelerative weight by about 20%
`in first gear and 2% in top gear.
`
`3
`
`Janus—its structure and operation
`
`A fundamental feature of Janus is the use of separate
`Fortran subroutines to represent individual vehicle com-
`ponents. The necessary subroutines can then be easily
`assembled into a master program to simulate a particular
`type of vehicle. This approach is adopted as many of the
`components required by the conventional internal com-
`bustion engined vehicle are encountered in an electric
`vehicle while components used in both these appear in the
`hybrid electric vehicle. Each component block is written to
`ANSI Fortran VII standard. The master program can also
`include additional Fortran statements if required. Such
`versatility is of prime importance, as it allows the user to
`increase the degree of output information to suit his own
`
`266
`
`IEE PROCEEDINGS, Vol. 132, Pt. A, No. 5, SEPTEMBER 1985
`
`Page 2 of 15
`
`FORD 1905
`
`

`
`particular requirements, or to restructure the control com-
`mands when simulating a hybrid vehicle.
`The advantages of this block structure approach can be
`demonstrated by considering the layout of the internal
`combustion engined vehicle shown schematically in Fig. 1.
`
`ICENG
`
`COUPL
`
`TRANS
`
`1 C engine
`
`gearbox
`
`Fig. 1
`
`Conventional internal combustion engine vehicle drive train
`
`The software blocks representing this vehicle are shown in
`the flow diagram of Fig. 2, and it is only the components
`within the dotted frame that are assembled by the user. To
`simplify program writing, the names of the software blocks
`have been selected to be similar to those of the com-
`ponents they represent. For operational flexibility each
`subroutine is divided into three sections termed, respec-
`tively, the initial section, dynamic section and output
`
`section. The initial and dynamic sections are further
`divided into two subsections. During the initialisation
`phase, each of the component subroutines is entered in
`turn, and the vehicle power train and driving cycle par-
`ameters specified. Unfortunately,
`in some subroutines
`information may be required from components upstream
`of the one currently being accessed, and as such will have
`yet to be denned. To overcome this problem a two pass
`system is used. On the first pass full details of the individ-
`ual components are specified, while on the second pass any
`calculations requiring information from an upstream com-
`ponent are completed. Such a system is necessary, for
`example, when automatic weight generation is used. In
`such circumstances, the weight of the gearbox and final
`drive depends on the specified engine torque and power,
`one of the last components to be defined.
`Having established the initial operating conditions the
`simulation enters the dynamic section, the main computa-
`tional part of the program. In this part of the program, the
`fuel efficiency of the vehicle and all the component effi-
`ciencies are calculated. For the majority of driving cycles,
`such as the ECE-15 urban cycle shown in Fig. 3, the
`vehicle velocity is explicitly defined as a function of time,
`
`reset initial
`conditions
`
`/ set t=0
`Vset flag for ,
`'initial pass/
`
`VEHICLE
`specifies all
`vehicle
`parameters
`
`DCYCLE
`acceleration and
`velocity at time
`t for specific
`driving cycle
`
`WHEELS
`torque and speed
`at wheels
`
`AXLE
`torque and speed
`at drive shaft
`account of axle
`efficiency
`
`TRANS
`transmission type,
`gear.torque and
`speed at engine
`
`COUPL
`friction clutch
`or torque conv.
`
`ICENG
`fuel consumption
`over interval
`At
`
`IEE PROCEEDINGS, Vol. 132, Pt. A, No. 5, SEPTEMBER 1985
`
`Page 3 of 15
`
`Fig. 2
`Flowchart for a conventional internal
`combustion engine
`
`267
`
`FORD 1905
`
`

`
`enabling the program to step through the driving cycle at
`one second intervals (default value) calculating vehicle
`
`60
`
`50
`
`40
`
`.c
`1 3°
`C 20
`o1
`
`10
`
`0
`
`20
`
`40
`
`60
`
`80 100 120 140 160 180 200
`time,s
`
`Fig. 3
`
`ECE-15 urban driving cycle
`
`velocity and acceleration directly from the driving-cycle
`data. The tractive effort at the road wheels is calculated at
`each time instant using eqn. 1 and converted into a torque
`and rotational speed demand in the subroutine WHEELS.
`This torque requirement is then reflected back through
`each of the drive train components to the engine output
`shaft. At each stage, account is taken of the gear ratio and
`instantaneous loss within each of the transmission com-
`ponents. Fuel usage is now obtained from the engine fuel
`map by assuming the engine load to be constant over the
`step interval. By sequentially repeating this process, the
`total fuel used over the driving cycle can be found.
