vehicle modeling using bg

Upload: danish-moin

Post on 07-Apr-2018

220 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/4/2019 Vehicle Modeling Using BG

    1/11

    Vehicle dynamics simulation using bond graphs

    Germn Filippini, Norberto Nigro and Sergio Junco

    Facultad de Ciencias Exactas, Ingenieray Agrimensura

    Universidad Nacional de Rosario. Av. Pellegrini 250, S2000EKE Rosario, Argentina.

    Abstract This work addresses the construction of a four-

    wheel, nonlinear vehicle dynamic bond graph model and itsimplementation in the 20sim modeling and simulation

    environment. Nonlinear effects arising from the coupling of

    vertical, longitudinal and lateral vehicle dynamics, as well asgeometric nonlinearities coming from the suspension system

    are taken into account. Transmission and (a simplified) engine

    models are also included. The modeling task is supported by amultibody representation where the parts are handled as rigid

    bodies linked by joints. The first step is the 3D-modeling of

    each, chassis, suspension units, tires and joints, as bond graph

    elements equipped with power ports for physical

    interconnection. This is done with the help of vector ormultibond graphs in order to exploit their compactness and

    simplicity of representation. These 3D-units are later

    programmed as 20sim bond graph subsystems whoseassembling through the power ports allows for an automated,

    modular approach to the construction of the overall vehicle

    model. Simulation experiments corresponding to standard

    vehicle dynamics tests are presented in order to show theperformance of the model.

    Index Terms Bond Graphs, Multibody systems, Vehicledynamics.

    I.INTRODUCTIONodeling and simulation has an increasing importance in

    the development of complex, large mechanical systems.

    In areas like road vehicles [1, 12, 15], rail vehicles [9], high

    speed mechanisms, industrial robots and machine tools [10,

    11, 6], simulation is an inexpensive way to experiment with

    the system and to design an appropriate control system.

    The above indicated kind of mechanical systems belong for

    a major part to the class of systems of rigid bodies or

    multibody systems. Such systems consist of a finite number of

    rigid bodies, interconnected by arbitrary joints. The latter may

    exhibit properties of rotational or translational freedom,

    Norberto Nigro whish to thanks CONICET for its support to this research.

    N. Nigro is with CIMEC-INTEC-CONICET and with the School of

    Mechanical Engineering, Facultad de Ingeniera (FCEIA), UniversidadNacional de Rosario (UNR), Argentina. (Corresponding author. Phone: 54-

    342-4511594; fax: 54-342-4550944; e-mail: [email protected]) .

    Germn Filippini, is fellow at CIMEC-INTEC and teaching assistant at the

    School of Mechanical Engineering, FCEIA-UNR ([email protected]).Sergio J. Junco is with the Department of Electronics, FCEIA-UNR,

    Argentina ([email protected]).

    damping and compliance, and are the place of the attachment

    of drives or external forces.

    In classical mechanics several procedures exist through

    which differential equations can be derived for a system of

    rigid bodies. In the case of large systems these procedures are

    labor-intensive and consequently error-prone, unless they are

    computerized [4].

    This work applies the multibody theory through the

    multibond or vector bond graph technique [2, 3, 5, 6, 7] to the

    modeling of a complex four-wheel vehicle system. Primarily, bond graphs (BG) represent elementary energy-related

    phenomena (generation, storage, dissipation, power exchange)

    using a small set of ideal elements that can be coupled together

    through external ports representing power flow. Thus, they are

    well-suited for a modular modeling approach based on

    physical principles. Hierarchical modeling becomes possible

    through coupling of component or subsystems models through

    their connecting ports. Besides these physical features

    capturing energy exchange phenomena, it is also possible to

    code on the graph the mathematical structure of the physical

    system, in the sense of showing the causal relationships (in a

    computational sense) among its signals [2]. On the one side,

    this allows connecting BG-models to signal flow graphs or

    block diagrams, and -on the other side- it turns the algorithmic

    derivation of mathematical and computational models from

    BG into a highly formalized task [2]. The conjunction of all

    these features make of BG a physically based, object-oriented

    graphical language most suitable for dynamic modeling,

    analysis and simulation of complex engineering systems

    involving mixed physical and technical domains in their

    constitution [7].

