An Atlas of Engineering Dynamic Systems, Models, and Transfer Functions
An Atlas of Engineering Dynamic Systems, Models, and Transfer Functions
This document presents the models and/or transfer functions of some real-world control systems.
Transfer functions are the mathematical vehicle of classical controls. Transfer functions are
defined as the Laplace transform of the output variable divided by the Laplace transform of the input
variable, with zero initial conditions. Transfer functions represent the system dynamics, as described by
the simplified model – they yield the simulated system output given various inputs. Transfer functions
can be derived for the open-loop, closed-loop, and/or smaller system components. Block diagrams are
used for graphical representation, where the blocks have transfer functions representing the dynamics of
certain system components, while the arrows represent system variables.
The models and transfer functions summarized in this document only give the bottom-line results,
without derivations or much explanation. The reader is referred to the various references for more details.
2
Atlas of Engineering Dynamic Systems, Models,
and Transfer Functions
Table of Contents
We start with a table of like variables playing the same role in different engineering systems.
rotational mechanical angular velocity (t) angular displacement (t) torque (t) angular impulse
incompressible fluid volume flow rate q(t) volume V(t) pressure p(t) none
compressible fluid mass flow rate qm(t) mass m(t) pressure p(t) none
thermal heat flow rate q(t) heat energy Q(t) temperature T(t) none
4
2. Zeroth-Order System Examples
accelerometer, low-frequency X ( s) 2
( k m ) x (t ) xIN (t )
2
(Dorf & Bishop) X IN (s) k m
T (s)
motor torque KT
(t ) K T i (t ) I (s)
VB ( s )
back emf KB
v B (t ) K B M ( t ) M (s)
5
Zeroth-Order System Examples (continued)
Q(s) V ( s) 1
capacitor (q(t) ~ charge) q(t ) Cv (t ) C
V (s) Q( s) C
V ( s) I (s) 1
resistor R
v (t ) Ri (t ) I (s) V (s) R
(s) I ( s) 1
inductor ((t) ~ flux) L
( t ) Li ( t ) I (s) (s) L
V2 ( s ) R2
potentiometer v1 (t ) R2 v2 (t )( R1 R2 )
V1 ( s ) R1 R2
V (s)
tachometer Kt
v (t ) K t (t ) (s)
V2 ( s )
DC amplifier, zero time constant KA
v2 (t ) K Av1 (t ) V1 ( s )
Name Diagram Model G(s)
x(t)
X (s) 1 c
k
massless translational mechanical system c f(t) cx (t ) kx (t ) f (t ) F ( s ) cs k k
x(t)
V ( s) 1 m
mv(t ) cv(t ) f (t )
c
m f(t)
F ( s ) ms c c
springless translational mechanical system
( s ) 1 cR
cR
kR
c R (t ) k R (t ) (t ) T ( s ) cR s k R kR
inertialess rotational mechanical system (t) (t)
(s)
J
1 J
cR
J (t ) cR (t ) (t ) T ( s ) Js cR cR
springless rotational mechanical system (t) (t)
R
di (t ) I (s) 1 L
+ L Ri (t ) v (t )
v(t)
-
i(t) L
dt V ( s ) Ls R R
LR series electrical circuit
R
1 Q( s) C
Rq (t ) q (t ) v(t )
v(t)
+
-
i(t) C C V ( s ) RCs 1 RC
RC series electrical circuit 1 I ( s)
C
Ri ( t ) i ( t ) dt v ( t ) Cs
RC