`Such a direct calculation method is adequate for the
`majority of situations, as the performance of each drive
`train component can be considered to be independent of
`the other components. Normally the component efficiency
`will depend only on the torque and speed it is required to
`transmit at that time instant. However, in some cases, the
`performance and behaviour of two or more of the drive
`train components may be strongly dependent. For
`example, in an electric vehicle, the behaviour of both the
`electronic controller and the traction motor depends on
`the battery terminal voltage. In turn, the battery terminal
`voltage depends on load current. In such instances, it may
`be necessary to iterate around the vehicle drive train until
`a stable operating condition is achieved, when component
`efficiencies can then be calculated. To accommodate this, a
`two-pass system is used at each time step. On the first
`pass, all the vehicle variables such as torques, speeds etc.
`are calculated from the road wheels to the energy source,
`using an iterative process to establish the stable operating
`condition. Once the stable operating condition has been
`reached the second dynamic pass is started, in which the
`
`TRANS
`
`fuel used during the last time step and the energy loss in
`each of the drive-train components is recorded. The time is
`then updated and the process repeated until the driving
`cycle is completed.
`Once the driving cycle is completed, the simulation
`enters the output section, where full details of the vehicle,
`driving cycle and the individual drive-train components is
`given with graphical presentation of time-varying quan-
`tities, engine fuel maps etc. Besides displaying component
`efficiencies, losses and the overall vehicle fuel economy, the
`percentage of the total cycle time spent in each area of the
`engine fuel map is also given. Such fuel map information is
`invaluable, particularly when detailed studies on the effect
`of the vehicle component sizing and control on fuel effi-
`ciencies are being undertaken. Alternatively, if a detailed
`component breakdown is not required, a reduced output
`facility may be selected which simply details the vehicle
`range and/or fuel consumption when operated contin-
`uously over a particular driving cycle.
`Once a simulation run has been completed, the control-
`ling software allows further runs to be conducted over a
`different driving cycle. Alternatively, modifications may be
`made to the individual power-train components and/or the
`vehicle parameters.
`The different component subroutines available within
`Janus are listed in Table 1 and, in many cases, contain
`
`Table 1: Component subroutine names
`
`Component
`
`Simulation name
`
`Vehicle definition
`Driving cycle
`Wheels
`Final drive
`Transmission
`Clutch or torque convertor
`Internal combustion engine
`Series DC motor
`Separately excited DC motor
`DC switched reluctance motor
`AC inductance motor
`DC generator
`Field chopper
`Armature chopper
`Traction battery
`'Gearing' for connecting two
`prime movers
`Battery switching
`Summing block
`Torque splitting module (hybrids)
`Power splitting module (hybrids)
`Vehicle controllers
`
`VEHICLE
`DCYCLE
`WHEELS
`AXLE
`TRANS
`COUPLE
`ICENG
`DCSER
`DCSHUNT
`DCREL
`ACINDUC
`DCGEN
`FCHOPR
`ACHOPR
`BATTERY
`DRIVE
`
`BATSWCH
`SUM
`TORQSPLT
`POWSPLT
`VEHCONT
`
`performance details and simulation algorithms on a
`number of related items. For example, the tree structure of
`Fig. 4 describes the specific components available within
`the transmission module, any of which can be selected
`from within the program.
`
`gearbox
`I
`
`automatic
`(as manual)
`
`manual
`(number of gears: 2-6,
`gear ratios, upchange
`schedule, downchange
`schedule)
`
`Page 4 of 15
`
`CVT
`
`hydromechanical
`
`Perbury
`
`'V belt
`
`cone
`roller
`
`constant
`efficiency
`
`Fig. 4
`
`Subdivision of transmission module
`
`1EE PROCEEDINGS, Vol. 132, Pt. A, No. 5, SEPTEMBER 1985
`
`FORD 1905
`
`

`
`4
`
`Component simulation
`
`4.1 Transmission
`The
`transmission subroutine (TRANS) simulates the
`behaviour of the vehicle transmission system and includes
`models of manual and automatic, fixed ratio, gearboxes
`and a variety of continuously variable transmissions
`(CVTs). The fixed ratio gearbox model is based on an effi-
`ciency algorithm
`that predicts the gearbox efficiency,
`depending on the transmitted torque and speed. The algo-
`rithm also allows the number of gears, gear ratios and the
`gear change schedule to be specified. Both the up-change
`and the down-change schedule can be specified as a func-
`tion of vehicle speed or, alternatively, the ECE-15 gear
`change schedule may be used. As the time taken during the
`gear change is small compared with the total cycle time,
`the simulation assumes instantaneous gear changes.
