Sunteți pe pagina 1din 27

This article was downloaded by: [Cinvestav del IPN] On: 17 April 2013, At: 10:29 Publisher: Taylor

& Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Vehicle System Dynamics: International Journal of Vehicle Mechanics and Mobility


Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/nvsd20

Understanding the limitations of different vehicle models for roll dynamics studies
Taehyun Shim & Chinar Ghike
a a a

Department of Mechanical Engineering, The University of Michigan-Dearborn, 4901 Evergreen Road, Dearborn, MI, 48128, USA Version of record first published: 30 Sep 2010.

To cite this article: Taehyun Shim & Chinar Ghike (2007): Understanding the limitations of different vehicle models for roll dynamics studies, Vehicle System Dynamics: International Journal of Vehicle Mechanics and Mobility, 45:3, 191-216 To link to this article: http://dx.doi.org/10.1080/00423110600882449

PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.tandfonline.com/page/terms-andconditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

Vehicle System Dynamics Vol. 45, No. 3, March 2007, 191216

Understanding the limitations of different vehicle models for roll dynamics studies
TAEHYUN SHIM* and CHINAR GHIKE
Department of Mechanical Engineering, The University of Michigan-Dearborn, 4901 Evergreen Road, Dearborn, MI 48128, USA
An accurate and realistic vehicle model is essential for the development of effective vehicle control systems. Many different vehicle models have been developed for use in various vehicle control systems. The complexity of these models and the assumptions made in their development depend on their application. This article looks into the development and validity of vehicle models for prediction of roll behavior and their suitability for application in roll control systems. A 14 DOF vehicle model that includes dynamics of roll center and nonlinear effects due to vehicle geometry changes is developed. The limitations, validity of simplied equations, and various modeling assumptions are discussed by analyzing their effect on the model roll responses in various vehicle maneuvers. A formulation of the popular 8 DOF vehicle model that gives good correlation with the 14 DOF model is presented. The possible limitations of the 14 DOF model compared with an actual vehicle are also discussed. Keywords: Vehicle dynamics; Rollover; Roll center; Suspension

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

Nomenclature Vehicle a b c h hrcf hrcr Jx Jy Jz m u/v/w

distance of c.g. from front axle (m) distance of c.g. from rear axle (m) track width (m) c.g. height (m) front roll center distance below sprung mass c.g. (m) rear roll center distance below sprung mass c.g. (m) roll inertia (kg m2 ) pitch inertia (kg m2 ) yaw inertia (kg m2 ) vehicle sprung mass (kg) longitudinal/lateral/vertical velocities of c.g. in body-xed coordinate (m/s) pitch angle (rad) roll angle (rad)

*Corresponding author. Email: tshim@umich.edu

Vehicle System Dynamics ISSN 0042-3114 print/ISSN 1744-5159 online 2007 Taylor & Francis http://www.tandf.co.uk/journals DOI: 10.1080/00423110600882449

192

T. Shim and C. Ghike

x /y /z

yaw angle (rad) roll rate/pitch rate/yaw rate of c.g. in body-xed coordinate (rad/s)

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

Suspension/tire bs suspension damping coefcient (Ns/m) Fx t /Fy t /Fzt tire longitudinal/lateral/vertical forces (N) longitudinal/lateral/vertical forces at tire contact patch in coordinate Fx g /Fy g /Fzg frame 2 (N) Fx gs /Fy gs /Fzgs longitudinal/lateral/vertical forces at tire contact patch in coordinate frame 1(N) Fx s /Fy s /Fzs longitudinal/lateral/vertical forces transferred to body in coordinate frame 1(N) Jw rotational inertia of each wheel (kg m2 ) ks suspension spring stiffness (N/m) kt tire stiffness (N/m) ls instantaneous length of strut (m), ls = lsi xs + xsi lsi initial length of strut (m), lsi = h (r xti ) mu unsprung mass (kg) r instantaneous tire radius (m) ro nominal radius of tire (m) s tire longitudinal slip T external torque applied at wheel (N m) ug /vg longitudinal/lateral velocities at tire contact patch in coordinate frame 2 (m/s) us /vs /ws longitudinal/lateral/vertical velocities at the suspension corner in coordinate frame 1 (m/s) uu /vu /wu unsprung mass longitudinal/lateral/vertical velocities in coordinate frame 1 (m/s) xs suspension spring compression (m) xt tire spring compression (m) xsi initial suspension spring compression (m) xti initial tire spring compression (m) tire lateral slip (rad) angular velocity of wheel rotation (rad/s) road wheel steer angle (rad)

1.

Introduction

With the growing popularity of sport utility vehicles (SUVs) and the increasing number of rollover accidents in SUVs, as well as trucks and buses, the reduction of vehicle rollover propensity has been an area of active research in recent years. Owing to the high fatality rate of rollover crashes, National Highway Trafc Safety Administration initiated a rollover resistance rating test as part of its New Car Assessment Program for passenger vehicles, including SUVs [1, 2]. A number of methods have been proposed and explored to prevent vehicle rollovers. Research on such rollover prevention systems can be categorized into two basic types: passive systems (e.g. curve speed warning systems) and active systems (e.g. roll stability control). Many passive warning systems use a prediction algorithm to determine the risk of impending

