Sei sulla pagina 1di 32

18BEI0004

GOUTHAM KRISHNA

Interacting and Non-Interacting System


18BEI0004

Experiment-1
Aim:
To study the dynamic response of single and multi capacity processes when connected in
interacting and non-interacting mode.

THEORY:
Description of system: The setup consists of a supply tank, pump for water circulation,
rotameter for flow measurement, transparent tanks with graduated scales, which can be
connected, in interacting and non-interacting mode. The components are assembled on the
frame to form tabletop mounting.

1
18BEI0004
GOUTHAM KRISHNA

2
18BEI0004
GOUTHAM KRISHNA

3
18BEI0004
GOUTHAM KRISHNA

4
18BEI0004
GOUTHAM KRISHNA
1.​ ​Step response of first-order systems arranged in non- interacting mode
In the non-interacting system, we assume the tanks have a uniform cross-sectional area
and the flow resistance is linear. To find out the transfer function of the system that relates
h2 to q, writing a mass balance around the tanks, we proceed as follows

5
18BEI0004
GOUTHAM KRISHNA
Experiment Procedure
• Prepare the set up for non-interacting mode operation.
• Switch on the pump and adjust the flow to 40 LPH. Allow the level of both the tanks (Tank
1 & tank 2) to reach steady-state and record the initial flow and steady-state levels of both
tanks.
• Apply the step change with increasing the rotameter flow by 10 LPH.
• Record the level of Tank 2 at the interval of 30 sec until the level reaches a steady state.
• Record final flow and steady-state level of Tank1
• Carry out the calculations as mentioned in the calculation part and compare the predicted
and observed values of the tank level.
Observations
The diameter of tanks: ID 92mm
Initial flow rate (LPH): 90lph
The initial steady-state level of Tank 1 (mm): 131.5mm
The initial steady-state level of Tank 2 (mm): 70.5mm
Final flow rate (LPH): 100lph
The final steady-state level of Tank 1 (mm): 166.5mm
The final steady-state level of Tank 2 (mm): 92mm
(Fill up columns H(t) observed and H(t) predicted after calculations)

6
18BEI0004
GOUTHAM KRISHNA
MATLAB CODE :

MATLAB OUTPUT:

7
18BEI0004
GOUTHAM KRISHNA
INFERENCE
In the output, the unit-step responses for the two cases are plotted to show the effect of
interaction. From this figure, it can be seen that interaction slows up the response. This
result can be understood on physical grounds in the following way: If the same size step
change is introduced into the two systems, the flow from tank 1 ( q1) for the noninteracting
case will not be reduced by the increase in level in tank 2. However, for the interacting case,
the flow q1 will be reduced by the buildup of level in tank 2. At any time t1 following the
introduction of the step input, q1 for the interacting case will be less than for the
noninteracting case with the result that h2 (or q2 ) will increase at a slower rate.

In general, the effect of interaction on a system containing two first-order lags is to change
the ratio of effective time constants in the interacting system. In terms of the transient
response, this means that the interacting system is more sluggish than the noninteracting
system.

Experiment No. – 2
Simulation of Nonlinear process model using SIMULINK Objective: A cylindrical tank as
shown in the figure has a cross-sectional area of 2 m2 and valve resistance is 0.1 s/m2.
Simulate the system and get the level response when the inlet flow suddenly increased to 1
m3 /s.

Assume: Qi, Qo is volumetric flow rates


Density is constant
The outlet valve is linear and R = √h/Qo
Mass Balance Equation
𝜌𝐴*(𝑑ℎ/ 𝑑𝑡) = 𝜌𝑖*𝑄𝑖 − 𝜌𝑜*𝑄𝑜
𝑑ℎ/𝑑𝑡 = 1 /𝐴 *(𝑄𝑖 − (√ℎ)/ 𝑅 )
𝑑ℎ/𝑑𝑡 = 1/2 *(𝑄𝑖 − (√ℎ)/ 0.1 )

8
18BEI0004
GOUTHAM KRISHNA

9
18BEI0004
GOUTHAM KRISHNA

10
18BEI0004
GOUTHAM KRISHNA

OUTPUT:

11
18BEI0004
GOUTHAM KRISHNA

Objective:
An interacting system as shown in the figure below has two cylindrical tanks. The process
parameters are given below.
A1=2 m2; R1=0.1 s/m2;
A2=1 m2; R2=0.2 s/m2; R3=0.1 s/m2;
Simulate the system and get the level responses when the first tank’s inlet flow suddenly
increased to 1 m3/s and the second tank’s inlet flow suddenly increased to 1 m3/s after 4
seconds.

12
18BEI0004
GOUTHAM KRISHNA
dh1/dt=1/A* (Qi1-(h1- h2)/R2 - h1/R2 )

dh2/dt=1/A* (Qi2+(h1- h2)/R2 - h2/R3 )

Simulation:

13
18BEI0004
GOUTHAM KRISHNA
OUTPUT RESPONSE

INFERENCE:
The output response from system 1 is influenced by the output from system 2.
The overall Transfer Function of the whole process is not just the product of the transfer
function in the series.

