22-Mec-A3 System Analysis and Control · December 2013
Question 5 of 6: Root locus of an unstable plant and the gain range for stability
Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)
Notes on this paper
Paper format. National Exams, December 2013 — 07-Mec-A3
System Analysis and Control. Three hours, closed book; a Casio or
Sharp approved calculator and semi-logarithmic graph paper are the only permitted aids.
Six questions are printed; the rubric states that any four questions constitute a
complete paper and that all questions are of equal value. A two-page table of Laplace
transform pairs is appended to the examination. Because this set is a study resource,
all six questions are solved in full below.
Reference texts.
K. Ogata, Modern Control Engineering, 5th ed. — Ch. 5 (transient response),
Ch. 5.8 (steady-state errors), Ch. 6 (root-locus analysis and design), Ch. 7 (frequency-response
methods, Bode diagrams, stability margins).
N. S. Nise, Control Systems Engineering, 8th ed. — Ch. 4 (time response),
Ch. 6 (Routh–Hurwitz stability), Ch. 7 (steady-state errors, system type),
Ch. 8 (root locus), Ch. 10 (gain and phase margins).
R. C. Dorf and R. H. Bishop, Modern Control Systems, 13th ed. — Ch. 5, 7, 9.
G. F. Franklin, J. D. Powell and A. Emami-Naeini, Feedback Control of Dynamic
Systems, 8th ed. — Ch. 3 (Laplace methods), Ch. 5 (root locus), Ch. 6 (frequency
response).
Question 5: Root locus of an unstable plant and the gain range for stability
Given. Loop gain function
$G(s)=K(s^2+4s+8)/[s^2(s-1)]$ in a unity-feedback loop. The open-loop plant has a
double pole at the origin and a real pole in the right half plane at
$s=+1$, so it is open-loop unstable.
Find. The root-locus sketch as $K$ varies from 0 to
$\infty$, and the range of $K$ giving a stable closed loop.
Figure 5.1 — Root locus of
$K(s^2+4s+8)/[s^2(s-1)]$. Crosses mark the open-loop poles ($0$ twice
and $+1$), circles the zeros at $-2\pm j2$. Two branches break away
from the real axis at $s=+0.629$, sweep left, cross the imaginary axis at
$\pm j3.464$ when $K=3$, and terminate on the zeros; the third branch
runs left along the real axis to infinity.
Approach. Construct the locus from the standard rules (real-axis
segments, asymptote count and centroid, breakaway point from
$dK/ds=0$, imaginary-axis crossing from the Routh array), then read the stability
range directly off the crossing.
Catalogue poles, zeros and branches. The open-loop poles are
$s=0$ (double) and $s=+1$; the zeros are the roots of
$s^2+4s+8$, namely $s=-2\pm j2$. With $n=3$ poles and
$m=2$ zeros the locus has three branches: two terminate on the finite zeros and
$n-m=1$ runs off to infinity.
Locate the real-axis segments and asymptote. A real point lies on the
locus if the number of real poles and zeros to its right is odd. Counting the double pole
twice, every real $s \lt 1$ satisfies this, so the entire real axis to the left
of $+1$ belongs to the locus. The single asymptote lies at
$180^\circ$, with centroid
$$\sigma_a=\frac{\sum p_i-\sum z_i}{n-m}=\frac{(0+0+1)-(-2-2)}{1}=+5$$
Find the breakaway point. Solving $1+G=0$ for the gain,
$K=-s^2(s-1)/(s^2+4s+8)$, and setting $dK/ds=0$ yields
$$-s\left(s^3+8s^2+20s-16\right)=0$$
Apart from the trivial root at the double pole, the only real root of the cubic is
$s=+0.6292$, with $K=0.01345$ there. Two branches (one leaving the
origin to the right, one leaving $s=+1$ to the left) meet at this point and break
away into the complex plane at $\pm 90^\circ$.
Form the characteristic equation. From $1+G(s)=0$,
$$s^2(s-1)+K(s^2+4s+8)=0
\quad\Longrightarrow\quad
s^3+(K-1)s^2+4Ks+8K=0$$
Apply the Routh test. The array is
$$\begin{array}{c|cc}
s^3 & 1 & 4K\\
s^2 & K-1 & 8K\\
s^1 & \dfrac{4K(K-1)-8K}{K-1}=\dfrac{4K(K-3)}{K-1} & 0\\
s^0 & 8K &\end{array}$$
The $s^0$ row needs $K \gt 0$; the $s^2$ row needs
$K \gt 1$; and with $K \gt 1$ the $s^1$ row needs
$K \gt 3$. The binding condition is the last, so
$$\boxed{\;K \gt 3\;\text{ for closed-loop stability}\;}$$
There is no upper limit: unusually, this plant is stabilised only by large gain.
Locate the imaginary-axis crossing. At $K=3$ the
$s^1$ row vanishes and the auxiliary polynomial from the row above is
$$(K-1)s^2+8K=2s^2+24=0
\quad\Longrightarrow\quad
s=\pm j\sqrt{12}=\pm j3.464$$
Factoring the cubic at $K=3$ confirms it:
$s^3+2s^2+12s+24=(s^2+12)(s+2)$, so the third pole is at
$s=-2$. Below $K=3$ the pair sits in the right half plane; above it, in
the left.
Read the sketch. As $K$ rises from zero the two right-hand
branches leave $s=0^+$ and $s=1$, meet at $+0.629$,
break away vertically, curve left, cross the imaginary axis at
$\pm j3.464$ when $K=3$, and finally converge on the zeros
$-2\pm j2$ as $K\to\infty$. The third branch leaves the origin
travelling left and follows the $180^\circ$ asymptote to
$-\infty$. Because the whole locus is in the right half plane for small
$K$, the closed loop is unstable at low gain — the mirror image of the more
usual situation.
Check: the plant is open-loop unstable (a pole at
$s=+1$) and only high gain stabilises it, so any practical implementation must
guarantee that the loop is never operated at reduced gain — during start-up, saturation
or actuator de-rating the system would be genuinely unstable. The stability margin should
therefore be quoted as a lower gain margin, and the design gain set comfortably above
$K=3$ (a factor of two or more).
Quantity
Result
Open-loop poles
$s=0$ (double), $s=+1$
Open-loop zeros
$s=-2\pm j2$
Asymptote
1 asymptote at $180^\circ$, centroid $\sigma_a=+5$