V ( s ) RCs 1
V2 ( s ) KA
DC amplifier with time constant v2 (t ) v2 (t ) K Av1 (t )
V1 ( s ) s 1
7
First-Order System Examples (continued)
dq ( t ) d (t ) I (s) V (s)
i (t ) v (t ) s s
dt dt Q(s) ( s)
x (t ) v (t ) dt v (t ) a (t ) dt X (s) 1
V ( s) 1
integrator V (s) s A( s ) s
q ( t ) i ( t ) dt ( t ) v ( t ) dt
Q (s) 1 (s) 1
I ( s) s V ( s) s
I (s) V ( s) 1
Cs
dv (t ) 1 V (s) I ( s ) Cs
C
i (t ) C v (t ) i (t ) dt
capacitor dt
Q(s) 1 V (s)
Rs
1 dq (t ) V ( s ) Rs Q(s)
q (t ) v (t ) dt v (t ) R
resistor R dt
1 di (t ) I (s) 1 V (s)
i (t ) v(t ) dt v (t ) L Ls
L dt V ( s ) Ls I (s)
inductor
YSENS ( s ) k
H (s)
y SENS (t ) ySENS (t ) ky (t ) Y (s) s 1
generic sensor k gain time constant
8
4. Second-Order System Examples
x(t)
k mx(t ) cx (t ) kx (t ) f (t ) X (s) 1
2
translational mechanical system c m f(t) F ( s ) ms cs k
dv (t )
m cv (t ) k v (t )dt f (t )
dt V ( s) s
2
F ( s ) ms cs k
x(t)
springless translational c
mechanical system m f(t) mx(t ) cx (t ) f (t ) X (s) 1
F ( s ) s ( ms c )
x(t)
damperless translational
k
mechanical system m f(t) mx(t ) kx(t ) f (t ) X (s) 1
2
F ( s ) ms k
damperless translational k
mechanical system, vertical g my(t ) ky (t ) f (t ) Y (s) 1
m 2
F ( s ) ms k
y(t)
f(t)
x(t)
mass-only translational X (s) 1
mechanical system m f(t) mx(t ) f (t ) F ( s ) ms 2
9
Second-Order System Examples (continued)
(t) (t)
springless rotational J
mechanical system cR
J (t ) c R (t ) (t ) ( s ) 1
T ( s ) s ( Js cR )
(t) (t)
damperless rotational J
inertia-only rotational
J
dv (t ) v (t ) 1 V (s) RLs
+ C v (t ) dt i (t )
parallel RLC circuit i(t) iR(t) R iL(t) L iC (t) C v(t)
dt R L I ( s ) CRLs Ls R
2
-
1 1 (s)
C(t ) (t ) (t ) i (t )
RL
R L I ( s ) CRLs Ls R
2
R di (t ) 1 I (s) Cs
L Ri (t ) i (t )dt v(t )
series RLC circuit + dt C V ( s ) LCs RCs 1
2
v(t) i(t) L
-
1 Q(s) C
Lq(t ) Rq (t ) q(t ) v (t )
C C V ( s ) LCs RCs 1
2
X ( s) s 2
accelerometer (Dorf & Bishop) mx(t ) bx (t ) kx(t ) mxIN (t ) 2
X IN (s) s (b m)s (k m)
d 2 x (t ) A( s )
a (t ) s2
double differentiator dt 2 X ( s)
X ( s) 1
double integrator
x (t ) a ( t ) dt
A( s ) s 2
11
5. Real-World Transfer Functions
L R
cR J
v(t) i(t)
(t) (t)
(t)
where we have ignored the motor back emf voltage. Usually the time constant for the electrical system is
much smaller than the time constant for the rotational mechanical system, which means that the electrical
system current i(t) rises much faster than the mechanical displacement (t). Therefore, we can ignore the
circuit dynamics ( L 0 ), so the electrical circuit model simplifies to Ri (t ) v (t ) , which is simply Ohm’s
Law.
In a DC servomotor, the generated motor torque is proportional to the circuit current, a linear
proportional relationship that holds good for nearly the entire range of operation of the motor:
(t ) KT i(t )
KT is the motor torque constant, which is stamped on the motor housing, available from the motor
manufacturer, or determinable by experiment.
