Professional Documents
Culture Documents
1 Intro
Example
Need for performance analysis and control
2 Modeling
Longitudinal dynamics
Road loads
Powertrain and Actuation
Tractive performance
3 Analysis
4 Summary
5 Examples
following slide.
Consider basic dynamics problem from Meriam and Kraige [4], where it is
assumed that the rear driven wheels are slipping and maximum
acceleration is to be found.
The value of this example is in showing what can be found given minimal
information in a given scenario: total mass, CG location, rough value of µ,
wheel size and inertia properties.
1 Longitudinal dynamics
2 Characterizing typical road loads
3 Powertrain and actuation
4 Modeling tractive effort
If slip is negligible, the vehicle mass and total rotational inertia of rotating
parts can be lumped. Consider this case.
Dynamics for 2D motion (xyz),
If Ftx exceeds ‘adhesion’ limit, wheel slips, and the traction force
takes a form such as, Ftx = µFz , where µ may be a function of ’slip’.
P
Fx = Ftf + Ftr − Fa − Frf − Frr − Fd − Fg
Ftf,r = tractive effort on front and rear
Fa = aerodynamic resistance force
Frf,rr = rolling resistance on front and rear
Fd = drawbar load
Fg = grade resistance = W sin θs
For small unmanned ground vehicles (mobile robots), it is usually not
necessary to consider aerodynamic or drawbar loads. More information for
full-scale vehicles can be found in [2, 6].
The forward velocity, vx , can only be determined once all the forces are
specified.
Rolling Resistance
Rolling resistance forces are
primarily caused by hysteresis due Hysteresis vs. adhesion (adapted from
to deflection of running gear [5])
materials while rolling.
In some applications, RR may also
account for some sliding, air
circulation, and fan effect of
l high-speed rolling tiresa
The rolling resistance coefficient is
defined as the ratio of the rolling
resistance force (e.g., applied at
center of rolling body) to the
normal load, fr = Fr /N .
a
For example, at 80-95 mph, 90-95% hysteresis, 2-10%
friction, 1.5-3.5% air resistance.
ME 379M/397 Cyber Vehicle Systems (Longoria)
Outline Intro Modeling Analysis Summary References Examples
1 GR2
mef f = mv + 2
I w + 2
Id
rw rw
with mv total translational mass, Iw wheel inertia, and Id inertia of
driveline parts.
Driven wheels/tires under load experience slip when they are driven. Slip
can be thought of as a deficit in distance traveled compared to a pure
rolling case.
Slip velocity for a wheel is defined as vs = rw ωw − vw , where ωw is the
wheel rotational velocity and vw is the translational velocity.
Slip is then defined,
vs
s=
max(rw ωw , vw )
taking a positive value in driving conditions and a negative value in
braking (referred to then as skid).
A rough estimate for passenger tires at steady high speed, for example, is about 3% [6].
The article below by Evans [1] suggests one way to make a rough estimate.
Some basic calculations for rolling resistance based on [1] for the small
rubber wheel on the Tamiya mouse vehicle are shown below.
Now we can build a general model for wheel/tire dynamics under both
driving torque and traction force effects.
P
ḣw = Iw ω̇w =P Tw = +Td ∓ PTt − Tb − Tf (ωw )
ṗx = mv̇x = Fx = ±Ft − Road loads
Td = drive torque, Tt = rw Ftx = traction torque, Tb = brake torque, Tf = friction torque
The wheel model suggests a simple performance model for the longitudinal
dynamics of vehicle that includes drivetrain dynamics and wheel slip. The
equations are,
Performance Analysis
Given the model basis described, there are various types of analysis that
can be performed. Some typical analysis problems include:
1 Given vehicle characteristics, loads, and tractive effort, predict
steady-state speed
2 Evaluate ability to traverse grade, overcome drawbar, etc.
3 Gear train evaluation/selection
4 Evaluate power and energy requirements, actuator selection and
control
5 Evaluate transient performance (acceleration, time to steady speed,
braking, etc.)
Summary
References
[1] I. Evans, “The rolling resistance of a wheel with a solid rubber tyre,”
British J. of Applied Physics, Vol. 5, pp. 187-188, 1954.
[2] T.D. Gillespie, Fundamentals of Vehicle Dynamics, SAE, Warrendale,
PA, 1992.
[3] D.C. Karnopp and D. Margolis, Engineering Applications of Dynamics,
Wiley, New York, 2008.
[4] J.L. Meriam and L.G. Kraige, Engineering Mechanics: Dynamics (4th
ed.), Wiley and Sons, Inc., NY, 1997.
[5] D.F. Moore, Friction and Lubrication of Elastomers, Pergamon Press,
New York 1972.
[6] J.Y. Wong, Theory of Ground Vehicles, John Wiley and Sons, Inc.,
New York, 1993 (2nd) or 2001 (3rd) edition.
Example problems
where rolling resistance forces are applied through the wheel centers at
height hr = h − rw , h being the CG height and rw the effective rolling
radius of the wheels. The first two equations are solved for the weight
distributed to the front and rear axles, Wf and Wr , respectively.
Solution of the three unknowns yields the two normal loads on the front
and rear axels and an ODE for the vehicle forward velocity, vx , in terms of
‘known’ quantities,
− Lh Ft + (l2 +h r fr )
Wf L W cos θ
Wr = h (l1 −hr fr )
L Ft + L W cos θ
v̇x 1
m [Ft − W sin θ − Fa − fr W cos θ]
The traction force will depend on the condition of the front and rear tires.
Note that Ft = Ftf + Ftr . It must be determined whether there is rolling
or slipping.
and,
l2 + hfr (v)
Ftf max = µW cos θ ,
L + hµ
where fr (v) is the rolling resistance force, generally a function of vehicle
velocity, v.
Initial assumptions/approach
This model can be used for the case where mass is concentrated on axle or
distributed, but in the latter the weight on the wheels must change.
Performance analysis
Simulations for the vehicle running on flat wood floor (θ = 0) with the two
different motors are shown below. The 100% skid µ was 0.19. Each graph
compares the vehicle forward velocity for no slip and slip case. Also
plotted is the wheel velocity for the slip case, r · ω.
Judging from the steady-state velocity, it is likely that the mouse uses the
18100 motor since a test showed that the steady velocity was ≈ 22 cm/sec.
Changing the slope does not affect the no-slip case, which predicts the
vehicle climbs reliably. However, when considering slip and including the
effect of mass distribution, the model shows a grade of 27 degrees can be
climbed (22 deg was roughly measured in a test).
You can also show that assuming no mass distribution will over-predict the
slope you can climb.