Limitations of different vehicle models

193

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

rollover using vehicle or axle roll angles, lateral load transfer, and/or lateral acceleration as inputs [310]. They provide some type of warning so that the driver can take corrective action. The active roll control systems reported in the literature [1116] can be categorized into about four different types the basis of on its actuation schemes: four wheel steering [11], active suspension [12, 13], active roll bar [14, 15], and differential braking [1618]. The active roll bar and active suspension are designed to directly control the vehicle roll motion; the four wheel steering and differential braking are designed to reduce the vehicle rollover tendency by controlling vehicle yaw motion. In the development of active/passive roll control systems, a vehicle model that can represent realistic roll behavior is essential to predict impending rollover as well as accurately applying the actuation force to avoid vehicle rollover. In the literature, a number of vehicle models have been introduced for the application of active/passive roll control systems [810, 1924]. The majority of these models are low-order roll-plane models derived with various assumptions and have limitations for their application. Figure 1 shows some roll-plane models such as a rigid vehicle model and a suspended vehicle model. The rigid vehicle model is the most common roll-plane model and is widely used due to its ease of understanding and parameterization. It assumes that there are no compliances (i.e. frame, suspension, or tire) and no degrees of freedom (prior to lift-off). It represents a theoretical upper bound on vehicle stability [21, 22]. The suspended vehicle model [810, 1921, 23, 24] differs from the rigid vehicle model in that it has a roll degree of freedom for the suspension that connects the sprung and unsprung masses. The sprung mass is typically assumed to rotate about a kinematic roll center axis [19], which connects the front and rear suspension roll centers. This model also assumes that the sprung mass is much greater than the unsprung mass and that the chassis and tires are much stiffer than the suspension. These models are based on a vehicle model which has a decoupled yaw-roll motion and one equivalent axle. In order to better represent the vehicle lateral and yaw dynamics as well as coupling of yaw-roll motion due to the transient lateral load transfer during extreme maneuvers, higher order models such as 8 DOF [2527] and 14 DOF [2831] are also used in rollover studies. A 14 DOF vehicle model, which considers the suspension at each corner, has the same benets of an 8 DOF vehicle model, with the additional capabilities of predicting vehicle pitch and heave motions. It also offers the exibility of modeling nonlinear springs and dampers and can simulate the vehicle responses to normal force inputs in the case of an active suspension system. Moreover, the 14 DOF model, unlike the 8 DOF model,

Figure 1.

Schematic of simple rigid and suspended vehicle models used in the rollover study.

194

T. Shim and C. Ghike

can predict vehicle behavior even after wheel lift-off and thus can be used in developing or testing the validity of rollover prediction/prevention strategies. In other words, even though the 14 DOF model has low degrees of freedom and consequently certain limitations when compared with a multi-body dynamics model, it can sufciently express the vehicle motions that are important in most active chassis control systems. This article looks into the development and validity of vehicle models for prediction of roll behavior and their suitability for application in roll control systems. A 14 DOF vehicle model that includes dynamics of roll center and nonlinear effects due to vehicle geometry changes is developed. The limitations, validity of simplied equations, and various modeling assumptions are discussed by analyzing their effect on the model roll responses in various vehicle maneuvers. A formulation of the popular 8 DOF vehicle model that gives good correlation with the 14 DOF model is presented. The possible limitations of the 14 DOF model compared with an actual vehicle are also discussed.

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

2.

Development of vehicle model (model 1)

Figure 2 exhibits the schematic of the two axle, 14 DOF, vehicle model used to investigate vehicle roll response to steering and torque inputs. This schematic includes 6 DOF at the vehicle lumped mass center of gravity and 2 DOF at each of the four wheels, including vertical suspension travel and wheel spin. The body is modeled as being rigid, with body-xed coordinates, xyz, attached at the center of gravity and aligned in principal directions (coordinate frame 1). u, v , and w indicate forward, lateral, and vertical velocities, respectively, of the sprung mass. There is roll angular velocity, x , pitch angular velocity, y , and yaw angular velocity, z . The attitude and position of body with respect to the inertial frame (XYZ ) can be determined through successive coordinate transformations through the cardan angles (i.e. the roll angle , the pitch angle , and the yaw angle ) as shown in gure 2. The coordinate frame 2 is obtained by rotating the inertial coordinate frame through the yaw angle . In other words, the body-xed coordinate frame 1 is obtained by rotating the coordinate frame 2 rst through the pitch angle and then through the roll angle . Figure 3 shows the force and velocity components in the right front corner of a vehicle. The velocities usrf , vsrf , and wsrf are the velocities of the right front strut mounting point in the longitudinal, lateral, and vertical directions, respectively, in the body-xed coordinate frame,

Figure 2.

Schematic of 14 DOF full vehicle model with one-dimensional suspension and coordinate frames.

Limitations of different vehicle models

195

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

Figure 3.

Description of forces and velocities at the right front corner of a vehicle.

which is attached to the sprung mass center of gravity (coordinate frame 1). These velocities can be obtained by transforming the c.g. velocities as c 0 0 2 x usrf u vsrf = 0 0 a y + v (1) c w wsrf z a 0 2 The velocities uurf , vurf , and wurf represent the velocities of the unsprung mass mu in the body-xed coordinate frame 1, and ugrf , vgrf , and wgrf are the lateral, longitudinal, and vertical velocities at the tire contact patch in coordinate frame 2. The forces Fxsrf , Fysrf , and Fzsrf are the forces transmitted to the sprung mass along the longitudinal, lateral, and vertical directions, respectively, of coordinate frame 1. The forces Fxgsrf , Fygsrf , and Fzgsrf are the forces acting at the tire ground contact patch in the same coordinate frame 1. These forces can be written in terms of the tire forces Fxgrf , Fygrf , and Fzgrf by projecting its components along coordinate frame 2 as Fx gsrf 1 0 0 cos 0 sin Fx grf Fy gsrf = 0 cos sin 0 1 0 Fy grf (2) 0 sin cos sin 0 cos Fzgsrf Fzgrf The forces Fx grf and Fy grf are obtained by resolving the longitudinal (Fx trf ) and cornering (Fy trf ) forces at the tire contact patch as Fx grf = Fx trf cos Fy trf sin Fy grf = Fy trf cos + Fx trf sin (3) (4)

where is the steering angle at the road wheels. The Magic Formula [32] tire model has been used in the development of the tire forces Fx trf and Fy trf . The longitudinal and lateral slips used in the tire model were calculated as follows: srf = (rrf rf (ugrf cos + vgrf sin )) |(ugrf cos + vgrf sin )| vgrf ugrf (5) (6)

rf = tan1

196

T. Shim and C. Ghike

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

Figure 4.

Bond graph representation of right front suspension.