The rotational mechanical system dynamic model is derived from a free-body diagram of the
rotating motor shaft, using Euler’s rotational dynamics law M J(t ) :
J (t ) c (t ) (t )
R
Substituting the electrical models into the rotational mechanical system dynamic model yields:
K (s) KT R
J ( t ) c R ( t ) ( t ) K T i ( t ) T v ( t ) G ( s )
R V ( s ) s ( Js c R )
This is a linear, lumped-parameter, constant-coefficient, second-order ODE. The same model written for
angular velocity (t) output is a first-order model:
K (s) KT R
J ( t ) c R ( t ) T v ( t ) G ( s )
R V ( s ) Js cR
12
DC Servomotor1
( s ) K
G ( s )
V ( s ) ( Ls R )( Js cR ) K 2
KT K B K
( s ) K
G ( s )
V ( s ) s ( Ls R )( Js cR ) K 2
L J
If we set the armature circuit time constant to zero relative to the mechanical system time constant
R cR
(since the mechanical system dominates), the above transfer functions are simplified to first- and second-
order, respectively (rather than the original second- and third-order systems):
( s ) K
G ( s )
V ( s ) JRs ( RcR K 2 )
( s ) K
G ( s )
V ( s ) JRs ( RcR K 2 ) s
1
R.L. Williams II and D.A. Lawrence, 2007, Linear State-Space Control Systems, John Wiley & Sons, Inc.
13
suitable for ME 3012 Term Projects
Inverted pendulum1
In order to derive the overall SISO transfer function for the inverted pendulum, take the Laplace
Transform of both sides of both of the linearized ODEs above. Then use algebra to eliminate W(s) between
the two equations and arrive at G(s). This process yields the following Type 0, second-order, unstable
open-loop transfer function:
( s ) 1
G (s)
F ( s ) m1 Ls ( m1 m2 ) g
2
( s ) 1.15s 0.18
G (s) 2
( s ) s 0.74 s 0.92
2
www.engin.umich.edu/group/ctm
14
Automobile Cruise Control2
V ( s) 1
G (s) u(t) road force input v(t) output speed
U ( s ) ms b
( s ) K
G ( s) 2 q(t) hydraulic fluid flow (t ) (t ) roll velocity
Q(s) s 4s 9
(t) roll (bank) angle
Diabetes Control3
B(s) s2
G (s) i(t) insulin input b(t) output blood-sugar level
I ( s ) s ( s 1)
3
Dorf and Bishop, Modern Control Systems, 11th edition, Pearson Prentice Hall, 2007.
15
Elevator Control3
V (s) 1
G (s) 2 (t) torque input v (t ) y (t ) elevator velocity
( s ) s 2 s 11
y(t) elevator displacement
( s ) s6
G (s) (t) torque input (t) output angular velocity
( s ) ( s 2)( s 4)
16
Fluid Heating System3
(s) (2 s 1)( s 2)
G (s) E ( t ) input angle error (t) output elevator angle
E ( s ) (0.5 s 1)( s 2)
0.25 0.50 human pilot time constant
Hydraulic Actuator3
Ak x A2
Y (s) K K B b K g g ( x, P ) flow
G ( s) kP kP
X ( s ) s ( ms B ) A piston area
g g
kx kP
x x0 P P0
18
Laser Printer Positioning3
Y (s) 4( s 50)
G(s) 2 (t) input control torque y(t) printer head displacement
( s ) s 30 s 200
0 (s) K
G (s) M e0 (t ) input motor voltage 0 (t ) windup roll velocity
E0 ( s ) s 1
L
K M motor constant motor time constant
R
19
Racecar Speed Control3
V (s) 100
G(s) c(t) throttle input v(t) output racecar speed
C ( s ) ( s 2)( s 5)
( s ) 2
G (s) (t) torque input (t) output elbow angle
( s ) s ( s 4)
20
Robot Position Control3
Ship Stabilization3
(s) 9
G(s) 2
f (s) s 1.2s 9
Y (s) 1
G (s) (t) torque input y(t) output displacement