`The efficiency of a gearbox depends on the power trans-
`mitted, the operational speed, the gear ratio and the gear
`profile and will vary with different gearbox designs.
`Although efficiency is high at normal full-load operating
`conditions, typically in excess of 96%, at part load, parti-
`cularly high-speed part-load operation, efficiency drops
`and can significantly effect the vehicle fuel economy.
`Gearbox losses can be broadly divided into two com-
`ponents, a no-load loss due to gears churning in the
`gearbox oil and a load friction loss due to the transfer of
`load between the gears. This last component of loss is also
`dependent on the gear profile. Such losses are approx-
`imated in Janus by the power loss expression
`
`PGBL = (1 - rtGB)Pi + 1.14 x 10" V W
`
`(7)
`
`where P, is the gearbox input power and rjGB is the partial
`efficiency relating to the friction loss. The second velocity-
`dependent term is associated particularly with the churn-
`ing loss. The friction loss component is dependent on gear
`ratio and in the simulation is changed such that non-
`meshing gears, i.e. gear ratio of 1:1, have a higher effi-
`ciency than the meshing gears. Typically, rjGB is taken as
`0.99 for nonmeshing gears and 0.98 for meshing gears. For
`an automatic gearbox, a single value of 0.97 is assumed.
`At high rotational speeds, particularly at low load, the
`value of the churning loss speed index v becomes critical as
`gearbox' efficiency is dominated by this loss term. To
`obtain gearbox efficiency values typical of those quoted in
`the literature [17, 22], a speed index of 2.1 is assumed.
`The transmission is coupled to the engine by the
`COUPLE routine, which contains models for both a
`torque convertor and a friction clutch.
`Efficiency curves similar to those described by eqn. 7 are
`also used in the final drive model. However, as the final
`drive may be constructed from either bevel or spur gears, a
`different friction loss component is required for each gear
`type, bevel gears having a higher frictional loss. The final
`drive simulation block also contains efficiency values rep-
`resentative of both chain and belt drives, as these have
`been used previously in electric vehicles [15, 20].
`
`4.2 Internal combustion engine
`Accurate engine maps giving the specific fuel consumption
`at different loads and speeds are essential if fuel consump-
`tion over urban driving cycles is to be accurately com-
`puted. Such fuel maps are commonly presented, either in a
`power/speed,
`brake-mean-effective-pressure
`(BMEP)/
`speed, or torque/speed form. An example of the latter is
`shown in Fig. 5 for a typical 1850 cc engine. For computa-
`tional purposes the fuel map is divided into twenty BMEP
`
`IEE PROCEEDINGS, Vol. 132, Pt. A, No. 5, SEPTEMBER 1985
`
`Page 5 of 15
`
`and speed increments and stored as a 20 x 20 two-
`dimensional array. At each time interval through the cycle,
`
`4000
`
`1000
`
`2000
`3000
`speed, rev/min
`Fig. 5
`Engine fuel map for a typical 1850 cc engine
`See Reference 18
`a Maximum torque curve
`b Constant SFC 400g/kWh
`c Power for constant 115 km/h
`d Power for constant 80 km/h
`e Power for constant 50 km/h
`
`the fuel consumption is calculated for the particular load
`and engine speed by linearly interpolating between the
`four specific fuel values adjacent to the operating point.
`The position within the fuel map array is then recorded. At
`the end of the cycle simulation, the total fuel used is calcu-
`lated and the fuel usage information presented graphically
`in the form of a fuel usage map. This map displays that
`percentage of the cycle time the engine spent in different
`parts of the operating region. As engine efficiency maps are
`obtained from steady-state load tests to obtain net engine
`torque during an acceleration interval, the engine inertia
`torque must be added algebraically to the output torque.
`As the engine inertia torque is proportional to the engine
`angular acceleration, this is readily achieved, except when
`a gear change takes place. As a gear change is assumed to
`occur instantaneously, a step change occurs in the engine
`speed which can lead to very high, false, inertia torques. In
`general, engine inertia has only a small influence on the
`vehicle fuel economy; thereby, allowing engine inertia
`effects to be neglected during the computational time step,
`the gear change takes place.