    The vehicle model developed in this paper considers the

    vertical, longitudinal and lateral vehicle dynamics, takes into

    account the geometrical non-linearities associated to the

    suspension system, and includes Pacejka models [14] for thebehavior of the pneumatic tires. Simple, adequate models for

    the engine and the transmission are also included to take into

    account the vehicle traction.

    The rest of the paper is organized as follows. Section II

    presents a brief general description of a four-wheel vehicle and

    of its constituents seen as rigid bodies, and discusses the

    modeling assumptions. Section III first deals with the standard

    mathematical modeling of these components and then with

    their bond graph modeling and discusses the construction of

    the 20sim library [18]. Section IV presents the full vehicle

    M

    mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]
  • 8/4/2019 Vehicle Modeling Using BG

    2/11

    model obtained by assembling the library components. Section

    V presents the simulation results and, finally, Section VI

    brings the conclusions.

    II. GENERALITIES AND MODELING ASSUMPTIONSA.Vehicle Chassis

    The vehicle chassis is modeled as a rigid body with a local

    coordinate reference frame (x, y, z) attached to the center ofmass and aligned with the inertia principal axes as is shown in

    figure 1. It has mass m, and the following principal inertia

    moments: roll (Jr) respect to the body x-axis,pitch (Jp) respect

    to the body y-axis andyaw (Jy) respect to the body z-axis.

    Figure 1: Full car model.

    As also the four suspension subsystems are modeled as

    spatial multibody systems, joint models are necessary to link

    them to the chassis. The joints are represented as flexible

    instead of rigid using a pseudo spring-damper system withelastic and damping constants. The flow and efforts actuating

    on the joints depend on the relative position and the relative

    orientation among the bodies.

    In order to link two rigid bodies at a given joint it is

    necessary to do some transformations (translations and

    rotations) between the reference frames associated to each

    body. In this way, the state variables expressed at the center of

    mass of each body are transformed to a local reference frame

    attached to the joint.

    B.Engine and TransmissionMost vehicles are propulsed by internal (spark or

    compression ignited) combustion engines which -for ourmodeling purposes- may be modeled through a given static

    curve relating the engine speed and the load with its torque

    and its power. Usually, these curves are obtained through

    testing the engine at partial and full load.

    The transmission is composed by the mechanical members

    connecting the engine crankshaft with the traction wheels, the

    gearbox and the planetary gear train including the differential

    [8]. The main phenomena taken into account are the speed and

    effort transformation among them.

    C.Pneumatic tiresAside from aerodynamic and gravitational forces, all other

    major forces and moments affecting the motion of a ground

    vehicle are applied through the running gear-ground contact.

    An understanding of the basic characteristics of the interaction between the running gear and the ground is, therefore,

    essential to the study of performance characteristics, ride

    quality, and handling behavior of ground vehicles. However, a

    detailed explanation about pneumatic tires is out of the scopeof this paper. The following figure shows a summary of the

    main forces, moments and angles that play a major role on themodeling of the pneumatic tire. Each one is associated with a

    corresponding local axis located at the center of the contact

    patch of the wheel. Traction (Fx), lateral (Fy) and normal (Fz)

    forces along X, Y and Z local axis respectively and overturning

    (Mx), rolling resistance (My) and aligning (Mz) torques alongthe same axis respectively are modeled in terms of the slip and

    the camber angles and the pneumatic characteristics.

    For more details about the fundamentals and the modeling ofpneumatic tires see [14, 1, 12].

    Figure 2: Tire axis reference system

    D.SuspensionsSuspension is the term given to the system of springs, shock

    absorbers and linkages that connects a vehicle to its wheels.

    This assembly is used to support weight, absorb and dampen

    road shock, and help maintain tire contact as well as proper

    wheel-to-chassis relationship. Without being a restriction for

    future extensions of the overall vehicle model, only static

    suspension systems are considered in this work.

    E.Aerodynamic forcesThe aerodynamic forces were very simplified in this

    analysis through the usage of empirical aerodynamic

    coefficients [1, 16, 17]. In this work only the drag force was

    included, being the lift and the pitching effects very similar in

    terms of their mathematical expressions. In the future not only

    the coefficients, also their sensitivity may be included.