T ( s ) s (0.2 s 1)
( s ) 20
G (s) 2 (t) torque input (t ) (t ) station velocity
( s ) s 20 s 100
(t) space station angle
22
Space Telescope Pointing Control3
(s) 1
G (s) (t) torque wheel input (t) output telescope angle
T ( s ) s ( s 12)
Y (s) 0.25
G (s) (t) input motor torque y(t) output steel thickness
( s ) s ( s 1)
v0 2000 ft/min nominal steel speed
( s) KT / R
G( s) v(t) input voltage (t ) output turntable speed
V ( s) Js b
K T motor torque constant R circuit resistance
23
Vehicle Steering Control3
V (s) 1
G (s) (t) steering wheel angle v (t ) y (t ) centerline velocity
( s ) s ( s 12)
y(t) centerline displacement
Y (s) 1
G ( s) (t) thruster input y(t) vertical displacement
T ( s ) s ( s 1)
24
Weld Bead Depth Control3
Y (s) K
G (s) i(t) input current y(t) output weld bead depth
I ( s ) (0.01s 1)(1.5s 1)
V (s) 75( s 1)
G (s) (t) torque input v (t ) y (t ) robot velocity
( s ) ( s 5)( s 20)
y(t) robot displacement
25
Antenna Azimuth Control4
( s ) 20.83
G (s) v(t) input voltage (t ) (t ) azimuth velocity
V ( s ) ( s 100)( s 1.71)
(t) azimuth angle
( s ) 0.13( s 0.44)
G (s) 2 e(t) input elevator angle (t) output pitch angle
e ( s ) s 0.23s 0.02
4
N.S. Nise, Control Systems Engineering, 2nd edition, Cummings, 1995.
5
Golnaraghi and Kuo, Automatic Control Systems, 9th edition, Wiley, 2010
26
Missile Roll Control5
P(s) l l l
G (s) p (t) aileron angle input p(t) output roll rate
( s ) s l p a s 1
1
a aerodynamic time constant
lp
l l p steady-state gain
X (s) 10
G(s) 2 p(t) input air pressure x(t) output displacement
P ( s ) s 10 s 29
27
Robotic Swivel5
( s ) K
G(s) 2 v(t) input voltage (t) output swivel velocity
V ( s ) ( s 4 s 10)
To ( s ) 1
G ( s) ti(t) input temperature to(t) output temperature
Ti ( s ) RCs 1
RC thermal time constant
Pneumatic System5
Pb ( s ) 1
G (s) p1(t) input pressure pb(t) output pressure
P1 ( s ) RCs 1
q mass flow rate RC pneumatic time constant
28
Magnetically-Levitated Ball6
X (s) 1
G (s) 2
I ( s ) As B
i0 i0
A B
2g x0
X ( s) K
G ( s) f(t) road force input x(t) backup distance
F ( s ) s ( ms c )
K open-loop gain
m car mass
c effective viscous damping coefficient
6
H. Huang, H. Du & W. Li, 2015, "Stability enhancement of magnetic levitation ball system with two controlled
electromagnets," Power Engineering Conference (AUPEC), 2015 Australasian Universities: 1-6.
29
Third-Order and Fourth-Order Systems
( s ) s 500
G (s) (t) torque input (t) angle output
( s ) s ( s 0.03)( s 2 2.57 s 6667)
( s ) 25( s 0.03)
G(s) (t) torque input (t) output pitch angle
( s ) ( s 0.4)( s 2 0.36 s 0.16)
30
Robot Force Control3
F ( s) K ( s 2.5)
G ( s) 2 (t) torque input f(t) force output
( s ) ( s 2 s 2)( s 2 4 s 5)
31
Third-Order and Fourth-Order Systems
Ball-and-Beam System1
Jb
p(t ) m g sin (t ) m p(t ) (t )2 0
r 2 m
m p(t )2 J J b (t ) 2 m p(t ) p (t ) (t ) m g p(t ) cos (t ) (t )
p(t
) non-linear
Jb
(t) p(t ) m g (t )
r 2 m 0
(t) m L2
4 J J b (t ) m g p(t ) (t )
linearized
L is the constant half-length of the beam, m and r are the mass and
radius of the ball, respectively, Jb and J are the mass moment of inertia
of the ball and beam, respectively.
