MEEN 617 Handouts: Appendix B 2013 Luis San Andrs, Make McDermott B-1

Appendix B. LINEARIZATION

Linearization is a tool that allows us to use linear models to approximate the
performance of nonlinear systems. Linear models are desirable because
there are many powerful tools to analyze and design linear systems, whereas
there are very few general tools available for analysis and design of
nonlinear systems.

The linearization technique will be explained using an example. The
example is the speed control/analysis of a car (could be a boat, airplane,
rocket, cycle, etc.). The main forces acting on the car are aerodynamic drag,
F
d
, and the driving force F
e
, produced by the engine and drive train and
exerted at the rear wheels. The free body diagram (FBD) is:




The equation of motion is

= = v M F F F
eq d e X

(1)


where
F
e
= k , driving force from engine.
k = gain (a constant),
= throttle angle (controlled by driver),
F
d
= C
d
A v
2
, drag force
= air density,
C
d
= drag coefficient,
A = projected frontal area of car,
v = car speed,
M
eq
= equivalent mass of car.
F
e

F
d

MEEN 617 Handouts: Appendix B 2013 Luis San Andrs, Make McDermott B-2
Thus, the equation of motion (EOM) becomes


( )
2
1
2
-
d eq
k C A v M u v =
(2)

This is a 1
st
order nonlinear ODE for the velocity (v). We wish to linearize
the EOM about some constant operating point denoted by v
*
. The throttle
angle corresponding to a steady state speed v
*
is
*
, i.e.

2
1
2
* * 0
d
k C Av u =
(3)

at
*
and v
*
, i.e. the thrust force balances the drag force, so v = 0


Define ov and o as small changes in velocity and throttle angle,

v = v
*
+ ov ov = v - v
*
(4)


=
*
+ o o = -
*


For example, if v
*
=60 mph =constant, and at time t
1
, v(t
1
) =72 mph, then
ov(t
1
) =12 mph. This is simply a change in independent variable from v(t)
to ov(t).

Expand F
e
and F
d
in Taylor series about
*
and v
*
and drop nonlinear terms
since these are second order, that is small,


| F k F
F
e e
e
= = + u
c
cu
ou
c
c u
ou
u
u u *
* *
+
F
e
2
2
1
2
2




*
e
F k k u ou ~ +
(5)


F C Av
d d d
d 2
= F +
F
+ 0 (
=
1
2
2

c
cv
ov ov
v v
* *
)


( ) F Av Av
d

1
2
C + C
d d
~ o v
*
*
2
(6)


MEEN 617 Handouts: Appendix B 2013 Luis San Andrs, Make McDermott B-3
Substitution of the forces into the EOM (1) gives

= = v M F F F
eq d e X




| | ( )
| | k Av Av
d
dt
u ou ov v ov
* * *
* + k -
1
C + C = M +
d d eq
2
2

(


and rearranging

( )
( )
( ) k Av Av
d
dt
d
dt
u ou ov v ov * *
*
-
1
2
C + k - C = M + M
d d eq
*
eq
2

`
)



Since v
*
,
*
define the steady state operating point, the first term in curly
brackets { } is zero (see Eq. 3). The resulting linearized equation of motion
is:

( ) k C Av
d
ou ov ov - = M
eq
*

(7)

where k is the slope of the curve of F
e
vs at the operating point
*
, and
( C
d
A v
*
) is the slope of the curve of F
d
vs v at the operating point v
*
,
i.e.,
( )
* *
; *
e d
d d
v
F F
k k C Av
v
u

u
| | | | c c
= = =
| |
c c
\ . \ .
(8)

Hence, Eq. (7) becomes

eq d
M v k v k o o ou + =
(9)

Note that if using some other curve fit other than a Taylor series about a
steady state operating point, the term in curly brackets would not be zero (it
would be a constant). By using a Taylor series about a steady state operating
point, the method eliminates this constant forcing term in the linearized
ODE.

Use engineering judgement to select the appropriate
operating point for each problem (situation)