You are on page 1of 2

Table 2: Manual settings for PID controller using on-line trial and error tuning.

Parameter

Value

Kc

8.25

I (s)

120

D (s)

Table 3: Performance comparison between the manual and automatic controller settings.
Controller settings
Manual (Trial-and-error)
Calculated (Ziegler-Nichols)

Quadratic Performance Index (QPI)


406.71
400.21

Table 4: Closed-loop specifications of the best controller.


Closed-loop specification
Gain Margin
Phase Margin

Value
0.91
-2.14

Discussion

When only considering the gain of the controller,

K c equal to 8 produced a level response

curve that resembled the most underdamped and oscillatory shape out of all the

K c values

tested.
An integral time of 60s and 120s produced responses that were both quite underdamped,

however, the graph for 60s displayed a response that appeared to decay better.
The final manual settings were chosen as those values in Table 2 because the level response
curve settled at the set-point value of 56% within 150s which was the best response time for
any setting. Also, the amount of overshoot was not more than 10% while the decay ratio from

peak to peak was approximately 0.20.


Upon comparison of the responses of the manual and calculated settings, the QPI for the
calculated (Ziegler-Nichols) settings was lower than the manual settings QPI. Hence, the
performance of the calculated settings is better, given that both values apply for a 10%
decrease step change. This is due to the fact that the ZN settings produce a level response that
contains less overshoot and tighter control than the manual settings.

From Table 4, it can be inferred that the best controller operates in a closed-loop system that
is marginally unstable. This is due to the fact that the phase margin is slightly less than 0 and
the gain margin is slightly less than 1.

You might also like