In order to derive the overall SISO transfer function for the ball-and-beam system, take the Laplace
Transform of both sides of both of the linearized ODEs above. Then use algebra to eliminate (s) between
the two equations and arrive at G1(s). This process yields the following Type 0, fourth-order, unstable
open-loop transfer function:
P(s) mg
G1 ( s )
T ( s ) J E1 J E 2 s 4 m 2 g 2
J m L2
where: J E1 2b m JE2 J Jb
r 4
This process could alternatively eliminate P(s) between the two equations and arrive at G2(s), the
following Type 0, fourth-order, unstable open-loop transfer function:
(s) J E1s2
G2 (s)
T (s) J E1J E 2 s4 m2 g 2
32
Missile Yaw Control3 (cannot use internal pre-filter for positive poles)
m mass 0.2 kg
k1 light sensor 1 V/m
k2 velocity feedback gain 0.1 Vs/m
r radius 0.15 m
L inductance 0 (ignore)
b rotational viscous damping 0.25 Nms/rad
R resistance 2
Km torque constant 2 Nm/A
J J motor J pulley inertia 0.01 kg-m2
34
6. Advanced Real-World Models
m1 m2 m3
c1 c2 c3 c4
q(t)
e
J
(t) m
y1(t) y2(t)
u1(t) u2(t)
y1 (t ) (c1 c2 ) y1 (t ) (k1 k2 ) y1 (t ) c2 y2 (t ) k2 y2 (t ) u1 (t )
m1
k1 k2
y2 (t ) c2 y2 (t ) k2 y2 (t ) c2 y1 (t ) k2 y1 (t ) u2 (t )
m2
m1 m2
c1 c2
35
Automobile Suspension System2
In this model, the Body Mass is m1, generally one-fourth of the vehicle mass (excluding tires). m2
is the Suspension Mass, the mass of one tire. As seen in the above transfer functions, this model is 4th-
order.
To make this model fit the second-order system controller designs we focus on in ME 3012:
Eliminate the tire mass (m2, the Suspension Mass) and include half of this into m1.
Then combine the two springs and two dampers in series to obtain one equivalent spring
stiffness and one equivalent damper coefficient.
36
Human Skeletal Muscle Model7
k2 (k k ) kk Fm (t ) k2
y (t )
y (t ) 1 2 y (t ) 1 2 y (t )
( FA (t ) Fm (t ) mg k1 L1R )
b m mb m mb
where:
7
Dr. Bob’s ME 4670 / 5670 Biomechanics NotesBook Supplement, derived by Elvedin Kljuno
37
Armature circuit / DC servomotor / gear box / robot joint1 – ME 3012 Term Example
L R n
cL
J L (t)
cM
+ + L (t)
v A (t) v B (t) JM
- - L (t)
i A (t) M (t) L (t)
M (t)
M (t)
KT
LJL (t ) ( Lc RJ )L (t ) ( Rc K T K B )L (t ) vA (t )
n
KT
LJ L (t ) ( Lc RJ ) L (t ) ( Rc K T K B )L (t ) vA (t )
n
c
where J J M J L2 and c cM L2 are the effective polar inertia and viscous damping coefficient
n n
reflected to the motor shaft.
Numerical Parameters
L (s) KT / n 5
G ( s ) 2
VA ( s ) LJs ( Lc RJ ) s ( Rc K T K B ) s 11s 1010
2
L ( s) KT / n 5
G ( s )
VA ( s ) s ( LJs ( Lc RJ ) s ( Rc K T K B )) s ( s 11s 1010)
2 2