******___________******

14
18BEI0004
GOUTHAM KRISHNA
EXAMPLE EXPERIEMENT: CONICAL TANK SYSTEM

AIM:

To study the level response of a conical tank system and plot it.

CALCULATIONS:

15
18BEI0004
GOUTHAM KRISHNA
CONICAL EQUATION
function
dh=coni(t,h)
global ​fin
teta=0
.5;
b=0.5;
x=1;
dh= (fin-(b*x*sqrt(h)))/(pi*tan(teta)^2*h^2)

MAIN CODE:

clc;
close ​all​;
clear​ all​;
global​ fin
%fin=0.2215;
%h=1; r=1; Cs=0.05; Xs=1; g=9.8;
h0=0.2;
tin=0;
tfin=0;
ts=0.02;
simt=30;
for​ k=1:(simt/ts);
h1(k)=h0;
if (k>0 && k<=300)
fin=0.27386;
​ end
​if ​(k>300 && k<=600)
fin=0.3536;
end
if​ (k>600 && k<=1500)
fin=0.4472;
​end
time(k)=tfin;
tfin=tin+ts;
finp(k)=fin;
[t,h] = ode45(@coni_s,[tin tfin],h0);
h0=h(length(t),:);
tin=tfin;
end
plot(time,h1);
hold ​on​;
plot(time,finp,​'r'​);
grid ​on

16
18BEI0004
GOUTHAM KRISHNA

OUTPUT RESPONSE:

INFERENCE:
1)The conical tank system is a non-linear system, ​due to the variation of the area
with respect to height.
2)​Since the radius of the cross-sectional area (r) at the surface is proportional to
the height of the level, it is expected that the decrease in the processing time
constant is much larger than the decrease in process gain with a decrease in level.

17
18BEI0004
GOUTHAM KRISHNA

DIGITAL ASSIGNMENT-2
PROCESS AUTOMATION AND CONTROL LAB
LAB SLOT:- L33+L34

NAME: GOUTHAM KRISHNA


REG.NO: 18BEI0004

EXPERIMENT NUMBER 3:​ ​PID CONTROLLER TUNING USING PROCESS REACTION


CURVE OR COHEN-COON METHOD

PID controller tuning using the Process Reaction Curve or Cohen-Coon


Method

AIM:
Consider the three-tank system as in fig. Below

According to this approach, first, an approximate First Order Plus Delay Time (FOPDT)
representation of the process is identified based on the process reaction curve and then
the PID controller parameters are obtained using the appropriate Cohen-coon tuning
rules.

Procedure:

Step 1. ​Build the Process model as an Open loop configuration in the SIMULINK model
and save it in the current MATLAB folder. An S-Shaped curve is obtained which is
known as ‘Process Reaction Curve’.

Step 2. ​Draw a tangent to the curve and manipulate the value of the gain K, time
constant τ, and time-delays (td) of the approximate model and Find the FOPDT
model of the plant.

Step 3. ​Plot the response of both the original plant model and approximate model in
such a way to obtain a response that is as close as possible to the original process

1
18BEI0004
GOUTHAM KRISHNA
response.

Step 4. ​Use the effective time constant, effective gain, and effective time delay
obtained in Step 2 (which provides the best approximations of the original third
order system) and calculate the PID controller parameters using the Cohen-coon
Tuning rules.

Step 5. ​Build a new SIMULINK diagram with P, PI, PID Controller using the calculated
tuning parameters from step 4 to simulate the closed-loop response. Compare the
closed-loop performance for all the cases.

Step 6. ​Calculate the P, PI, PID controller settings using the C-C Method as in following
Table T 2.1

Step 7. ​Observe the closed-loop performance for P, PI, PID controller using the
obtained values by substituting in the SIMULINK Model, and tabulate the performance.

2
18BEI0004
GOUTHAM KRISHNA
CALCULATIONS:

K=6, τ=15.3, td=2.73

TABLE:

Type of Control Kc τI τD

Proportional (P) 0.99 - -

Proportional-Integral (PI) 0.85 6.63 -

Proportional-Integral-Deriv 1.29 6.26 0.96


ative (PID)

3
18BEI0004
GOUTHAM KRISHNA

COHEN-COON
APPROACH
Controllers P PI PID

%overshoot 62.64 20.14 67.4


1
Settling Time 84.15 249.2 84.3
4 6
Rise 3.06 4.42 2.61
Time
Peak 8.9 9.7 7.4
Time
Offset 1-0.8559=0. 1-0.9007=0.1 1-1=
14 0
=0(approx.)