Figure 4 shows the bond graph representation of the right front suspension shown in gure 3. The longitudinal and lateral velocities at the tire contact patch, ugrf and vgrf , can be determined from gure 4 as ugrf = cos (uurf y rrf ) + sin (wurf cos + sin (x rrf + vurf )) vgrf = cos (vurf + x rrf ) wurf sin (7) (8)

where rrf is the instantaneous radius of the tire. The longitudinal (uurf ) and lateral (vurf ) velocities of the unsprung mass in the body-xed coordinate frame used in the earlier equations are simply written as uurf = usrf lsrf y vurf = vsrf + lsrf x (9) (10)

where lsrf is the instantaneous length of the strut as indicated in gure 3. The unsprung mass vertical velocity wurf represents the degree of freedom corresponding to the suspension deection and can be expressed by applying Newtons law for the vertical motion of the unsprung mass as urf = cos (cos (Fzgrf mu g) + sin Fx grf ) sin Fy grf Fdzrf mu w xsrf ksf x srf bsf mu (vurf x uurf y ) (11)

where ksf is the suspension stiffness, bsf the suspension damping coefcient, and xsrf the instantaneous compression of the right front suspension spring. The force, Fdzrf , represents

Limitations of different vehicle models

197

the additional load transfer that occurs at the wheels through the suspension links because of the reaction force to the force transmitted to the sprung mass through the roll center. This will be considered in detail at a later stage. The instantaneous suspension spring deection xsrf is given as x srf = wsrf + wurf (12)

The vertical force Fzgrf acting at the tire ground contact patch in coordinate frame 2 can be written in terms of the tire stiffness (ktf ) and the instantaneous tire deection (xtrf ) as Fzgrf = Fztrf = xtrf ktf The instantaneous tire deection xtrf in equation (13) is given as (13)

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

x trf = wgrf wuirf = wgrf (cos (wurf cos + vurf sin ) uurf sin )

(14)

where wuirf is the vertical velocity of the wheel center in the inertial coordinate frame. For the simulations in this article, it is assumed that the vertical velocity wgrf at the tire contact patch is zero (smooth road). It should be noted that even though the tire is assumed to remain at a xed angle with the strut, the vertical stiffness of the tire, ktf , is always considered to be normal to the ground between the ground and the wheel center ( point C). Thus, even though the compliance element 1/ktf is located between the one junctions of the ground vertical velocity (wgrf = 0) and the velocity w1rf in gure 4, the vertical velocity of the wheel center in the inertial frame, wuirf , is used in place of the velocity w1rf in equation (14). The instantaneous tire radius is then determined as ro xtrf rrf = (15) cos cos To account for the wheel lift-off, when the tire radial compression becomes less than zero, the tire normal force Fzgrf is set equal to zero. In addition, the instantaneous tire radius is considered equal to the nominal tire radius until the tire returns to the road surface. If xtrf < 0 then Fzgrf = 0 and rrf = r0 (16)

As the tire normal force becomes zero, no lateral (Fy grf ) and longitudinal (Fx grf ) tire forces are developed at that contact patch. Thus, the only forces acting at that suspension corner are the weight and inertia forces of the unsprung mass, which are very small in magnitude. As the cardan angles and appropriate coordinate transformations between the body-xed coordinate frame 1 and the coordinate frame 2 at the tireground contact patch (gure 2) are considered for all the forces and velocities in the system, the model is able to simulate vehicle behavior after wheel lift-off and during the rollover event with just the modication mentioned in equation (16). The instantaneous length of the strut lsrf used in equations (9) and (10) is given as lsrf = lsif (xsrf xsif ) (17)

where lsif is the initial length of strut and xsif is the initial suspension spring deection. The initial length of the strut lsif is taken such that lsif = h (ro xtif ) where xtif is the initial tire compression. (18)

198

T. Shim and C. Ghike

The initial spring compression xsif and the initial tire compression xtif are determined from the static conditions as xsif = xtif mb 2(a + b)ksf (mb/2(a + b)) + muf = ktf (19) (20)

The forces Fx srf and Fy srf transmitted to the sprung mass along the u- and v -axes of the bodyxed coordinate frame are obtained after subtracting the components of the unsprung mass weight and inertia forces from the corresponding forces Fx gsrf and Fy gsrf acting at the tire contact patch as

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

Fx srf = Fx gsrf + mu g sin mu u urf + mu z vurf mu y wurf Fy srf = Fy gsrf mu g sin cos mu v urf + mu x wurf mu z uurf The vertical force Fzsrf transmitted to the sprung mass through the strut is given as Fzsrf = xsrf ksrf + x srf bsrf

(21) (22)

(23)

Figure 5 shows the forces and velocities in the roll plane of, for example, the front suspension. Generally, the roll center height is dened with reference to the ground. However, for the development of this model, the front and rear roll centers are assumed to be xed at distances hrcf and hrcf , respectively, below the sprung mass c.g. along the negative w -axis of the bodyxed coordinate frame 1. Moreover the roll center is simply considered to be a point of application of the forces transmitted to the sprung mass through the suspension links and not as a kinematic constraint. In gure 5, Fzslf and Fzsrf are the forces transmitted to the sprung mass through the struts. Fy slf and Fy srf represent the lateral forces transmitted to the sprung mass through the suspension links. In the absence of a roll center, i.e. when the roll center is assumed to be in the ground plane, the total roll moment transmitted to the sprung mass at, for example, the right front

Figure 5.

Forces and velocities in the front suspension roll plane.

Limitations of different vehicle models

199

corner along the x direction is given as Mx rf = Fy gsrf (lsrf + rrf ) (mu g sin cos + mu v urf mu x wurf + mu z uurf ) lsrf = Fy gsrf rrf + Fy srf lsrf and the force due to the lateral load transfer through suspension links, Fdzrf = 0, (mu g sin + mu u urf mu z vurf + mu y wurf ) When a roll center is modeled as shown in gure 5, the roll moment Mx rf transmitted to the sprung mass by the right front corner suspension is given as (24)

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

Mx rf = Fy srf hrcf

(25)

Thus, the inclusion of a roll center reduces the total roll moment transferred to the sprung mass by the front suspension. The difference between the roll moments in the absence of the roll center (equation (24)) and when the roll center is considered (equation (25)) acts directly on the unsprung mass and is responsible for the link load transfer forces ( jacking forces), Fdzlf and Fdzrf . These forces can be estimated as Fdzrf = Fdzlf = Fy gsrf rrf + Fy srf lsrf + Fy gslf rlf + Fy slf lslf (Fy srf + Fy slf )hrcf cf (26)

