ASSIGNMENT-5

%Assignment submitted by
%Name: Jeswin Tom Joseph
%Roll_no: CH17B052

1. The process is

The measuring element is

Effective open loop transfer function is

a.
The root locus sketch
Asymptotic Angles

here, P=4 and Z=2

Hence

Therefore, the asymptotic angles are

Centroid

Hence the centroid is at (-5,0)

Break in points

For break-in points, gain attains local maximum.

for local maxima, differentiating wrt s, and solve.

Differentiating and equating to zero, we have

Using a solver to solve the equations, we have

on real axis which is not a part 0f the root locus.

Hence,
 Angle of arrival

Angle of arrival at a zero is

For the zero

Hence angle of arrival at

b.

%defining the transfer function

Gp = tf([1 4 8],[1 14 43 30 0]);

%root locus plot

hrl = rlocusplot(Gp);
%to increase visibility of plot
 a_rl = findobj(gca,'type','line');
set(a_rl,'Linewidth',1,'Markersize',8)

c.

Using the 'rltool' command, we can find find the point at which the system achieves a damping

%rltool(Gp);

One of the poles is at -0.899+2.06i

%using rlocfind to get the gain

Kc_1=rlocfind(Gp,-0.899+2.06i)

Kc_1 = 27.4402

Hence

Further details are as shown in the following image.

d.

PI Controller is used.

For getting ultimate gain, we need to find the crossover point into RHP for the root locus.

The given transfer function is

For the PI controller, we have effective open loop transfer function as

assuming to br the design parameter

Since , we have

Finding the crossover point for the given open loop transfer function
%Defining the numerator and denominator part of transfer function
s=tf('s');
num1=s^2+4*s+8;
den1=s*(s+1)*(s+3)*(s+10);

%Effective Open loop Transfer function

Gs1=num1/(s*(den1+Kc_1*num1))

Gs1 =

s^2 + 4 s + 8
----------------------------------------------
s^5 + 14 s^4 + 70.44 s^3 + 139.8 s^2 + 219.5 s

Continuous-time transfer function.

%Root locus plot for the transfer function

Plot_1d=rlocusplot(Gs1)

Plot_1d =

resppack.rlplot

%to increase visibility of plot

a_r2 = findobj(gca,'type','line');
set(a_r2,'Linewidth',1,'Markersize',8)
 %Finding crossover frequency and associated design parameter
[Gm1,~,Wc1,~] = margin(Gs1)

Gm1 = 156.0781
Wc1 = 3.9121

Clearly crossover frequency is 3.9121 rad/s

Hence, we have the ultimate equal to 156.0781.

2.

The process transfer function is

For a P type controller, assuming a gain of , we have the closed loop transfer function as

a.

The gain margin of the process is 8.2 dB.

Open Loop tranfer function,

Kc_2=1.0825;

%defining transfer function

Gp_2=tf([2 2],[10 7 1],'iodelay',2);
%defining PController
Gc_2a=Kc_2;
%defining the system
Lg_2a=series(Gc_2a,Gp_2)

Lg_2a =

2.165 s + 2.165
exp(-2*s) * ----------------
10 s^2 + 7 s + 1

Continuous-time transfer function.

%Plotting the bode plots for the system

Plot_2a=bodeplot(Lg_2a)

Plot_2a =

resppack.bodeplot

Plot_2a.showCharacteristic('MinimumStabilityMargins')
 Clearly from the plot, the Phase margin is 69.3 degrees

b.

The PI controller to be designed has the form

Also,

Hence the transfer function is

x=fsolve(@gain_find,[1 11]);
%function to find the gain and frequency
function f=gain_find(x0)
Kp = 1.0825;
%defining the initial values
w = x0(1);
k = x0(2);
f(1) = 2*sqrt(w^2+1)*sqrt(Kp^2+(Kp/(w*k))^2)- (sqrt(25*w^2+1)*sqrt(4*w^2+1));
f(2) = pi+atan(w)-2*w +pi-atan(1/(w*k)) - atan(5*w) - atan(2*w) + 2*pi/3;
end

Equation solved.

fsolve completed because the vector of function values is near zero

as measured by the value of the function tolerance, and
the problem appears regular as measured by the gradient.