`Experience has shown
`the use of BMEP/
`that
`normalised speed fuel maps are to be preferred to power
`maps, as, with the latter, significant errors can be intro-
`duced at low-torque low-speed conditions; conditions that
`are common in urban driving cycles. This arises because
`under such conditions the power demand is low, possibly
`less than 10% of maximum power, so that if equal power
`increments are used in the storage process, accurate spe-
`cific fuel values corresponding particularly to the lower
`left-hand corner of Fig. 5 can be lost. By normalising the
`engine speed, the fuel maps can be 'stretched' to either a
`different engine cubic capacity, and hence power output, or
`to a different maximum engine speed other than their
`design value. Although stretching of the engine maps over
`too large an engine range, and particularly over a different
`compression ratio range, is not recommended, it does
`provide substantial additional flexibility with negligible
`additional programming complexity.
`At some stage during the driving cycle, it is possible for
`the demand torque to be greater than the full throttle
`
`269
`
`FORD 1905
`
`

`
`capability of the engine or for the engine speed to be above
`the engine's maximum. If this occurs, a warning is flagged
`to the user's VDU screen and the cycle velocity and accel-
`eration reduced until the maximum allowable engine
`output is reached. The modified driving cycle can then be
`compared with the demand cycle in the graphical output.
`At low engine torques the operating point may lie between
`the speed axis and the first set of specific consumption data
`stored in the fuel map. It is then necessary to use as one of
`the interpolation points fuel flows at high engine speeds
`but zero output torque. Such fuel flows are calculated by
`assuming the idle fuel consumption to be increased in the
`ratio of operating speed to idle speed.
`During both the overrun and stationary periods of the
`cycle, idle fuel flows are consumed by the engine and, as
`will be shown in Section 5, can be a significant fraction of
`the total urban fuel used. To evaluate the magnitude of
`this effect, a 'fuel off at idle' and/or a 'fuel off on overrun'
`option is included in the engine algorithm.
`The engine algorithm also includes a cold start option
`as, during such conditions, fuel flows can be 200 or 300%
`of those present in the warm engine [37]. As the actual
`penalty varies both with ambient temperature and distance
`travelled, this effect is included by multiplying the warm
`engine fuel flows by the factor fT, where
`
`fr —fo — ^ U20 —fo)
`
`(8a)
`
`and/ 0 a n d / 20 are the equivalent modifying factors at 0°C
`and 20°C, respectively, given by
`/ 0 = 0.88e1-42/(1-42 + d'
`/ 20 = 0.48e137/(13-7+d)
`(8fc)
`d is the cumulative distance travelled in kilometres and T
`the ambient temperature in °C.
`During deceleration, compression braking within the
`engine is normally present. The value of this compression
`braking torque depends on the coefficient of expansion
`and compression within the cylinders, the engine cubic
`capacity and compression ratio CR and can be expressed
`
`TCB = cc^0.081(C°-2 - 1) - 0.05 ^n
`
`Nm
`
`(9)
`
`where cc = engine cubic capacity, cm3.
`A torque loss equivalent to 2.5% of the maximum power
`when the engine is running at maximum speed is also
`included in the braking algorithm to represent the braking
`effect of accessory load. During the overrun part of the
`cycle, an iterative braking process is necessary to deter-
`mine the energy split between the compression-braking
`effect and the friction brakes, the actual energy dissipated
`in the brakes being displayed in the output. As, during this
`part of the cycle, flow in the venturi of the carburettor is
`sonic, idle fuel consumption is assumed.
`
`4.3 Electric traction motors
`The modelling method used to represent the DC and AC
`traction motors is similar to that used for the internal
`combustion engine in its use of efficiency maps, fuel usage
`maps, interpolation techniques and inertia loss calculation.
`However, a significant difference is that the power output
`of the electric motor is not directly related to the primary
`energy used as in the internal combustion engine, but to
`the electrical power input to the motor from the battery.
`The behaviour of the battery must now be included in the
`simulation.
`
`Page 6 of 15
`
`Four types of electric traction motor are currently avail-
`able within the simulation; the DC separately excited
`motor, the DC series motor, the DC switched reluctance
`motor and the AC squirrel-cage induction motor. The DC
`series motor has traditionally been used for electric vehicle
`traction as its inherent
`torque characteristic, varying
`inversely as a function of speed, best suits the requirements
`of the road vehicle. However, the series motor is not so
`well suited to regenerative braking, as the separately
`excited motor and also tends to have a lower efficiency at
`low speed. Consequently, with the advent of power elec-
`tronics there is an increasing use of the separately excited
`DC motor, particularly with transistor chopper control of
`field current and SCR armature chopper control as, for
`example, used in the Lucas hybrid electric car [24], The
`advances in power electronics have also paved the way for
`more advanced traction motor types such as electronically
`commutated DC motors, the DC switched reluctance
`motor and the AC induction motor. In the Janus simula-
`tion the DC switched reluctance and the AC induction
`motors are both represented as being favourable advanced
`traction motor drives.
