You are on page 1of 50

PROIECT I.R.A.

Nume student: Grupa Nota:

FLOW AND TEMPERATURE CONTROL OF


A GRANULAR MATERIAL
PROIECT I.R.A.

Nume student Grupa

A.7. IDENTIFICATION PROBLEMS FOR THE FIXED PART

Identification of the flow control loop


We start with the analysis of the components of the loop and represent the results of the
identifications.
The loop is made of:
a – conveyor (TM);
b – transporter with cups (TC);
c –gravimetrical doze (DG) with adaptor.
a) concernig the conveyor:
a.1.the power amplifier receives the Ua voltage at the input and generates the voltage Um at
the output; it has the structure from the figure, in which:
Ua Um Kap =
Kap* e-sap
ap =
a.2.the conveyor motor receives at the input the voltage Um and generates at the output the
quantity tm. It has the structure from the figure, in which:
K1 = [Nm/V]
Mr K2 = [rad/sec/Nm]
Um K1 - K2 tm Tm1 = [sec]
Tm1s+1 Tm2s+1 Tm2 = [sec]

a.3.the tahogenerator for measuring the revolution (TG) and its adaptor receive at the input
the revolution tm and generate by means of the adaptor a voltage U. It has the structure from the
figure, in which:

2
PROIECT I.R.A.

Nume student Grupa

Kt = [U/rad/s]
tm Kt U Tt = [sec]
Tts+1

a.4. conveyor (TM) receives at the input the revolution tm and provides steered
intermediate flow of granular material. The structure is given in the figure below:

Pm1 0.05 TQ1 = TTM = [sec]


TBs(TQ1s+1) Ktm = [kg/sec/rad/sec]
TB = [sec]

Ktm + Qi
Ttms+1 +

b) The conveyor with bowls (TC) has at the input the intermediate flow (Q 1), and at the
output the flow (Qm). The structure is given in the figure below, in which:

K=
Q1 Qm
T= [sec]
m = [sec]

c) Gravimeter dose with controller: we consider the flow (flow capacity) Qm as the input
and the electric current iq as the output of the controller, such as, in the fugure below, we have:

Qm iQm

KG = [mA / kg / sec]

3
PROIECT I.R.A.

Nume student Grupa

TG = [sec]

4
PROIECT I.R.A.

Nume student Grupa

Identification of the temperature control loop

We start with the analysis of the components of the loop and represent the results of the
identifications.
a) The pneumatic vent (PV) and the electropneumatic convertor (EPC)
- we apply the current iE at the input and the output is the flow of the marsh gas (q)
pg
Kpg

Kpg =
Ie Kce.Kv + q Kce*Kv = [Nm3/s/mA]
+ Tv = [sec]
Tv.s+1

b) The furnace (C) will be analized together with the conveyor with bowls (TC). The
structure is reprezented in the figure below:
Kc = [o C/Nm3/sec]
KT =
I
K
KZ =
Tzs+1 TZ= [sec]
Tc = [sec]
Kc.e -sc +
q Kst.e -st m c = [sec]
Tcs+1 + Ts+1
TT = [sec]

T = [sec]

c) The temperature transducers: for the pyrometer, the input is the temperature m, the
output of the controller is the electric current Im, and for the thermoresistance (TR), the input is the
air temperature from the furnace and the output is the electric current Ic.
5
PROIECT I.R.A.

Nume student Grupa

- for the pyrometer


m Km lm Km = [mA/oC)
Tms+1 Tm = [sec]

- for the thermoresistance:


c KC lc Kc = [mA/oC]
TCs+1 Tc = [sec]

6
PROIECT I.R.A.

Nume student Grupa

4. COMPUTING THE REGULATORS USING THE METHOD OF


ALLOCATING POLES AND ZEROS

1.1. CONSIDERATIONS ON THE FIXED PART

The method, bazed on the equivalent order two system, is applied to the regulating loop of
the revolution:

Mr
Ua KAPe-sAP Um K1 - K2 TM
TM1s+1 TM2s+1

KT
TTs+1

Then, we apply a series of transformations in order to bring the fixed part to the form
nedeed by the Guillemin – Truxal method:
a) we neglect disturbances (Mr =0);
b) we aproximate the delay :
−τ AP s 1
e ≃ =
1+τ AP
c) we transfigure the system into the form of the unitary negative feedback.
Because Tt <<TM2, then:
K2 KT Ω K ¿M
⋅ ≃ ¿ =
T M 2 s+1 T T Ω s+1 T M s+1

in which : TM*=TM2+TT = and KM*=K2KT =


7
PROIECT I.R.A.

Nume student Grupa

d) Because the constant (TT ) remains much smaller then the constants of the loop, the
inputs and outputs have approximately the same dinamic evolution, so we can have a
simplified structure for the regulating loop of the revolution:

u0 HR TM
-

This means:

u0 HR TM
-

In this situation the majority of the performances imposed on the output TM are also valid
for (u), excepting stv for which:
stv (u)=kT· stv (TM)

1.2. COMPUTING THE REGULATOR HR1(s) IN CASE OF THE SECOND


ORDER EQUIVALENT SYSTEM WITHOUT CORECTION

a) We impose the set of performances:


stp = [V];
 [%];
tr  [sec];
B  [rad/sec];
8
PROIECT I.R.A.

Nume student Grupa

stv  [V];
b) We determine the parameters of the second order system
ω2n
H 02 ( s )= 2 =
s + 2ξω n s +ω 2n
=
n =

starting from the following performances:

c) The imposed testing:

stp = [V]; fulfilled unfulfilled

 [%]; fulfilled unfulfilled

tr  [sec]; fulfilled unfulfilled

B  [rad/sec]; fulfilled unfulfilled

stv  [V]; fulfilled unfulfilled

d) The analytical determination of the regulator is based on the figure from the beginning
of the chapter in which the element (1/KT ) is ignored.
¿
K AP⋅K 1⋅K M
H f ( s)= =
( τ AP s+1 )( T M 1 s +1 ) ( T M s +1 )
¿

9
PROIECT I.R.A.

Nume student Grupa

ω2n
H d (s )= =
s ( s+ 2ξω n )
ωn
2ξ 1
H R 1 (s)= ⋅ =

( )
1 H f ( s)
s s+1
2 ξωn

The step response and the ramp response of the closed system are:

The form of the regulator HR1(s) is to complicated, therefore we can make some
simplifications so that the relation obtained has the form of some typical regulators:
d.1) We want to make some minor changes to the coefficient (1/2n) to the form
(1/2’’n) so that we can simplify the pole from HR1(s) with one of the three zeros, without
changing the performances.
10
PROIECT I.R.A.

Nume student Grupa

’ =
’n =

'
H R 1=

H '02=

d.2) We "attach" the smallest constant (AP) to the biggest one, for example TM*:

} } ={}} {¿¿
H R1 ¿

''
H 02=

In this case the regulator can be realized: it is a PID with first order filter.
e) The tuning parameters of the regulator are determined by identification with the typical
forms of the PID type regulator and we obtain the following parameters:

VR =

i =

d =

TN =
We lay-out the graph of the closed system response H 02’ and H02” for a unity step input,
respectively ramp in comparison with the closed system response H02.

11
PROIECT I.R.A.

Nume student Grupa

Concluzions:

12
PROIECT I.R.A.

Nume student Grupa

1.3. COMPUTING THE REGULATOR HR2 FOR THE CASE OF THE


CORRECTED SECOND ORDER SYSTEM (DIPOL CORRECTION)

a) The set of the imposed performances is more severe:


stp = ;
 [%];
tr  [sec];
B  [rad/sec];
stv  [V];
e) We determine the parameters of the second order system
2
ωn
H 02 ( s )= 2 =
s + 2ξω n s +ω 2n
=
n =

starting with the following performances:

We test the performances of the system and we find that the following performances are not
fulfiled:

13
PROIECT I.R.A.

Nume student Grupa

Because the performances are not fulfiled, we apply the correction of the closed system and we
obtain the form:
2
ωn s+ z C pC
H 0C ( s )= ⋅ ⋅
s +2 ξωn s+ ωn s+ pC z C .
2 2

Computing the values of the correction pole and zero:

pc =
zc =

b) The testing is valid only for this value of the response time:
tr = [sec]; so tr < tr imposed

c) Analitic determination of the controller HR2(s)

H 0 C( s ) 1
H R 2 (s )= ⋅ =
1−H 0C (s ) H f (s )

14
PROIECT I.R.A.

Nume student Grupa

The unitary step and ramp responses of the closed system are:

The form of the controller HR2(s) is to complicated, so we need to make some


simplifications so that the obtained relation has the form of a typical controller:

Ts+1
1
β s+1
1

( )
β [ ( ) ]
1
≃ T − s+1
β
T >5
1
()
d.1) We simplify the group β if β

15
PROIECT I.R.A.

Nume student Grupa

Thus:

H 'R 2 =

H '0C =

d.2) We "attach" the reduced time constant to the big constant:

H }R2} ={}} {¿¿ ¿

''
H 0C =

Ts+1 Ts+1 1
= ≃

d.3) We search for a simplificaton of the groups


s+ β 1
β s+ 1
β
β
( )
allowable on

1
T 1
∈ ÷5
5 ( )
general if β

'} } ={}} { ¿¿
H R2 ¿

'''
H 0C =

16
PROIECT I.R.A.

Nume student Grupa

d) Determining the tuning parameters of the controller:

We lay-out the graphic of the H 0C’, H0C” şi H0C”’ closed system response for unitary step,
respectively unitary ramp input in comparison with the response of the H0C closed system.

Conclusions:

17
PROIECT I.R.A.

Nume student Grupa

5. COMPUTING THE REGULATORS USING FREQUENTIAL


METHODS BASED ON THE EQUIVALENT SECOND ORDER
SYSTEM

2.1. CONSIDERATIONS ON THE FIXED PART

We analyze the structure from the figure, in which we apply a series of simplifications:

u0 HR TM
-

K AP K1 Kf
⋅ ≃ =
τ AP s +1 T M 1 s+1 T f s+ 1

{T f =τ AP +T M 1=
K f =K AP⋅K 1 =
Also, we neglect the viscous friction of the rotor, so that:
¿ ¿
KM KM 1 1
¿ ≃ ¿ = ¿ = =
T M s+1 T M s T M Ts
s +1
K ¿M
T ¿M
T= =
K ¿M
resulting the structure in the following figure, the performances beeing imposed to (U ) because
computing them using (TM) is very simple.

18
PROIECT I.R.A.

Nume student Grupa

UA U

2.2. DETERMINING THE FACTOR (VR) OF A P REGULATOR

a) We impose the following set of performances:


stp = [V];
 [%];
tr  [sec];
B  [rad/sec];
cv  [V];
b) The computation of the gain factor (VR)
We represent in logaritmic diagrams the fixed part H f(s) using the module diagram, the
system beeing of minimal phase. At the breaking frequency rezults the point F.

19
PROIECT I.R.A.

Nume student Grupa

- We determine the cutting frequency ( ωt ) and the breaking frequency ( ω f ).


ωt =
ωf =
¿
- We determine the ( ξ )factor corresponding to the imposed overshoot σ=σ and we compute
the value |A| :
=
1
|A|≃
4 ξ2 =
which then we reprezent in decibels. At ω=ω f results the point N.
- We drag the initial characteristic (Hf) so that we have the breaking in N, rezulting the final form
of the direct, open loop ( H r =1 ). It is obvious that:
FN=V R|dB =
paying attention to the sense of the necessary translation.
- The necessary tests aim at the following performances:
o The response time: requires reading the frequency ( ωt ) from the logaritmic
ω
ωt = n
diagrams and because 2 ξ rezults that ω n=2 ξω t , so we need that
4 ¿
t r= ≤t
ξωn r ;

o The coefficient of the speed stationary error, which can be read directly from the
¿
logaritmic diagrams at ω=1 , beeing neccesary that c v ≥c v ;

o The band width can be taken orientatively Δω B≃ωt .

20
PROIECT I.R.A.

Nume student Grupa

Following-up, we reprezent the closed loop system responses for unitary step, respectively unitary
ramp inputs:

21
PROIECT I.R.A.

Nume student Grupa

Conclusions:

22
PROIECT I.R.A.

Nume student Grupa

2.3. DETERMINING THE PARAMETERS OF A PI REGULATOR

a) We impose the following set of performances:


stp = [V];
 [%];
tr  [sec];
B  [rad/sec];
cv  [V];
We try to use a simple P regulator, but this cannot satisfy all the imposed performances,
especially the one reffering to the coefficient of the speed stationary error. In theese conditions we
recomend using a PI regulator.

b) Computing the regulator parameters.


The structure of this regulator is:
1+sT z
H PI ( s )=V R
1+ sT P .