The moments My rf and Mzrf transmitted to the sprung mass at, for example, the right front corner by the suspension along the y and z directions can be given as My rf = (Fx sgrf (lsrf + rrf ) (mu g sin + mu u urf mu z vurf + mu y wurf ) lsrf ) = (Fx sgrf rrf + Fx srf lsrf ) Mzrf = 0 (27) (28)

The equations of motion for the 6 DOF of the sprung mass model can now be derived from the direct application of Newtons laws for the system as m(u + y w z v) = m(v + z u x w) = m(w + x v y u) = x = Jx Jy y = Jz z = + (Fx sij ) + mg sin (Fy sij ) mg sin cos (Fzsij + Fdzij ) mg cos cos (Mx ij ) + (Fzslf + Fzslr Fzsrf Fzsrr )c 2 (29) (30) (31) (32) (33)

(My ij ) + (Fzslr + Fzsrr )b (Fzslf + Fzsrf )a (Mzij ) + (Fy slf + Fy srf )a (Fy slr + Fy srr )b (Fx slf + Fx srf Fx slr + Fx srr )c 2

(34)

where m is the sprung mass and the subscript ij denotes left front (lf), right front (rf), left rear (lr), and right rear (rr).

200

T. Shim and C. Ghike

The cardan angles , , needed in the aforementioned equations are obtained by performing the integration of the following equations, = y cos z sin = y sin z cos + cos cos = x + y sin tan + z cos tan (35) (36) (37)

The model developed is a highly complex, non-linear full vehicle model with one-dimensional suspension. The next section shows the validation of this model with CARSIM. In the subsequent sections, we will consider the effect of various modeling assumptions and simplications on the responses of the vehicle model and understand their applicability and validity.

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

3. Vehicle model validation The 14 DOF full vehicle model is validated with CARSIM [33] and ADAMS/Car [34] for a J-turn maneuver at 50 kph. Figure 6 shows the steering wheel angle input for the maneuver and the comparative roll angles, lateral acceleration, and yaw rates. Figure 7 shows the comparative normal forces at the four tires. It can be seen that the developed model correlates very well with CARSIM until wheel lift-off. The slight difference in the responses prior to wheel lift-off and the deviation that occurs later may be due to the fact that the roll center in CARSIM is assumed to be xed with respect to the ground, whereas in the model developed here, the roll center is assumed to be at a xed distance from the sprung mass center of gravity. At extreme roll angles and certainly after wheel lift-off, this will have an impact on the responses. The difference

Figure 6.

Comparison of vehicle responses among the models during the J-turn maneuver at a speed of 50 kph.

Limitations of different vehicle models

201

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

Figure 7.

Comparison of tire normal forces among the models during the J-turn maneuver at a speed of 50 kph.

in the responses when compared with ADAMS/Car can be attributed to the variation in roll center height that occurs due to the suspension geometry in the multi-dynamics model.

4.

Effect of linearization of trigonometric terms (Model 1a)

A small angle assumption is widely employed in the development of vehicle models for various vehicle control systems. In this section, a small angle assumption is made for the cardan angles and , and its effect on the vehicle roll dynamics is investigated. No small angle assumption is made for the steering angle as this could cause differences at high steering inputs which are not truly representative of the limitations of the vehicle model as such. Several equations are simplied due to the linearization of the trigonometric terms with the small angle assumption. Equation (2), which represents the coordinate transformation of the ground forces, can be simplied to equation (38). Here it is assumed that sin sin = = 0. Fx gsrf 1 Fy gsrf = 0 Fzgsrf 0 1 Fx grf Fy grf 1 Fzgrf

(38)

Equations (7) and (8) for the longitudinal and lateral velocities at the tire ground contact patch are, respectively, modied to equations (39) and (40). ugrf = usrf y (lsrf + rrf ) + (wurf ) vgrf = vsrf + x (lsrf + rrf ) wurf (39) (40)

202

T. Shim and C. Ghike

The terms containing and in the above equations are very small and can be dropped without introducing any signicant errors to yield equations (41) and (42) subsequently. ugrf = usrf y (lsrf + rrf ) vgrf = vsrf + x (lsrf + rrf ) (41) (42)

Equation (11) for the vertical motion of the unsprung mass is simplied to equation (43) subsequently. mu w urf = xtrf ktrf mu g + Fx grf Fy grf Fdzrf (xsrf ksrf + x srf bsrf ) mu (vurf x uurf y ) Equation (14) for the tire deection xtrf is simplied to equation (44). (43)

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

x trf = wuirf = (wurf + vurf uurf ) The instantaneous tire radius is then simply given as rrf = ro xtrf

(44)

(45)

The forces Fx srf and Fy srf transmitted to the body by the suspension are now given by equations (46) and (47). Fx srf = Fx gsrf + mu g mu (u urf z vurf + y wurf ) urf x wurf + z uurf ) Fy srf = Fy gsrf mu g mu (v (46) (47)

The equations of motion for the three translational degrees of freedom of the sprung mass model are modied as m(u + y w z v) = m(v + z u x w) = m(w + x v y u) = The angles , , can be obtained as = y z = y + z = x + z (51) (52) (53) (Fx sij ) + mg (Fy sij ) mg (Fzsij + Fdzij ) mg (48) (49) (50)

Figures 8 and 9 show the effect of the small angle assumption on the vehicle responses. It can be seen that for the step steer input in gure 8, the small angle assumption makes no difference. In the ramp steer responses in gure 9, the small angle assumption seems to affect the responses only when the roll angle exceeds 8 10 . In general, the small angle assumption will not affect the vehicle responses while the concerned angles, and , remain less than 8 10 . The pitch angle is unlikely to reach such values except in very hard braking scenarios and, in most cases, the vehicle would already have seen wheel lift-off by the time the roll angle reaches 8 10 . Thus, we can conclude that the small angle assumption is signicant only for studies of vehicle behavior after wheel lift-off has occurred.

Limitations of different vehicle models

203

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

Figure 8. Comparative responses step steer effect of small angle assumption and ignoring lateral and longitudinal inertia forces of unsprung masses.

Figure 9. Comparative responses ramp steer effect of small angle assumption and ignoring lateral and longitudinal inertia forces of unsprung masses.

204

T. Shim and C. Ghike

5.