`The torque/speed characteristic and efficiency map for
`the separately excited motor used in the General Electric
`ETV-1 electric car is shown in Fig. 6 [25, 26]. Such a char-
`acteristic can be split into four operating regions; regions 1
`and 2 associated with motoring action and regions 3 and 4
`with regenerative braking. When operating in regions 1
`and 4 the field current, and hence flux, is maintained con-
`stant and speed control of the motor is achieved by chop-
`ping the armature supply voltage. In regions 2 and 3 the
`battery is connected directly across the motor armature
`terminals and speed control is achieved by field weakening.
`
`1.0
`
`0.6
`
`0.8
`O.A|Rb2 0.6
`0.2
`normalised speed n/nmO)I
`
`1.0
`
`region
`
`generating
`
`Fig. 6
`Torque/speed curves and efficiency contours for
`separately excited motor
`
`the ETV-1
`
`IEE PROCEEDINGS, Vol. 132, Pt. A, No. 5, SEPTEMBER 1985
`
`FORD 1905
`
`

`
`This is a particularly attactive feature, as the controller
`does not now handle the large armature currents resulting
`in a high motor/controller efficiency.
`Although the efficiency map allows the required motor
`input power to be calculated from the known shaft power,
`a knowledge is also required of the currents and voltages
`within the DC motor. These can be obtained from the fam-
`iliar torque and EMF equations if the torque and EMF
`constants for the machine are known. In the simulation,
`the motor efficiency map is stored in a normalised torque/
`speed form and, as it is assumed to be typical of DC
`separately excited traction motors, the actual torque and
`EMF constants will vary depending on the specified
`rating. Nevertheless, both torque and EMF constants can
`be satisfactorily deduced from the efficiency map, if the
`armature current is assumed to be continuous and other
`pertinent assumptions with regard to power loss and satu-
`ration are made as described in Appendix 10.1.
`Once the torque and EMF constants for the motor have
`been evaluated, the voltages and currents can be found at
`any operating condition, using the standard torque and
`EMF equations in conjunction with the efficiency map.
`Calculations in the armature control region are straight-
`forward as the field current is constant at 1 p.u., but, in the
`field weakening region, the field current, the degree of satu-
`ration and the back EMF are unknown and an iterative
`solution is necessary.
`During the regeneration phase, a similar set of equa-
`tions are used as for the motoring condition. However, a
`lower maximum armature current can be specified to allow
`the rating of the regenerative step-up chopper to be
`reduced relative to the motoring step-down chopper. In
`calculating the regenerative efficiency, the same efficiency
`map is used as in the motoring mode.
`A similar modelling technique is used for the DC series
`motor, but is somewhat simplified, as there is only one
`electrical input to this traction motor. In addition, the user
`can select limits to be imposed on both the maximum
`speed and the maximum torque during regeneration, to
`allow the benefits of regeneration to be studied in detail.
`For the DC switched reluctance motor and the AC induc-
`tion motor, the drive electronics are an integral part of the
`drive system and efficiency data representing the behaviour
`of the combined motor/controller subsystem are used.
`
`4.4 DC-DC chopper controller
`The control of field current and the armature current in a
`DC traction motor is generally achieved by 'chopping' the
`supply voltage to vary the average voltage applied to the
`winding. If the chopper has an efficiency of r\c, then,
`assuming continuous conduction, the chopper input and
`output currents are related as
`
`VaIa
`
`(10)
`
`/„ =
`where
`Ia = average motor armature current
`Vb = battery voltage
`Ib — average battery current
`The efficiency of the chopper depends not only on the type
`of chopper used, transistor or SCR, but also on the chop-
`ping frequency, instantaneous voltage ratio (mark-space
`ratio) and
`the average current being passed by the
`chopper.
`To account for such variations, Janus contains efficiency
`data for typical SCR and transistor-based chopper designs
`[26, 28]. In particular, the efficiency variation of the
`
`ETV-1 transistor armature chopper obtained from Refer-
`ences 26 and 38 is shown in Fig. 7. Curve fitting to this
`100
`
`motoring
`
`90
`
`„ 80
`>;
`1 70
`
`% 60
`
`50
`
`t-0.1 0.2 0.3 0.4 0.5
`
`0.6 0.7 0.8
`duty cycle
`Armature chopper efficiency data
`JPL experimental dalal ^^
`r a ( io _ , Q
`Janus simulation
`)
`Janus simulation—current ratio = 0.1
`
`Fig. 7
`
`0.9 1.0
`
`data allows the chopper efficiency dur