OUTPUT RESPONSES FROM SIMULINK

The Open-loop block diagram for Process Reaction curve:

MATLAB RESPONSE

4
18BEI0004
GOUTHAM KRISHNA

 
Open Loop Block Diagram of both Original & Approximate (with time delay) Plant
Model:

RESPONSE FROM THE SIMULINK BLOCK DIAGRAM

5
18BEI0004
GOUTHAM KRISHNA

Closed Loop Block Diagram for Controllers:

The block diagram inculcating all the controllers is given below:

6
18BEI0004
GOUTHAM KRISHNA

The Closed Loop Responses for each controller is produced below:

P-CONTROLLER
The response is given below

P-I CONTROLLER
The response is given below

PID CONTROLLER
The response is given below

7
18BEI0004
GOUTHAM KRISHNA

MATLAB CODE TO GET THE REQUIRED OUTPUT

>>plot(t,P)
>>hold ​on
>>plot(t,PI);
>>plot(t,PID)
>>hold ​off

OUTPUT RESPONSE FROM MATLAB

8
18BEI0004
GOUTHAM KRISHNA

Inference:
1)The Cohen-Coon tuning rules work well on processes where the dead time is less than
two times the length of the time constant.
2)Cohen-Coon tuning rules are suitable for use on the self-regulating process if the
control objective is having a fast response.
3)The approximate plant model has a smoother curve than the original plant model,
although it settles at the same final value
4)Integral action makes the system sluggish and highly oscillatory, as can be seen in
the plot. It has a large settling time. In the PID controller, the offset is eliminated and
the system settles fast.

*******________*******

9
18BEI0004
GOUTHAM KRISHNA

EXPERIMENT 4: PID CONTROLLER TUNING USING ZEIGLER-NICHOLS TECHNIQUE

Aim:
Consider the three-tank system as in fig. Below

Tune a PID controller using Ziegler-Nichols tuning rules. It is based on the similar idea
as the ZN method described in the lecture, with the difference that the ultimate gain
and ultimate period are determined experimentally, not analytically.

Procedure:

Z-N Tuning: Practical Approach

Consider the given plant transfer function as Gp(S) and controller Transfer function
Gc(S) with gain Kc or Kp. Therefore, the characteristics equation for the system with the
controller and process only is given by

Step 1. ​To find the value of ultimate gain (Ku) the closed loop system is at the verge of
instability, equation 1 can be written as

Step 2. ​Substitute S=ωj into the characteristic equation, we have

Step 3. ​Rearrange in the Cartesian form: Re+ Im j and consider it as equation(2)

Step 4. ​This equ.2 can be satisfied if and only if both the real and imaginary parts are
equal to zero. Solve the equations to get a value for Kc and ω. Hence, the ultimate gain is
Ku = Kc =

10
18BEI0004
GOUTHAM KRISHNA

Step 5. ​And the frequency of sustained oscillations at that point is: ω The ultimate period
Pu is given by

Step 6​. Calculate the P, PI, PID controller settings using the Z-N Method as in the following
Table T.1.1.

Step 7. ​Observe the closed-loop performance for P, PI, PID controller using the
obtained values by substituting in the SIMULINK Model. Record the necessary
information in the format given.

CALCULATIONS:

11
18BEI0004
GOUTHAM KRISHNA

Type of Control Kc τI τD

Proportional (P) 0.835 - -

Proportional-Integral (PI) 0.7515 10.46 -


7
Proportional-Integral-Deriv 1.002 6.28 1.57
ative (PID)

Parameters Entered:

Control P I D

P 0.835 - -

PI 0.7515 0.0717 -

PID 1.002 0.1595 1.5734

Performance Comparison:

Controllers P PI PID

% 55.4 61.41 44.32


overshoot
Settling 66.56 125.99 34.07
Time
Rise Time 3.37 3.62 2.92

Peak Time 9.10 9.9 7.5

Offset 0.167 0 0

SIMULINK BLOCK DIAGRAM:

12
18BEI0004
GOUTHAM KRISHNA

P CONTROLLER RESPONSE:

P-I CONTROLLER RESPONSE:

13
18BEI0004
GOUTHAM KRISHNA
PID CONTROLLER RESPONSE:

OVERALL RESPONSE INCLUDING ALL THREE CONTROLLERS:

14
18BEI0004
GOUTHAM KRISHNA

Inferences:
1)The Ziegler-Nichols rules work well only on processes where the dead time is less
than half the length of the time constant.
2)Ziegler-Nichols tuning rules work well only on processes with very high time
constants relative to their dead times, and on level control loops.
3)Integral action makes the system more sluggish. It has the highest settling time.
However, it eliminates offset.
4)Overshoot increases in the proportional-integral controller.
5)Integral action may lead to instability and makes the response more oscillatory.
6)that rise time and the percentage overshoot is reduced compared with the use of P and
PI controller. And also percentage overshoot is reduced unlike in the case of Cohen Coon
method were the overshoot is larger than that of the P controller.

COHEN COON VS ZIGLER NICHOLS

​COHEN COON ​ZIGLER NICHOLS

1)Tuning suited to a wide variety of 1)Tuning limited to some processes only


processes
2)More Flexible 2)Less Flexible
3)Designed to give fast response 3)Can be slow in some case in giving a
response

15

Potrebbero piacerti anche