Process automation laboratory - Root locus, PI control

Kjartan Halvorsen

March 26, 2020

Repetition: The RC-circuit - a first-order system
1
R UC (s) = U(s)
s RC +1
|{z}
τ

s-plane
u(t) C uc (t)
×
− τ1

t
uC (t) = 10(1 − e τ ), for u(t) step of size 10
10 u(t)

5
V

uC (t)
0
0 1 2 3 4 5 6
t
A concept to keep in mind

In a system with an integrator steady-state can only exist if the signal to the integrator
is zero
R

u(t) C uc (t)

i(t) uC (t) 1
R
C uC (t) = uC (0) + C i(s)ds
Feedback control

v (t)

controller plant
yref (t) e(t) u(t) y (t)
F (s) G (s)
P P

For the closed-loop system we get

Go (s) G (s)
Y (s) = Yref (s) + V (s),
1 + Go (s) 1 + Go (s)

where Go (s) = G (s)F (s) is called the loop gain.

Feedback control
v (t)
controller plant
yref (t) e(t) u(t) y (t)
F (s) G (s)
P P

Go (s) G (s)
Y (s) = Yref (s) + V (s), Go (s) = G (s)F (s).
1 + Go (s) 1 + Go (s)

Let G (s) = 1
s and F (s) = K .
1. Will there be a steady-state control error (limt→∞ e(t) 6= 0) if yref (t) = 0 and
v (t) is a unit step? Why? Answer on Socrative
2. What is the characteristic equation for the closed-loop system?
3. Sketch the location of the poles in the imaginary plane as the gain K varies from 0
to ∞. Group exercise in breakout room
How to get rid of the steady-state error

Use a proportional-integral controller (PI controller)

Ki
F (s) = Kp + .
s
This gives closed-loop system
Kp
v (t)
yref (t) e(t) Ki
u(t) y (t)
G (s)
P P P
s
−
How to get rid of the steady-state error

Use a proportional-integral controller (PI controller)

Ki
F (s) = Kp + .
s
This gives closed-loop system
Kp
v (t)
yref (t) e(t) Ki
u(t) y (t)
G (s)
P P P
s
−

The only way that steady-state can exist is if the input to the integrator of the
controller is zero.
Root locus

Given loop gain

Q(s)
Go (s) = K
P(s)
how does the solutions to the characteristic equation

1 + Go (s) = 0 ⇔ P(s) + KQ(s) = 0

(i.e. the poles of the closed-loop system) depend on K ?

The root locus for PI-control of the integrator
Root locus question
Root locus definition

Let
(
P(s) = s n + a1 s n−1 + · · · + an = (s − p1 )(s − p2 ) · · · (s − pn )
, n≥m
Q(s) = s m + b1 s m−1 + · · · + bm = (s − q1 )(s − q2 ) · · · (s − qm )

The root locus shows how the roots to the equation

P(s) + K · Q(s) = 0, 0≤K <∞ (1)

depend on the parameter K . The root locus consists of the set of all points in the
complex plane that are roots to (1) for some non-negative value of K .
Characteristics of the root locus

The polynomial P(s) + KQ(s) = 0 above will always have n roots. Each gives a branch
in the root locus. Since the polynomials P(s) and Q(s) have real-valued coefficients, all
roots are either real or complex-conjugated pairs. This means that the root locus is
symmetric about the real axis. Other characteristics
I Start points - marked by crosses
I End points - marked by circles
I Asymptotes
I Pieces of the real axis
Start- and end points

Start points These are the n roots of P(s) + KQ(s) for K = 0, i.e. the roots of P(s).
These are the open-loop poles, and are marked with crosses ’×’
End points These are the m (finite) roots of P(s) + KQ(s) when K → ∞, and are
hence the roots of Q(s). The end points are marked with circles ’◦’
The real axis

Those parts of the real axis that have an odd number of real-valued start- or end points
to the right (including multiplicity) belong to the root locus.
Asymptotes

With n starting points and m end points, then m of the branches will go to end points.
The rest will go out towards infinity along n − m asymptotes. The asymptotes go out
symmetrically from a point on the real axis.
Asymptotes, directions

The directions of the asymptotes are given by the expression

(2k + 1)π
θk = arg s = , k ∈Z
n−m
Example: 6 start points and 3 end points gives n − m = 6 − 3 = 3 and the directions

k =0

π
3,

θ = π, k =1 .

 π
− 3 , k = −1
Asymptotes, intersection with the real axis

Pn
− m
P
i=0 pi i=0 qi
i.p. = ,
n−m
where {pi } are the starting points (open-loop poles) and {qi } are the end points
(open-loop zeros).
PI-Control of the integrator
Write the controller
Ki 1  sTi + 1
F (s) = Kp + = K (1 + = K /Ti ,
s sTi s
and let Ti = 2. The characteristic equation can be written
K
s2 + (2s + 1) = 0
2

I Start points: n = 2, in s = 0
I End points: m = 1, s = − 12
I Asymptotes: m − n = 1, with directions θ = π
I The real line: The real-line left of the end-point is part of the root locus.
PI-Control of the integrator
Do on your own: First-order system

Instead of the plant being an integrator

1
F (s) =
s
consider a stable first order system
1
F (s) =
s +a
How does the root locus change?