Effect of the unsprung masses: ignoring the lateral and longitudinal inertia (Model 1b)

The terms u urf and v urf in equations (46) and (47), which are required to compute the inertia forces of the unsprung masses, can be determined either by differentiation of the velocities uurf and vurf , respectively, or by the equations given subsequently srf y u urf = u l y lsrf srf x + v urf = v + a z + l x lsrf (54) (55)

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

Differentiation is not desirable in simulations as it slows down the computation time, and it is also not suitable for the development of control logic. Using the equations above will also add = BX + C . Therefore, for the sake of several off-diagonal terms to the A-matrix in the form AX simplicity, in addition to the small angle assumption, the terms mu (u urf z vurf + y wurf ) and urf x wurf + z uurf ) representing the inertia forces of the unsprung mass are dropped mu (v from equations (46) and (47), respectively, to yield the simplied equations subsequently. Fx srf = Fx gsrf + mu g Fy srf = Fy gsrf mu g (56) (57)

Also, equation (43) for the unsprung mass vertical velocity can be simplied to equation (58) by dropping the term mu (vurf x uurf y ) without any signicant effect prior to wheel lift-off. mu w urf = xtrf ktrf mu g + Fx grf Fy grf Fdzrf (xsrf ksrf + x srf bsrf ) (58)

It can be seen from the step steer responses in gure 8 that ignoring the unsprung mass inertia forces increases the net transient roll moment acting on the sprung mass and thus slightly increases the peak transient roll angles and lateral load transfers. This increase in the roll moment would also result in signicantly quicker rollover of the vehicle after two wheel liftoff if the changes in equations (56) and (57) are applied without modifying equation (58) as shown. However, the modication in equation (58) opposes this tendency for quicker rollover and, as seen in gure 9, the vehicle actually rolls over at almost the same time. It should be noted here that the unsprung mass is NOT lumped with the sprung mass in equations (48) and (49). Doing so causes deviation in the responses.

6.

Effect of further simplications to the 14 DOF full vehicle model

Even after the small angle assumption and ignoring the inertia forces of the unsprung masses, the equations for the 14 DOF model are fairly complex. In this section, certain additional simplications and their effect on the vehicle roll response are studied. 6.1 Simplifying the equations for the forces transmitted to the sprung mass (Model 1c-i)

The equations ((38), (56) and (57)) for the lateral and longitudinal forces transmitted to the sprung mass can be simplied by dropping the terms containing and as Fx srf Fy srf = Fx gsrf Fy gsrf = Fx grf Fy grf (59)

Limitations of different vehicle models

205

This simplication reduces the roll moment on the sprung mass, which reduces the predicted transient as well as steady-state roll angles as seen in the step steer responses in gure 10. There is also a consequent reduction in the lateral load transfer at the tires. For small roll angles, this effect is not signicant and may be deemed acceptable. However, as seen in the ramp steer responses in gure 11, at higher roll angles, there are considerable errors, which in this particular case make the difference between vehicle rollover and stability.

6.2 Simplifying the equation for the unsprung mass vertical velocity (Model 1c-ii) The terms containing and are dropped from the equation for the vertical motion of the unsprung mass to yield equation (60).

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

mu w urf = xtrf ktrf mu g Fdzrf (xsrf ksrf + x srf bsrf )

(60)

Dropping the term Fy grf causes an increase in the vertical acceleration of the unsprung mass and a consequent decrease in the tire compression. This reduces the predicted transient and steady-state tire normal forces, as is clearly seen in both the step steer and ramp steer responses in gures 10 and 11, respectively. The reduced tire normal forces result in lower lateral force generated at the tires and consequently reduced roll moment acting on the sprung mass. However, as the normal forces on the loaded tire are more affected by the simplication in equation (60), the effective roll stiffness of the model is also reduced. Thus, the change in roll angle is not as signicant as the change in the tire normal forces. However, it should be noted that this simplication will also affect the response of the vehicle to any normal force inputs at the suspension corners from an active suspension system.

Figure 10.

Comparative responses step steer effect of simplications indicated in Model 1c-i and Model 1c-ii.

206

T. Shim and C. Ghike

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

Figure 11.

Comparative responses ramp steer effect of simplications indicated in Model 1c-i and Model 1c-ii.

6.3 Simplifying the equation for the tire deection (Model 1c-iii) Equation (44) for the tire deection is simplied to equation (61) subsequently. x trf = wuirf = wurf (61)

Along with this simplication, it is necessary to change the equations for the angles of the sprung mass as = y = z = x (62) (63) (64)

The removal of the term vurf from equation (44) results in increased tire compression prediction. As can be seen in the step steer and ramp steer responses in gures 12 and 13, this results in signicantly greater tire normal forces at high roll angles. Moreover, the model can no longer predict wheel lift-off and rollover as is apparent from the ramp steer outputs. 6.4 Simplifying the equations for velocities at the tireground contact patch (Model 1c-iv) The longitudinal and lateral velocities at the tire are taken to be the same as those at the corresponding sprung mass corner. Thus, equations (41) and (42) are simplied as shown subsequently. ugrf = usrf vgrf = vsrf (65) (66)

Limitations of different vehicle models

207

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

Figure 12.

Comparative responses step steer effect of simplications indicated in Model 1c-iii and Model 1c-iv.

As can be seen from the step steer responses in gure 12, this simplication changes the roll frequency of the model entirely. However, there is no change in the steady-state roll angles or tire forces and the ramp steer responses in gure 13 are identical. The change in the roll frequency is very important when considering the response of the vehicle in non-steady state

Figure 13.

Comparative responses ramp steer effect of simplications indicated in Model 1c-iii and Model 1c-iv.

208

T. Shim and C. Ghike

maneuvers such as a double lane change test and also in the development of control logic for active steer or active normal force controls. However, equations (41) and (42) can be simplied by using the static c.g. height of the sprung mass h in place of the instantaneous lengths lsrf + rrf without introducing any errors either in the steady state or transient responses prior to wheel lift-off. Similarly, the initial strut length lsif can be used in place of the instantaneous strut length lsrf in equations (9) and (10) for the unsprung mass lateral and longitudinal velocities. The unsprung mass velocities can also be replaced by the corner velocities to further simplify the equations. However, this results in a slight difference in the roll frequency of the model. This error is very small and can generally be ignored.

