Department of Engineering Technology GET 431/EET 441 Controls Lab Spring, 2018
Department of Engineering Technology GET 431/EET 441 Controls Lab Spring, 2018
Theory
The plant, as represented in figure 1, can be approximated as a
first order system. Heat (𝑄̇ ) is provided at a constant rate from
a resistive heating element. This energy is transferred to the air
stream, the heating element, supporting structure, and the
temperature sensor (identified collectively as Mass in figure 1).
𝑑𝑇𝑜 (𝑡) 1 1
+ 𝜏 𝑇𝑜 (𝑡) = 𝜏 𝑇∞ + 𝐴 ∙ 𝑄̇ (1)
𝑑𝑡
Since all the terms on the right hand side of equation (1) are constant, they will be combined
into new parameters (𝒯𝑖 /𝜏) where 𝒯𝑖 carries units of temperature. Equation (1) can then be
rewritten as,
𝑑𝑇𝑜 (𝑡) 1 𝒯𝑖
+ 𝜏 𝑇𝑜 (𝑡) = (2)
𝑑𝑡 𝜏
𝑇𝑜 (𝑠) 1 1
𝐺𝑝 = = 𝜏 ∙ (𝑠+1⁄𝜏) (3)
𝒯𝑖 (𝑠)
and the closed-loop control system is presented in figure 2 where the controller (Gc) can be of
any type. The control action to be
examined in today’s laboratory is
proportional control.
𝐺𝑐 = 𝐾𝑝 (4)
𝑇𝑜 𝑝/𝜏 𝐾
= 𝑠+(1+𝐾 (5)
𝒯𝑖 )/𝜏 𝑝
1+𝐾𝑝
𝑟=− (6)
𝜏
This root increases in magnitude as the gain increases; it moves further and further to the left
on the real axis of the complex plane as Kp increases. Real valued roots appear as coefficients
of time in the exponential function. This means the response speed increases as the gain
increases.
Steady State Error. Since the open loop transfer function is type 0, we know there will a steady
state error due to a unit step input. This error will be,
1
𝐸𝑠𝑠 = 1+𝐾 (7)
𝑝
Equation (7) shows that the steady state error decreases as the gain increases and approaches
zero as Kp approaches infinity.
𝑇𝑜 |𝑛+1 −𝑇𝑜 |𝑛 1 𝒯𝑖
+ 𝜏 𝑇𝑜 |𝑛 = (27)
∆𝑡 𝜏
𝑇𝑜 |𝑛+1 −𝑇𝑜 |𝑛 𝐾𝑝 𝐾𝑝
+ (𝑇𝑜 |𝑛 + 1) ( ) = ( ) 𝒯𝑖 (28)
∆𝑡 𝜏 𝜏
∆𝑡 ∆𝑡
𝑇𝑜 |𝑛+1 = 𝑇𝑜 |𝑛 ∙ (1 − (1 + 𝐾𝑝 )) + 𝐾𝑝 ∙ 𝒯𝑖 ∙ ( 𝜏 ) (29)
𝜏
Equation (29) shows that progressive values of To are affected by the gain (Kp) as well as the
time increment (Δt). For instance, if Δt is very small, then the effect of Kp is diminished and
successive changes in To are very small. This approximates the continuous function case.
Stability. Increasing the time increment (Δt) has an effect similar to that of increasing
Kp. When the time increment (Δt) and proportional gain (Kp) take on values such that,
∆𝑡
(1 − (1 + 𝐾𝑝 )) < −1 (30)
𝜏
Or,
∆𝑡
(1 + 𝐾𝑝 ) > 2 (31)
𝜏
Or,
2𝜏
𝐾𝑝 > ∆𝑡 − 1 (32)
then, oscillations between successive points will occur, the values will increase in magnitude,
and the recursive series will never stabilize. At this point, the system is declared unstable.
Steady State Error. If this system is stable, the steady state value can be determined by
setting 𝑇𝑜 |𝑛+1 = 𝑇𝑜 |𝑛 . That is, if a steady state response is reached, then successive values
from equation (29) are repetitive. We obtain the steady state value by setting 𝑇𝑜 |𝑛+1 = 𝑇𝑜 |𝑛 =
𝑇𝑜−𝑠𝑠 in equation (29),
∆𝑡 ∆𝑡
𝑇𝑜 |𝑛 = 𝑇𝑜 |𝑛 ∙ (1 − (1 + 𝐾𝑝 )) + 𝐾𝑝 ∙ 𝒯𝑖 ∙ ( 𝜏 ) (33)
𝜏
Solve for 𝑇𝑜 |𝑛 ,
𝐾𝑝
𝑇𝑜−𝑠𝑠 = 𝒯𝑖 ∙ 1+𝐾 (34)
𝑝
𝐾𝑝 1
𝐸𝑠𝑠 = 𝑇𝑜−𝑠𝑠 − 𝒯𝑖 = 𝒯𝑖 (1 − 1+𝐾 ) = 𝒯𝑖 (1+𝐾 ) (35)
𝑝 𝑝
Equation (35) is directly comparable to equation (7) for the steady state error of a unit step
input.
𝑇𝑜 𝐾𝑝 (𝑠+𝐾𝑖 )/𝜏
= 𝑠2 +(1+𝐾 (9)
𝒯𝑖 𝑝 )𝑠⁄𝜏 +𝐾𝑝 𝐾𝑖 /𝜏
To convert this transfer function to a difference equation, we will use equation (26) to
approximate the first derivative and equation (36) to approximate the second derivative.
The difference equation of the s-domain closed loop transfer function, equation (9), is,
1 𝜏 𝜏
𝑇𝑜 |𝑛+1 = 𝜏 ∙ [(2 ∆𝑡 + 1 + 𝐾𝑝 − 𝐾𝑝 𝐾𝑖 ∆𝑡) 𝑇𝑜 |𝑛 − ∆𝑡 𝑇𝑜 |𝑛−1 + 𝒯𝑖 𝐾𝑝 𝐾𝑖 ∆𝑡] (40)
+1+𝐾𝑝
∆𝑡
Steady State Error. If this system is stable, the steady state value can be determined by
setting 𝑇𝑜 |𝑛+1 = 𝑇𝑜 |𝑛−1 = 𝑇𝑜 |𝑛 . That is, if a steady state value is reached, then successive
values from equation (40) are repetitive. We obtain the steady state value by setting 𝑇𝑜 |𝑛+1 =
𝑇𝑜 |𝑛−1 = 𝑇𝑜 |𝑛 = 𝑇𝑜−𝑠𝑠 and in equation (40). After some manipulation we find,
𝑇𝑜−𝑠𝑠 = 𝒯𝑖 (41)
Equation (42) is consistent with our understanding of a type 1 system subjected to a step input.
2. When the output has stabilized at room temperature, plug in the heating element. You
should observe the “S” shaped response similar to the Ziegler-Nichols first method
figure 10-3 in the tuning handout.
3. When the output temperature has stabilized, terminate the program and open the
“fanout.csv” file in Excel. Use the first method of Ziegler-Nichols to estimate the gain
for 1) proportional only control and 2) proportional + integral control. Save the
response for each of these control actions.
2. Repeat step #1 above with a time increment of 1 second. Save the response file. Repeat
this process, increasing the sample time increment by 0.5 second each time until
instability occurs.
3. Use the proportional and integral gains determined in part 3: step #3 and run the
program with PI control and a sample time increment of 0.25 seconds. Save the
response file.
4. Repeat step #3 above with a time increment of 1 second. Save the response file. Repeat
this process, increasing the sample time increment by 0.5 second each time until
instability occurs.