For the grapho-analytic determination of the parameters, first we have to represent H f ( jω ) in


logaritmic diagrams, the modulus diagram and the line |A| , rezulting the points F and N (see
chapter 2.2). Then we determine graphicaly the frequency ( ωt ) and the coefficient (c ) at ω=1 .
v

23
PROIECT I.R.A.

Nume student Grupa

We place the frequencies ( ω z ) and ( ω P ) such that:

{
ω z≈0,1 ωt =
c
ω p = v¿ ω =
cv z
ω <ω
in which ( p z ). With these frequencies we can determine the open, direct structure of the
system with a PI regulator (denoted HdC). The parameters of the regulator are:

{
V R|dB =
1 1
T z= = =
ω z 0,1 ωt
¿
1 1 cv
T P= = ⋅ =
ωP ωz cv
The necessary testing:

In the following we represent the closed loop system responces for a step input, respectively ramp
input:

Conclusions:
24
PROIECT I.R.A.

Nume student Grupa

2.4. DETERMINING A PD REGULATOR

a) We impose the following set of performances:


stp = [V];
 [%];
tr  [sec];
B  [rad/sec];
cv  [V];
We try using a simple proportional type of regulator, but it cannot satisfy all the imposed
perormances, especially the one reffering to the time responce. In these conditions it is recomended
using a PD regulator.

b) Computing the regulator parameters


The structure of this regulator is:
1+τ D s
H PD =V R
1+T N s .

For determining of the parameters, first we have to represent H f ( jω ) in logaritmic diagrams, the
modulus diagram and the line |A| , rezulting the points F and N (see chapter 2.2).

25
PROIECT I.R.A.

Nume student Grupa

We compute t1 =
2
t r= 2
from which ξ ωt 1 =
¿ 2 t
t r= 2 ωt 2 =ω t 1 r¿
and from ξ ω t 2 results that tr =

We place on the frequency axis ω=ω t2 , representing the final open structure.
We deduct the parameters of this PD:

{
V R|dB =
τd=
T N=
Testing the performances:

In the following we represent the closed loop system responces for a step input, respectively ramp
input:

Conclusions:

26
PROIECT I.R.A.

Nume student Grupa

2.5. DETERMINING THE PARAMETERS OF A PID REGULATOR

a) We impose the following set of performances:


stp = [V];
 [%];
tr  [sec];
B  [rad/sec];
cv  [V];
Having the previous examples, it is obviuos that it is imposible satisfing the imposed
performances using a proportional regulator. More than that, nor a PI or a PD regulator can solve
the problem, so we must use a combined regulator.

b) Computing the parameters of the regulator:


The structure of this regulator is:
1+sτ d 1+ sT z
H R ( s )=V R ⋅
1+sT N 1+ sT P .

First we represent H f ( jω ) in logaritmic diagrams.

27
PROIECT I.R.A.

Nume student Grupa

We determine (), the position of the line |A| , rezulting the point N and the structure Hd1.
=
A=

We compute (t2) and we move the structure (Hd1) to the right until (Hd2), this impling a PD
regulator
t2 =
¿
cv
(>1 )
We read cv given by Hd2, we compute cv

{
ω z≈0,1 ωt 2 =
c
ω p = v¿ ω =
cv z
We place
Rezulting the complet structure of (HdC), in which the modified PI regulator is added.

{
The parameters of the regulator will be:
V R|dB =
1 1
T z= = =
ω z 0,1 ωt
¿
1 1 cv
T P= = ⋅ =
ωP ωz c v
τ d=
ω
T N =τ d t 2 =
ωt 1

The necessary testing:

28
PROIECT I.R.A.

Nume student Grupa

In the following we represent the closed loop system responces for a step input, respectively ramp
input:

Conclusions:

29
PROIECT I.R.A.

Nume student Grupa

6. COMPUTING THE REGULATORS USING FREQUENTIAL


METHODS WITH THE ASSURANCE OF A PHASE MARGIN

3.1 CONSIDERATIONS ON THE FIXED PART

Method can be applied for the control loop of the avon temperature as shown in (fig 3.1). An
advantage of this method is that there are no necessary simplifications.