7.

Some other modeling considerations

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

7.1 Effect of the roll center The inclusion of the roll center is very important in accurately predicting the roll behavior of the vehicle. As seen in the step steer and ramp steer responses in gures 14 and 15, the roll angles predicted are very different if no roll center is considered in the model. The steadystate tire normal forces are not substantially different but the transient peaks of the tire normal forces differ considerably. This can make the critical difference in wheel lift-off/rollover in the shhook maneuver, for example, and can also affect the predicted trajectory. The roll frequency does not change signicantly. Naturally, the difference in the responses will be greater if the roll centers are further away from the ground. Thus, it is recommended that the roll center height be always considered even in cases where roll angle is not calculated or important, such as in the lower order models (7 DOF/8 DOF) used in yaw control sytems.

Figure 14.

Comparative responses step steer effect of roll center and tire inclination angle.

Limitations of different vehicle models

209

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

Figure 15.

Comparative responses ramp steer effect of roll center and tire inclination angle.

In gure 14, even though the model without roll center rolls more and predicts one wheel lift-off slightly earlier than the model with roll center, it actually rolls over at a later stage. This discrepancy exists because the roll center considered in the 14 DOF model is xed with respect to the sprung mass c.g. and not the ground. 7.2 Effect of tire inclination angle In the vehicle model developed, the tire was considered to remain normal to the sprung mass. In the case of a McPherson-type suspension, for example, this may be a reasonable assumption. However, for a double wishbone or multilink suspension, the camber angle changes are very small when compared with the roll angle of the body, and it may be more accurate to consider the tire remaining normal to the ground, rather than normal to the sprung mass, until wheel lift-off. Once wheel lift-off occurs, the tire may be assumed to remain normal to the line joining the right and left tire bottoms. To incorporate the above assumption certain equations need to be modied. For example, the longitudinal and lateral velocities at the tire contact patch are now given as ugrf = cos (usrf (y lsrf + (y cos tf z sin tf )rrf )) tf rrf sin tf ) + sin (wurf cos + (x lsrf + vsrf ) sin + tf rrf cos t wurf sin vgrf = cos (vsrf + x lsrf ) + and the instantaneous tire radius is calculated from the tire spring deection as rrf = ro xtrf cos cos tf (69) (67) (68)

210

T. Shim and C. Ghike

The equations for the moments acting on the sprung mass and for the lateral load transfer forces are also modied as shown subsequently Mx rf = Fy gsrf rrf + Fy srf lsrf Fdzlf = Fdzrf = Fy gsrf rrf + Fy srf lsrf Fy srf hrcf cf /2 (70) (71) (72) (73)

My rf = (Fx sgrf rrf cos( tf ) + Fx srf lsrf ) Mzrf = Fx gsrf rrf sin( tf ) where

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

Fy gsrf = (cos tf Fy grf + sin tf (Fzgrf cos + Fx grf sin ))

(74)

The angle tf in these equations is the inclination of the tire with respect to the ground in the roll plane and can be approximated as tf = tan1 min(xtrf , 0) min(xtlf , 0) cf cos (75)

tf can be approximated as and the angular velocity if tf = 0, tf = 0 else tf = x (76)

As seen in the step steer response in gure 14, this change in the tire inclination angle changes the roll frequency of the model. This effect is similar to that seen in the Model 1c-iii. Also, even though the steady-state roll angle does not change, the tire normal forces at this roll angle are different. There is less lateral load transfer for the same roll angle when the tire is considered to remain normal to the ground. This effect can also be seen in the ramp steer responses in gure 15. Owing to this, the wheel lift-off and rollover occur later if the tire is considered to remain normal to the ground. In these simulations, the camber angle input to the tire model (needed for calculating the tire lateral forces) was kept equal to zero in both cases. Thus, the comparison indicates how the camber angle variation can affect the vehicle rollover limit in addition to changing the lateral forces at the tires. If it is found suitable or desirable to consider the tire to remain normal to the ground rather than normal to sprung mass, or if the given camber angle variation is desired to be incorporated into the model, the aforementioned equations can be used. The equations can also be simplied along similar lines as the original vehicle model. 7.3 Effect of link load transfer equations

In the vehicle model developed, the link load transfer forces are considered symmetrical (equation (26)). It can be argued correctly that these forces will actually not be symmetrical. The tire developing the greater lateral force will have greater jacking force component. However, it is seen in our simulations that it is not the distribution of the left and right jacking forces, but the sum of their magnitudes, |Fdzrf Fdzlf |, which has a signicant impact on the vehicle responses. If the net moment transferred by the jacking forces, (Fdzrf Fdzlf )cf , is the same, then the actual distribution of the left and right forces does not have a signicant impact. Therefore, for sake of simplicity, the link load transfer forces can be considered symmetrical.

Limitations of different vehicle models

211

8. The 8 DOF full vehicle model An 8 DOF vehicle model is often used as a simplied lower order model for studying vehicle handling in scenarios which do not involve signicant longitudinal accelerations. In this section, a formulation for the 8 DOF model, adapted from various references, that can match the 14 DOF model reasonably accurately is presented and its limitations are studied. Figure 16 shows the schematic of the 8 DOF full vehicle model. The model has four degrees of freedom for the chassis velocities and one degree of freedom at each of the four wheels representing the wheel spin dynamics. The chassis velocities include the longitudinal velocity, u, the lateral velocity, v , the roll angular velocity, x , and the yaw angular velocity, z . The pitch and heave motions are not modeled and the front and rear suspensions are represented simply by their respective equivalent roll stiffness (k f /k r ) and roll damping coefcients (b f /b r ). The equations of motion for the chassis velocities are obtained as given subsequently: mt (u z v) = mt (v + z u) =
2 Fx gij + (muf a mur b)z 2hrc mz x

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

(77) (78)

Fy gij + (mur b muf a) z + hrc m x (Fx grf Fx glf )cf 2

Jz z + Jxz x = (Fy glf + Fy grf )a (Fy glr + Fy grr )b + +

(Fx grr Fx glr )cr + (mur b muf a)(v + z u) 2

(79)

Figure 16.

Schematic of 8 DOF full vehicle model.