    III. MATHEMATICAL AND BGMODELINGThis section presents the subsystem modeling according to

    the hypothesis of Section II. Previously, the essential issues

    concerning the bond graph modeling of multibody systems as

    Fx

    x

    y

    z

    roll

    yaw

    pitch

    Fy

    Kt

    Kt

    Kt

    Kt

    Bs

    Bs

    Bs

    Bs

    Ks

    Ks

    Ks

    Ks

    (1)

    (2)

    (3)

    (4)

    1r

    1h

  • 8/4/2019 Vehicle Modeling Using BG

    3/11

    used in this paper are addressed.

    To determine the spatial motion of the rigid body the well

    known Euler equations are used [2], which appear in (1) and

    (2) in both, their intrinsic way of representation and their

    tensorial counterparts.

    The first one represents the conservation of linear

    momentum, written as:

    kjijk

    rel

    ii

    i

    rel

    pdt

    dp

    dt

    dpF

    pdt

    pd

    dt

    pd

    F

    +==

    +== (1)

    where pF ,, represents the external forces, the angular

    velocity vector and the linear momentum vector respectively;

    d/dt, d/dt|rel, ijk represent the derivative respect to the inertialframe, the derivative respect to the body (vehicle) attached

    frame and the Levi-Civita tensor used to express the cross

    product in tensor notation.

    The second of the Euler equations sets up the conservation

    of angular momentum:

    kjijk

    rel

    ii

    i

    rel

    hdt

    dh

    dt

    dhM

    hdt

    hd

    dt

    hdM

    +==

    +==(2)

    where hM, represent the external torque and the angular

    momentum vector.

    Figure 3: BG representation of the spatial rigid body

    dynamics.

    The BG representation of the 3-dimensional motion of a

    rigid body based on the Euler equations is shown in figure 3.

    The port variables of the above model are defined with respect

    to the system attached to the center of mass of the rigid body.

    Referring these port variables to the coordinates of

    interconnection to other bodies is a must in order to be able to

    couple the corresponding models.

    Figure 4 shows the bond graph implementation of the

    equation system representing how the port variables of two

    arbitrary points 'A' and 'B' of a given spatial body transform

    each other. The equations relating the linear and rotational

    efforts are the following:

    kjijki rFM

    rFM

    =

    = (3)

    For the flow variables the equations are the following:

    kjijki rv

    rv

    =

    =(4)

    Figure 4: Power variables transformation between two points

    A, Bbelonging to a given 3-dimensional rigid body.

    To transform the dynamic equations from those expressed in

    the body attached frame of reference (roll, pitch and yaw axes)

    to a spatially fixed frame of reference (X,Y,Z : inertial frame)

    it is necessary to choose some parameterization for the

    rotations. Among the multiple possibilities, Euler angles are

    used in this work. To transform these rotations the following

    equations are used:

    = (5.a); = (5.b); =G (5.c)

    vv = (5.d); vv = (5.e); vv G = (5.f)

    where,

    =

    cossin0

    sincos0

    001

    =

    cos0sin

    010

    sin0cos

    =

    100

    0cossin

    0sincos

  • 8/4/2019 Vehicle Modeling Using BG

    4/11

    =

    z

    y

    x

    =

    z

    y

    x

    =

    z

    y

    x

    =

    Z

    Y

    X

    G

    =

    z

    y

    x

    v

    v

    v

    v

    =

    z

    y

    x

    v

    v

    v

    v

    =

    z

    y

    x

    v

    v

    v

    v

    =

    Z

    Y

    X

    G

    v

    v

    v

    v

    Its bond graphs representation is observed in figure 5, where

    the power variables are transformed from (x,y,z) axes to the

    rotated ones (X,Y,Z) according to the angles , , respect to

    each axis. It may be observed that while the flow variables are

    rotated from (xyz) to (XYZ), the effort variables are

    transformed back from (XYZ) to (xyz).

    Next, the models used for the different subsystems

    belonging to the whole vehicle model are presented [2, 12]. Figure 5: 3D-rotation equations expressed in terms of power

    variables.