KBZ
Pg Bi TBZs+1
Kpg

ie + q Kc e-STc
+ c
KCE KV
+ +
TVs+1 TCs+1

ic KBZ

TBZs+1

3.2. COMPUTING THE PARAMETERS OF A PI REGULATOR

a) We impose the following performances:


 the sistem works in a stabilizing regime, so the temperature c= co=ct. The main function
being that of rejecting the external perturbation pg and I in this case the effect is
absolutely necessary, so *stp=0 .
 the phase margine k*=45..500, so the system is relatively weakly damped but it is
acceptable because the technologic process is not pretentios, the gain margine mk*>5 dB;

b) Computing the parameters of a PI regulator:

30
PROIECT I.R.A.

Nume student Grupa

HR=VR(1+1/si)
dequires determinig the values VR and i .
We reprezent in logaritmic diagrams the original fixed part given by:

Hf(s)=[KceKv/(Tvs+1)][Kce-sc/(Tcs+1)][Kc/(Tcs+1)]=

From the module and phase diagram we determine the frequency


'=
for which:
Hf(j')=-180o+15o+k*
We also measure:

31
PROIECT I.R.A.

Nume student Grupa

|Hf(j')|=|Hf|=
With this the parameters of the regulator will be:
VR=1/|Hf|=
i=4/'=

HR=

c) Checking the performances requires testing:


k>45o;
mk >5 dB;

3.3. COMPUTING THE PD REGULATOR

32
PROIECT I.R.A.

Nume student Grupa

a) the system works mostly in a traking regime and it must follow a precise profile o(t),
and this automatically recommends a PD regulator;
 the disturbance of the stationary errors (which are not zero) is insignificant;
 because of the frequent switching on a big phase margine, for a small overshooting it is
recommended;
*k>50..60o
and simultaneously :
m*k>5 dB .
b) Computing the PD regulator:
HR=VR[(ds+1)/(ds+1)]= VR[(ds+1)/(TNs+1)]
requires determining the paremeters VR and d because =0.120 .
From the logaritmic diagram of the structure (3.2) we determine the frequency
0=
for which :
Hf(j0)=-180o
and the value:
|Hf0(jo)|=

The tuning parameters will be:


VR=
d=
N=
so
HR(s)=

c) for the necessary testing we need to represent the compete open loop in logaritmic
diagrams and to test the performances.

33
PROIECT I.R.A.

Nume student Grupa

3.4. COMPUTING THE PARAMETERS OF A PD REGULATOR

a) The imposed performances reffer to:


 the stationary error to position is: stp=0 ;
 a good damping: k*>55..60o
mk*>5dB .

b) Computing the parameters of the regulator:

HR(s)=VR[(1+sd)/(1+sd)][(1+is)/si]

uses the reprezentation of the structure (3.2) into logaritmic diagrams, from which we determine
the frequency:
'=
34
PROIECT I.R.A.

Nume student Grupa

and
o=
and the values |Hf’|= [dB]

and |Hfo|= [dB]


having this, it is recomended:
VR=0.228/|Hfo|=
i=1.2 *To=
d=0.5*T0=
T0=2*/0=

so HR(s)=

c) Testing the performances requires us to represent the compete loop in logaritmic diagrams and
to test the performances.

Conclusions:
4. COMPUTING THE REGULATORS USING QUASIOPTIMUM
METHODS
35
PROIECT I.R.A.

Nume student Grupa

The „symmetry” and „modulus” methods will be used, applied to the control loop of the
frequency from the control system of the flow rate.

4.1. CONSIDERATIONS ON THE FIXED PART

We use the same considerations as in point (1.1), adapting the structure (fig. 1.3) in which
the direct loop will be:

Hd=HR*Hf=HR(KAP/( APs+1))*(K1KM*/(TM1s+1)(TM*s+1))=

4.2 COMPUTING THE REGULATORS USING THE „MODULUS” METHOD


We use the optimal form of the direct loop:

Hd*=1/[2Ts(Ts+1)]=
in which: T=
a) Computing a PID regulator is neccesary if the fixed part contains two big time
constants, so:

HR(s)=(TM1s+1)(TM*s+1)/2TKAP.K1.KM* =

and the parameters of the PID regulator:


VR=
i=
d=
TN 0

so HR(s)=
36
PROIECT I.R.A.