212

T. Shim and C. Ghike

+ hrc m(v (Jx + mh2 x + Jxz z = mghrc (k f + k r ) (b f + b r ) + z u) (80) rc ) where hrc = hrcf b + hrcr a a+b (81)

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

In these equations, the forces Fx gij and Fy gij are the longitudinal and lateral forces at the four tire contact patches and the subscript ij denotes lf, rf, lr, and rr. As before, hrcf and hrcr are the vertical distances of the front and rear roll centers below the sprung mass c.g., and thus hrc is the vertical distance from the sprung mass c.g. to the vehicle roll center. It should be noted that as equation (80) for the roll degree of freedom is written by considering moments acting about the vehicle roll center rather than the sprung mass c.g., the roll inertia of the sprung mass about the vehicle roll center (Jx + mh2 rc ) is considered in equation (80). The equations for the wheel dynamics and the longitudinal and lateral tire forces are the same as those used in the 14 DOF model (equations (3)(6)). The longitudinal and lateral velocities at, for example, the right front tire contact patch required in these equations are given as ugrf = u + vgrf z cf 2 = u + z a (82) (83)

The normal forces at the four tires are determined as Fzglf = muf g mgb + 2(a + b) 2 Fzgrf = mb(hcg hrcf ) muf huf (k f + b f ) (v + z u) + cf cf (a + b) cf (84)

(mhcg + muf huf + mur hur )(u z v) 2(a + b)

mgb muf g + + 2(a + b) 2

mb(hcg hrcf ) muf huf (k f + b f ) (v + z u) + + cf cf (a + b) cf (85)

(mhcg + muf huf + mur hur )(u z v) 2(a + b)

Fzglr =

mga mur g + 2(a + b) 2 +

ma(hcg hrcr ) (k r + b r ) mur hur + (v + z u) cr cr (a + b) cr (86)

(mhcg + muf huf + mur hur )(u z v) 2(a + b)

Fzgrr =

mur g mga + + 2(a + b) 2 +

ma(hcg hrcr ) mur hur (k r + b r ) + (v + z u) + cr cr (a + b) cr (87)

(mhcg + muf huf + mur hur )(u z v) 2(a + b)

These equations are fairly simple and linearized. It is possible to include several additional terms in the equations for the chassis velocities as well as the tire forces. However, in our experience, these terms have very little effect on the vehicle responses and can be ignored. As seen in the step steer responses in gure 17, the 8 DOF model as described is able to match

Limitations of different vehicle models

213

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

Figure 17.

Comparative responses step steer 14 DOF model and 8 DOF model.

the responses of the 14 DOF fairly accurately. However, there is no provision in the 8 DOF model to simulate wheel lift-off as was possible in the 14 DOF model through equation (16). Thus, as can be seen from the ramp steer responses in gure 18, the 8 DOF model cannot simulate vehicle behavior beyond wheel lift-off. Nevertheless, the model is valid for applications

Figure 18.

Comparative responses ramp steer 14 DOF model and 8 DOF model.

214

T. Shim and C. Ghike

which do not involve wheel lift-off such as active steering and active throttle/brake systems for yaw control.

9.

Limitations of the 14 DOF model

There can be differences between the responses of the 14 DOF vehicle model developed when compared with an actual vehicle or a multi-body dynamics based vehicle model. These errors are due to several effects that exist in an actual vehicle suspension but are not represented in the vehicle model. Compliance elements are not considered in the vehicle model. The tire and suspension struts are considered to remain normal to the sprung mass. In an actual vehicle, the camber angle variation with wheel travel will affect the tire force generation and roll behavior. Similarly, the orientation of the struts with respect to the sprung mass also changes in a real suspension. This affects the equivalent spring/roll stiffness. The lateral and vertical movements of the roll center that occur in an actual vehicle when the suspension deects are also not considered in the 14 DOF model. These affect the net roll moment acting on the sprung mass as well as the load transfer at the tires. This effect can be included in the 14 DOF vehicle model developed by including the roll center movement as an additional input into the model or by expressing the roll center location as a function of the wheel travel. However, the roll center is also the instantaneous center of rotation of the sprung mass with respect to the ground. If in the actual suspension there is substantial lateral movement of the roll center, then it will affect the location, velocity, and acceleration (lateral and more importantly vertical) of the sprung mass c.g. For example, if the roll center moves in the positive v -direction during a positive roll, then it will result in a drop in the vertical position of the sprung mass c.g. which is not modeled in the 14 DOF model mentioned earlier. This may signicantly affect the vehicle roll response and the tire normal forces in extreme maneuvers. A vehicle model that considers the roll center not just as a point of application of forces but also as a kinematic link would be more accurate in such cases. In spite of these limitations, the 14 DOF model developed is a fairly good representation of the vehicle roll and handling behavior in most cases. It can be used in anticipating vehicle rollover and for the development of integrated chassis controls that include active suspension/active roll control systems and active steering and active throttle/brake control.

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

10.

Conclusions

A 14 DOF full vehicle model with one-dimensional suspension and xed roll center, which can simulate vehicle rollover, was developed and validated with CARSIM. The effect of simplications of certain key equations on the vehicle responses was studied to understand the limitations and extent of validity. Similarly, the effects of including the roll center, the tire inclination angle, and the symmetry/asymmetry of jacking forces were also analyzed. A formulation for a lower order 8 DOF model is presented, which is able to match the 14 DOF model fairly accurately until wheel lift-off. Finally, the possible limitations of the 14 DOF model when compared with an actual suspension are discussed. It is expected that the results and analyses presented here prove useful in answering questions regarding the validity and applicability of various vehicle models and the effect of certain key modeling assumptions. The limitations of the 14 DOF model with one-dimensional suspension developed also indicate a need for a vehicle model that can incorporate the effects of a

