Due Date: 23/09/2022 CH 246: Homework 3 Total Marks: 100

Problem 1. (30 points) An electrically heated, stirred-tank system is shown in Figure 1 Using the given
information, do the following:

Figure 1: Continuous flow stirred tank chemical reactor

(a) (10 points) Draw a block diagram for the case where T3 is the controlled variable and voltage signal
V2 is the manipulated variable. Derive an expression for each transfer function.
The block diagram using transfer functions obtained from additional information is shown in
Figure 2

Figure 2:

(b) (10 points) Repeat part (a) using V1 as the manipulated variable.
The block diagram using transfer functions obtained from additional information is shown in
Figure 3

(c) (10 points) Which of these two control configurations would provide better control? Justify your an-
swer.
The control configuration in part (a) will provide the better control. As is evident from the block
diagrams above, the feedback loop contains, in addition to Gc, only a first-order process in part
(a), but a second-order- plus-time-delay process in part (b). Hence the controlled variable responds
faster to changes in the manipulated variable for part (a).
1 of 4
 Figure 3:

Available Information:
1. The volume of liquid in each tank is kept constant using an overflow line.
The outlets of both the tanks have flow rate q0 at all times.

2. Temperature T0 is constant.
T0 (s) = 0

3. A 0.75-m3 /min decrease in q0 ultimately makes T1 increase by 3 0 C. Two-thirds of this total tem-
perature change occurs in 12 min. This change in q0 ultimately results in a 5 0 C increase in T3 .
T′ (t)
T′ (∞)
= 1 – e–t/τ1 or 23 = 1 – e12/τI , τI = 10.9 min
T1 (s) o 3 pm)
Qo (s) = 3 C/(–0.75m
10.9s+1
–4
= 10.9s+1
T3 (s) o 3 pm)
= (5–3) C/(–0.75m
τ2 s+1 = τ –4
Qo (s) 2 s+1

4. A change in V1 from 10 to 12 volts ultimately causes T1 to change from 70 to 78 0 C. A similar test

5. A step change in T2 produces a transient response in T3 that is essentially complete in 50 (=5 τ)

Since inlet and outlet flow rates for tank 2 are qO and volume of the tanks are equal
T3 (s) 1
T (s) = 10s+1
2

2 of 4
 6. The thermocouple output is amplified to give V3 = 0.15T3 + 5, where V3 [=] volts and T3 [=] 0 C.
V3 (s)
T3 (s) = 0.5

7. The pipe connecting the two tanks has a mean residence time of 30 s.
 
T2 (t) = T1 t – 30
60 = T1 (t – 0.5)
T2 (s)
T1 (s) = e–0.5s

Problem 2. (70 points) The dynamic behavior of a polymerization process has been approximated by
the transfer function model:
–550.0 –4.5
Y(s) = U(s) + D(s) (1)
11.0s + 1 (5.0s + 1)(25.0s + 1)
where in deviation variables, Y is the product’s weight average molecular weight, U is the chain transfer
agent flow rate (in kg/hr), D is the impurities concentration in the solvent stream (in ppm); time is in
minutes.
(a) (10 points) If a feedforward control strategy alone is to be used for this process, design this controller
and show in a block diagram how it will be implemented. Assume that the Y and D measurements
are both available instantaneously, with no measuring device dynamics. Is this feedforward con-
troller realisable? Justify your answer briefly.
Upon introducing the elements of the transfer function model given for the polymer process, we
obtain the following expression for the feedforward controller:
  h i
4.5(11s+1)
u(s) = 11s+1
–550 YS P – 550(5s+1)(25s+1) D(s)
u(s) = Gs tYS P + Gff D(s)
The implementation block diagram is show in Figure 4. The indicated controllers are implementable
because there arre no predictive elements, neither are there any unstable elemants; Gst is imple-
mentable as a PD controller. Gff as a second order System with a single zero

Figure 4:

(b) (20 points) Given now that a feedback controller with transfer function:
 
1
GC (s) = –0.004 1 + (2)
12.0s
is to be used on the process, and that in reality, the "Real" process gain is actually -600.0, and that the
disturbance gain is - 4.0, but the time constants were correctly estimated, implement this feedback
controller alone on the "Real" process, and obtain the closed-loop response to a step change of 20
ppm in the impurities concentration. Plot this response.
The response is shown in Figure 5:

(c) (20 points) For the "Real" process as given in part (b), implement the feedforward controller designed
in part (a); obtain and plot the control system response to the same step change of 20 ppm in the
impurities concentration. Comment on the performance of this control scheme. What effect, if any,
does the modelling error have on the performance?
The response is shown in Figure 5 and the resulting offset is evident. Note that in practice,
however, such an offset of abut 20 in the number average molecular weight will not be noticeable.
The offset is clearly a direct result of the model error.

3 of 4
 Figure 5:

(d) (20 points) Now implement feedback/feedforward controller ( the feedback controller of part (b)
and the feedforward controller of (a) ) on the process; obtain the overall system response to the
same step change of 20 ppm in the impurities concentration. Compare this response with the ones
obtained in parts (c) and (d), and with reference to this specific example, briefly comment on the
strengths and weaknesses of each control of the schemes investigated: feedback alone, feedfor-
ward alone, and feedforward/feedback.
The response is shown in Figure 5. As illustrated here, clearly feed forward alone is quite suscep-
tible to the effect of model errors. The amount of offset will depend on the amount of model error.
Feedback alone will leave no offset (if there is integral action) but observe that the process upset
lasted for close to 100 minutes: feedback control is unable to correct for upsets until the process
is affected. When the feedferward scheme is augmented with feedback, we see the anticipatory
nature of feed forward cutting down the maximum amount of deviation from set -point; we also
see the feedback eliminating the offset. This combination yielded the best results.

4 of 4