<stopping criteria details>

 tau = x(2)

tau = 0.0334

Clearly,

%defining the PI controller

Kc_2b=Kc_2*(1+(1/(s*tau)))

Kc_2b =

0.03617 s + 1.083
-----------------
0.03342 s

Continuous-time transfer function.

%defining the transfer function

Lg_2b=Kc_2b*((2*(s+1)*exp(-2*s))/((5*s+1)*(2*s+1)))

Lg_2b =

0.07235 s^2 + 2.237 s + 2.165

exp(-2*s) * -----------------------------------
0.3342 s^3 + 0.2339 s^2 + 0.03342 s

Continuous-time transfer function.

%Plotting the bode diagrms to get Gain Margin

Plot_2b=bodeplot(Lg_2b)

Plot_2b =

resppack.bodeplot

Plot_2b.showCharacteristic('MinimumStabilityMargins')
 Hence the corresponding Gain Margin is 3.37 dB

c.

Step response of the system

step(Lg_2b)
d.

The sensitivity function is

Here

Bode Sensitivity integral for P controller

%Finding the sensitivity integral for Gc_2a=Kc2=1.0825

%vector for finding the bode response

w = linspace(0.001,50, 1000000);

%Sensitivity integral for Kc2=1.0825

S_2d1 = 1/(1 + Gp_2*Kc_2);

%data points for computing the integral

[magnitude_2d1,~,~] = bode(S_2d1, w);
integral_2d1=trapz(l, log(abs(magnitude_2d1(:))))

integral_2d1 = -7.0842e-04

Clearly, we find that the sensitivity integral holds for a P controller

Bode Sensitivity integral for PI controller

%Finding the sensitivity integral for

Gc_2b=Kc2(1+1/(tau*s))=1.0825(1+1/(0.0334*s))

%Sensitivity integral for Kc2=1.0825

S_2d2 = 1/(1 + (Gp_2*Kc_2b));

%data points for computing the integral

[magnitude_2d2,~,~] = bode(S_2d2, w);
integral_2d2=trapz(l, log(abs(magnitude_2d2(:))))

integral_2d2 = -5.3354
3.

The given transfer function us

a.

Ignoring the delay, we can write the characteristic equation as

b.

The given transfer function us and controller is

%defining the transfer function

Gs_3=tf(2,[1 3 -10],'iodelay',1)

Gs_3 =

2
exp(-1*s) * --------------
s^2 + 3 s - 10

Continuous-time transfer function.

 %plotting the nyquist diagrams to calculate frequency
Plot_3b=nyquistplot(Gs_3)

Plot_3b =

resppack.nyquistplot

Plot_3b.showCharacteristic('MinimumStabilityMargins')

Clearly from the plot we have

Hence

We know,

Therefore

c.

The two controllers are

Gc1=6;
Gc2=4.207;

%Transfer function neglecting delay term

Gs_3a=tf(2,[1 3 -10]);

%Specifying closed loop systems for both controllers

L_3a=Gc1*Gs_3a/(1+Gc1*Gs_3a);
L_3b=Gc2*Gs_3/(1+Gc2*Gs_3);

%Plotting step response of system with controller Gc1

Plot_3c1=stepplot(L_3a)

Plot_3c1 =

resppack.timeplot

Plot_3c1.showCharacteristic('PeakResponse')

%Plotting step response of system with controller Gc2

Plot_3c2=stepplot(L_3b)

Plot_3c2 =

resppack.timeplot

Plot_3c2.showCharacteristic('PeakResponse')
Clearly, from the plots, we can see that :

• For using a controller neglecting the delay term, the system is stable, with settling time
of 7 seconds.

• Using a controller renders the system unstable.

Using Pade's second order approximation

is a root of this equation,

 Gc3=2.2105;

%Specifying closed loop system

L_3c=Gc3*Gs_3a/(1+Gc3*Gs_3a);

%Plotting step response of system with controller Gc3

Plot_3c3=stepplot(L_3c)

Plot_3c3 =

resppack.timeplot

Plot_3c3.showCharacteristic('PeakResponse')

From the plot, we can see that by using Pade's Second order approximation and controller the
system becomes unstable