Limitations of different vehicle models

215

moving roll center which acts both as a point of application of forces as well as a kinematic constraint. Vehicle parameters Sprung mass m = 1440 kg Sprung mass roll inertia Jx = 900 kg m2 Sprung mass yaw inertia Jy = 2000 kg m2 Sprung mass pitch inertia Jz = 2000 kg m2 Distance of sprung mass c.g. from front axle a = 1.016 m Distance of sprung mass c.g. from rear axle b = 1.524 m Sprung mass c.g. height h = 0.75 m Front/rear track width cf = cr = 1.5 m Front suspension stiffness ksf = 35,000 N/m Front suspension damping coefcient bsf = 2500 Ns/m Rear suspension stiffness ksr = 30,000 N/m Rear suspension damping coefcient bsr = 2000 Ns/m Front/rear unsprung mass muf = mur = 80 kg Front/rear tire stiffness ktf = ktr = 200,000 N/m Nominal tire radius ro = 0.285 m Tire/wheel roll inertia Jw = 1 kg m2 Front roll center distance below sprung mass c.g. hrcf = 0.65 m Rear roll center distance below sprung mass c.g. hrcr = 0.6 m

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

References
[1] Ponticel, P., 2003, Dynamic rollover testing on the way. Automotive Engineering International, 111, 2628. [2] Forkenbrock, G.J., Garrott, W.R., Heitz, M. and OHara, B.C., 2003, An experimental examination of J-turn and shhook maneuvers that may induce on-road, untripped, light vehicle rollover, SAE, 2003-01-1008. [3] Eger, R. and Kiencke, U., 2003, Modeling of rollover sequences. Control Engineering Practice, 11, 209216. [4] Dahlberg, E., 2000, A method determining the dynamic rollover threshold of commercial vehicles. SAE, 2000-01-3492. [5] Frimberger, J., 2000, Algorithm concepts for rollover detection to activate advanced restraint systems. SAE, 2000-01-0057. [6] Rakheja, S. and Piche, A., 1990, Development of directional stability criteria for an early warning device. SAE, 902265. [7] Nalecz, A. Lu., A. and dEntremont, K., 1993, An investigation intod dynamic measures of vehicle rollover propensity. SAE, 930831. [8] Das, N., Suresh, B. and Wambold, J., 1993, Estimation of dynamic rollover threshold of commercial vehicles using low speed experimental data. SAE, 932949. [9] Chen, B. and Peng, H., 1999, Rollover warning for articulated vehicles based on a time-to-rollover metric. Proceedings of ASME International Mechanical Engineering Congress and Exposition, Nashville, TN, November 1999. [10] Hyun, D. and Langari, R., 2003, Modeling to predict rollover threat of tractor-semitrailers. Vehicle System Dynamics, 39(6), 401414. [11] Furleigh, D.D., Vanderploeg, M.J. and Oh, C.Y., 1988, Multiple steered axles for reducing the rollover risks of heavy articulated trucks. SAE, 881866. [12] Watanabe, Y. and Sharp, R.S., 1999, Mechanical and control design of a variable geometry active suspension system. Vehicle System Dynamics, 32, 217235. [13] Cecj, I., 2000, Anti-roll and active roll suspension. Vehicle System Dynamics, 33, 91106. [14] Konik, D., Bartz, R., Barnthol, F., Bruns, H. and Wimmer, D., 2000, Dynamic drive-the new active roll stabilization system from the BMW group-system description and functional improvements. Proceedings of AVEC 2000, 5th International Symposium on Advanced Vehicle Control, August 2000. [15] Sampson, D.J.M. and Cebon, D., 2003, Active roll control of single unit heavy road vehicles. Vehicle System Dynamics, 40(4), 229270. [16] Wielenga, T., 1999, A method for reducing on-road rollovers anti-rollover braking. SAE, 1999-01-0123. [17] Palkovics, L., Semsey, A. and Gerum, E., 1999, Rollover prevention system for commercial vehicles-additional sensorless function of the electronic brake system. Vehicle System Dynamics, 32, 285297.

216

T. Shim and C. Ghike

Downloaded by [Cinvestav del IPN] at 10:29 17 April 2013

[18] Chen, B. and Peng, H., 2001, Differential-braking-based rollover prevention for sport utility vehicles with human-in-the loop evaluations. Vehicle System Dynamics, 36(4,5), 359389. [19] Dixon, J.C., 1996, Tires, Suspension and Handling (2nd edn) (Warrandale, PA: Society of Automotive Engineers Inc.). [20] Hac, A., 2002, Rollover stability index including effects of suspension design. SAE, 2002-01-0965. [21] Gillespie, T., 1992, Fundamentals of Vehicle Dynamics (Warrandale, PA: Society of Automotive Engineers Inc.). [22] The National Highway Trafc Safety Administrations Rating System for Rollover Resistance: An Assessment, Transportation Research Board, 2002, Special Report 265. [23] Bernard, J., Shannan, J. and Vanderploeg, M., 1989, Vehicle rollover on smooth surfaces. SAE, 891991. [24] Hac, A., Brown, T. and Martens, J., 2004, Detection of vehicle rollover. SAE, 2004-01-1757. [25] Song, J., 2005, Performance evaluation of a hybrid electric brake system with a sliding mode controller. Mechatronics, 15, 339358. [26] He, J., Crolla, D., Levesley, M. and Manning, W., 2004, Integrated active steering and variable torque distribution control for improving vehicle handling and stability. SAE, 2004-01-1071. [27] Cooper, N., Manning, W., Crolla, D. and Levesley M., 2005, Integration of active suspension and active driveline to ensure stability while improving vehicle dynamics. SAE, 2005-01-0414. [28] Shim, T. and Toomey, D., 2004, Investigation of active steering/wheel torque control at the rollover limit maneuver. SAE, 2004-01-2097. [29] Shim, T. and Margolis, D., 2005, Dynamic normal force control for vehicle stability enhancement. International Journal of Vehicle Autonomous System, 3, 114. [30] Drozdz, W. and Pacejka H., 1991, Development and validation of a bond graph handling model of an automobile. Journal of the Franklin Institute, 38, 941957. [31] Day, T., Roberts, S. and York, A., 2001, SIMON: a new vehicle simulation model for vehicle design and safety research. SAE, 2001-01-0503. [32] Pacejka, H., 2002, Tire and Vehicle Dynamics (Warrandale, PA: Society of Automotive Engineers Inc.). [33] Mechanical Simulation Corporation. Available online at: http://www.carsim.com/. [34] MSC software. Available online at: http://www.mscsoftware.com/products/adams.cfm.

S-ar putea să vă placă și