    Figure 6: BG modeling of Vehicle Chassis

    A.Vehicle ChassisFigure 6 shows the vehicle chassis model composed by a

    rigid body model, a coordinate system transformation to the

    global system where the vehicle weight is imposed and four

    translational transformations ( ir) to the pivots of the

    suspension at each wheel, each one with their corresponding

    model Engine and Transmission

    B.Engine and TransmissionEngine. In bond graphs representation the engine is

    modeled as an effort source (the engine output torque) variablewith the engine velocity normally expressed in rpm

    (revolutions per minute) and the position of the butterfly valve

    (accelerator command) as written by equation (6).

    )()1()(),,( rpppprp TATAATTT += (6)

    where Ap is the accelerator position ( 10 pA , Tp and Trare the engine output torque and the resistant torque

    respectively for a given engine speed ( ) .

  • 8/4/2019 Vehicle Modeling Using BG

    5/11

    While the engine curve, the accelerator position and the

    resistant torque are data for the model, the engine speed is

    computed by the whole model. In the Engine Torque model

    show in figure 7 the output torque is computed by equation (6)

    and its result Tis used to modulate the source MSe.

    Figure 7: BG modeling of Engine

    Figure 8: BG modeling of Transmission

    Transmission. Figure 8 shows the main components of the

    transmission and its structure. The gearbox (figure 9) is

    modeled using a modulated transformer (MTF) that relates the

    input and output effort variables through a variable

    transformation factor that depends on the gearbox ratios

    supplied to the model in advance.

    Figure 9: BG modeling of Gearbox

    The differential (figure 10) is modeled by a transformer(TF) modulated by the planetary drive train (differential) ratio.

    The 0 junction imposes the same torque to both traction

    wheels.

    Figure 10:BG modeling of Differential

    C.Pneumatic TiresFor all forces and moments acting on the tire the Pacejka

    model [14], inspired by a lot of experiments carried out using

    different types of pneumatic tires is used.

    ( )

    ( )

    2

    1/

    ( ) 1

    /

    1.12 ; 0.625 ; 1

    46 ; 5 ; 0.6

    b

    n

    x sign A e c D

    B K d

    A C D

    K d n

    = +

    =

    = = =

    = = =(7)

  • 8/4/2019 Vehicle Modeling Using BG

    6/11

    This model is briefly described with the expressions (7),

    with -1<

  • 8/4/2019 Vehicle Modeling Using BG

    7/11

    Figure 12: Bond graphs modeling of Suspension

    Figure 13: Bond graphs modeling of Revolution Joint

    E.Aerodynamic ForcesEquation 9 represents the aerodynamic drag force where

    is the air density, Cx is the (experimentally computed) drag

    coefficient,Af is the vehicle frontal area and V is the relative

    velocity between the vehicle and the wind.

    2

    2

    1

    VACF fxaero

    x = (9)

    This is represented in the Bond Graph formalism by a

    resistance component applied to the coordinate system fixed

    to the center of mass of the rigid body that models the chassis,

    as it is shown in figure 14.

    Figure 14: Bond graphs modeling of Rigid Body with

    Aerodynamics Resistance

    IV. FULL VEHICLE VECTOR BGMODEL AND 20SIMIMPLEMENTATION

    In figure 15 the whole system model is shown, which isbuilt via assembling the previous submodels, pretty much in

    the same way as a real vehicle is constructed. This is the

    powerful modular or objected-oriented modeling propertyof the BG technique.

  • 8/4/2019 Vehicle Modeling Using BG

    8/11

    Figure 15: Bond graphs modeling of a vehicle

    V.SIMULATION RESULTSAfter the validation of the bond graphs multibody

    toolbox developed in this work, a vehicle dynamics test,

    extracted from the bibliography is performed [1].

    The data, coming from a Renault Clio RL 1.1 car, are the

    following:

    Aerodynamics coefficient 0.33

    Frontal area 1.86 m2

    Distance between axes 2.472 mVehicle weight 8100 N

    Centre of mass height 0.6 m

    Front axis weight 5100 N

    Rear axis weight 3000 N

    Maximum engine torque 78.5 Nm at 2500 rpm

    Maximum engine power 48 CV at 5250 rpm

    Planetary drive train (differential) ratio 3.571

    First gearbox ratio 3.731

    Second gearbox ratio 2.049

    Third gearbox ratio 1.321

    Four gearbox ratio 0.967

    Five gearbox ratio 0.795

    Reverse gearbox ratio 3.571Tires, type and dimensions 145 70 R13 S

    Wheelbase 1.650 m

    Maximum speed 146 km/h

    Acceleration 0-100 km/h in 17 s

    Time spent to do 1000 meters 38 s

    Distance from centre of mass to front axes 0.916 m

    Distance from centre of mass to rear axes 1.556 m

    Pneumatic tire radius (unloaded) 0.2666 m

    Air density 1.225 kg/m3

    Unsprung masses (at each wheel) 38.42 kg

    Tire Vertical stiffness 150 000 N/m

    Tire inertia 1.95 Kgm2

    Damper coefficient 475 N s / m

    Suspension stiffness 14 900 N/m

    Sprung mass - Yaw Inertia 2345.53Kg m2

    Sprung mass - Pitch Inertia 2443.26Kg m2

    Sprung mass - Roll Inertia 637.26 Kg m2

    The first test is a sudden motion starting from rest. With

    the engine butterfly valve fully opened and once the clutch

    is released the vehicle response may be assessed in terms of

    the vehicle acceleration in a straight road. Figures 16 to 21

    show the results for this case. Figure 16 shows the engine

    speed (rpm) in time where it may be noted the times at

    which the gearbox is used. Figure 17 shows the longitudinal

    vehicle speed in time reaching 100 km/h in less than 20

    seconds. Figure 18 plots the longitudinal sliding of one of

    the traction wheels. The friction force may be evaluated

    from this figure and the vertical load on the tire (see figure

    19). In figure 20 the chassis pitching angle is shown.

    During the first 10 seconds until the fourth gear is selected

    the pitching is oscillating reaching a negative smaller value

    after this time when five gear is selected showing an

    acceleration behavior. Figure 21 shows the rear wheel load.The time spent to reach 100 km/h was 17 seconds, agreeing

    very well with the road test published by the manufacturer.

  • 8/4/2019 Vehicle Modeling Using BG

    9/11

    Figure 16: Time evolution of the engine speed.

    Figure 17: Longitudinal vehicle velocity [m/s] in time.

    Figure 18: Evolution of one of the traction wheels

    sliding.

    Figure 19: Evolution of the load [N] over a traction

    wheel.

    Figure 20: Chassis pitching as a function of time.

    Figure 21: Load [N] over one of the rear wheels.

    In the second test the vehicle is forced to follow a curved

    road. The simulation starts with the vehicle in third gear at

    65 km/h; the wheel drive in turned such that the front

    wheels turn 1 degree in 10 sec following a time law as

    shown in figure 22.

    Figure 22: Evolution of the turning angle [rad] of the

    front wheels.

    Figure 23: Trajectory in x-y [m]

  • 8/4/2019 Vehicle Modeling Using BG

    10/11

    Figure 24: Sliding angle of one of the front wheels as a

    function of time.

    Figure 25: Sliding angle of one of the rear wheels as a

    function of time.

    Figure 26: Yaw vehicle response as a function of time.

    After 5 seconds the wheel drives return to the original

    position (0 degree). In figure 24 the lateral slip angle of one

    of the front wheels is shown and in figure 25 the same for

    one of the rear wheels. Figure 13 plots the trajectory

    followed by the vehicle and figure 26 the yaw response ofthe vehicle.

    Figure 27: Front wheels slip angle [rad] as a function of

    time

    Figure 28: Trajectory x-y [m]

    Figure 29: Slip angle of one of the front wheels [rad].

    Figure 30: Slip angle of one of the rear wheels [rad].

  • 8/4/2019 Vehicle Modeling Using BG

    11/11

    Figure 31: Yaw angle [rad] as a function of time.

    In the third test a zigzag maneuver is assessed. The

    vehicle start the simulation at 65 km/h in third gear the

    driver turn the wheel drive following a time law as shown

    in figure 27, the front wheels turn an angle of one degree in

    10 seconds.

    After this time the wheel drive is turned as to reach one

    degree in the opposite direction in 10 seconds and then

    return to the original position in the remaining 5 seconds.Figure 29 shows the lateral slip angle of one of the front

    wheels and figure 30 does the same for one of the rear

    wheels.

    Figure 28 shows the vehicle trajectory and figure 31

    shows the vehicle yaw angle.

    The behavior of this case presents some similarity with

    the second one and it may be observed that in both cases the

    directional stability is a priori acceptable.

    However to do a more rigorous analysis we should

    analyze the eigenvalues of the directional response matrix

    that may be computed making a sensitivity analysis with

    simulations. This task is left for future work.

    For the vehicle application it is observed that formaneuvering in straight road the response of the vehicle

    with third, four and fifth gears were very smooth

    concluding that the vehicle is well optimized to be used at

    velocities greater than 50km/h. For curves the transient

    response allows for evaluating the directional stability with

    changes in the position of the wheel drive. According to the

    results obtained it would be more desirable to have a less

    oscillatory behavior in turns.

    VI. CONCLUSIONSOne of the main goals of this paper was the extension of

    this formalism to include large spatial (3-dimensional)rotations. Several elements oriented to multibody systems

    were developed allowing working with different reference

    frames, operating with them through the usage of

    translations and general transformations. This toolbox

    works acceptable in the vehicle dynamics prediction and it

    was successfully applied to another project based on vehicle

    fault diagnostics [13].

    REFERENCES

    [1] F. Aparicio Izquierdo, C. Vera Alvarez, V. Das Lpez, Teora de los

    vehculos automviles. U. P. de Madrid[2] Ronald C. Rosenberg, Donald L. Margolis, Dean C. Karnopp.

    Modeling and Simulation of Mechatronic Systems, A Wiley-Interscience

    Publication, New York.

    [3] Albert M. Bos. Modeling Multibody Systems in terms of MultibondGraphs, with application to a motorcycle, PhD thesis at Twente Univ. 1986

    [4] Ahmed A. Shabana. Dynamics of Multibody systems. A Wiley-Interscience Publication.

    [5] Jinhee Jang, Changsoo Han. Proposition of a Modeling Method forConstrained Mechanical Systems Based on the Vector Bond Graph.

    Journal of the Franklin Institute. Vol. 335B, No 3, pp. 451 469, 1998.[6] T. Erial, J. Stein, L. Louca. A Bond Graphs Based Modular Modeling

    Approach towards an Automated Modeling Environment for

    Reconfigurable Machine Tools. IMAACA 2004

    [7] F. Cellier, "Hierachical nonlinear bond graph: A unified methodologyfor modelling complex physical systems", Simulation, Vol 58, No. 4, pp.

    230-248.[8] J. M. Mera, C. Vera, J. Flez. 2WD Power Train Modeling with Bond

    Graph applied to Vehicular Dynamics . Universidad Politcnica de

    Madrid, Spain.

    [9] W. Kortuem , A. Urzt, Simulation of active suspensions in ground

    transportation - Application to Maglev vehicles, 11th IMACS world

    congress, Oslo, Norway, 5-9 Aug, 1985[10] B. Paul, Analytical dynamics of mechanisms - a computer oriented

    overview, Mechanism and Machine Theory, vol 10, pp. 481-507, 1975.

    [11] B. Paul, Computer oriented analytical dynamics of machinery,

    Proceedings of the NATO advanced study institute on computer aidedanalysis and optimization of mechanical system dynamics, ed. E.J. Haug,

    IOWA City, Aug 1983, Springer, pp. 41-87, 1984

    [12] G. Filippini. Dinmica Vehicular mediante bond graphs. Proyecto

    final de carrera de grado. Escuela de Ingeniera Mecnica, Universidad

    Nacional de Rosario, 2004[13] D. Delarmelina, L. Silva, S. Junco. Fault Diagnosis in Vehicle based

    on its Dinamic Model. Escuela de Ingeniera Mecnica y Escuela de

    Ingeniera Electronica, Universidad Nacional de Rosario, 2005.

    [14] H. Pacejka. Tyre Modeling for Use in Vehicle Dynamics Studies.SAE Paper, No. 870421.

    [15] H.B. Pacejka, Principles of plane motion of automobiles, IUTAMsymposium on the dynamics of automobiles, Delft Univ. of Tech. pp. 33-

    59, 1975

    [16] W.H. Hucho, editor. Aerodynamics of road vehicles. SAE, (1998).

    [17] R. H- Barnard, Road vehicle aerodynamic design, Second Edition,MechAero, England, (2001)

    [18] Getting Started with 20-sim 3.6, Controllab Products B.V., Enschede,

    Netherlands. Internet: www.20sim.com, 2005.