Nume student Grupa

b) Computing a PI regulator is necessary if there is only one preponderant time constant:

Hf=KapK1Km*/(APs+1)(TMs+1)=

HR(s)=Hd*/Hf=

The parameters of the PI regulator:

VR=
i=

In the case of the „modulus” method, the performances to a unitary step are: =4.3%, tr=6.75AP,
stp*=0.
Testing the performances:

4.3. COMPUTING THE REGULATORS USING THE „SYMMETRY” METHOD

37
PROIECT I.R.A.

Nume student Grupa

The optimal form of the direct loop is:

Hd*=(4Ts+1)/8T2s2(Ts+1)=

where T=

In order to use only a typical regulator, we adopt for the fixed part the structure Hf'(s) from the
relation (4.5), so:

Hf'(s)=

HR(s)= Hd*/Hf’=
with the parameters
VR=
i=
d=

In the case of the „symmetry” method, the performances of the closed system for a unitary step are:
tr*=11.1T=0.111; =43.17%, stv=0.
The graphical testing of the performances:

Conclusions:

38
PROIECT I.R.A.

Nume student Grupa

5. COPUTING THE REGULATORS USING CASCADE CONTROL

We will apply this method to the control loop of the granular material flow variant A and to
the control system of the temperature of this material.

5.1 CONSIDERATIONS ON THE FIXED PART

For the flow rate control we use .A8, in which corresponding to the relation:
KM*=K2.KT =
TM*=TM2+TT =
We simplify the structure:
(5.1) [(TTs+1)/KT]KTM/(KTMs+1)KTM*/(TTM*s+1)=
in which:
KTM*=KTM/KT =
TTM*=TTM-TT =
and respectively if the delay time m is approximated with a first order element, we have:
(5.4) [Ke-sm/(Ts+1)]Kg/(Tgs+1)[K/(Ts+1)(ms+1)]Kg/(Tgs+1)K*/(T*s+1)=
in which: K*=K.KG=
T*=T+m+TG=
resulting the structure from fig. 5.1.
M1 Pm1
Ua Kap K
- + K
iq TGs+1 Qm
KM* KTM*

aps+1 TM1s+1 TMs+1 TTMs+1 T*s+1 Kg


U
iQ

39
PROIECT I.R.A.

Nume student Grupa

Following, the performances will reffer to the output values (u ) and (iQ) of the respective
transducers (see chapter 4).
In case of the control system of the teperature we use the structure from A 12 neglecting the
perturbation pg and bringing the internal loop to the from of the unitary negative feedbak.
Following this purpose we make the sympifications:
KceKvKce-scKc/(Tvs+1)(Tcs+1)(Tcs+1)Kf Kce-sc/(Tfs+1)=
in which:
Kf= KceKvKc Kc =
Tf=Tv+Tc+Tc=
and respectively:
(Tcs+1)Kt e-st/[Kc(TTs+1)]Kt* e-st/(TT*s+1)=
in which: KT*=KT/Kc=
TT*= TT - Tc =
The structure of the temperature control loop part is the same as in fig. 5.2 .

ie Kf ic Kt*e -st m


Tfs+1 Tts+1

ic
im Km
Tms+1

5.2. COMPUTING THE REGULATORS OF THE FLOW RATE SYSTEM

We begin with the interior loop for which we have:


(4.4) HR=[(TM1s+1)(TM*s+1)]/2apaKapK1KM*=

40
PROIECT I.R.A.

Nume student Grupa

The complete structure is represented in fig. 5.3

Figure. 5.3 .
MR Pm2
iQc HRQ HR K K
AP 1
- K *
M
K *
TM
+ K*

( S+1)(T S+1) T *S+1 T *S+1 T*S+1


AP M1 M TM

Because of the „modulus” method we have that:

=
But:
AP<<TTM* si AP<<T*, then:
(5.10') H0(s)1/(2aps+1)=
In accordance with the computing method we can write that:
(5.11) H0(s)KTM*/(TTM*s+1)=KTM*/(2aps+1)(TTM*s+1)KTM*/(TTM*s+1)=
in which:
TTM**=TTM*+2AP=
and TQ=TTM**=
analogous with chapter 4.2 then for TQ=TTM** we have that:
(5.14) HRQ(s)=(T*/2TTM**KTM*K*)[(T*s+1)/T*s)
Speaking of a PI regulator.
Identifing the tuning parameters for the two structures:
(5.15) HR=
rezulting:

41
PROIECT I.R.A.

Nume student Grupa

1=
2=
VR =

and respectively:
VRQ=
iQ=
so HRQ=
As for the testing of the performances, the analysis of the responce to the perturbation (Mr)
applied to the interior loop and comparing the effect of this perturbation with the case of the
monoloop control with a single regulator have an unusual importance.
Accoring to the fig 5.3 with iQ=0; u0=0
u(s)=
For this transfer function we draw the responce to a 0.1/s step.
If we didn’t use the cascade structure, we would compute a PI regulator assuring a given
k*>450 phase margin for the fixed part appearing in figure 5.3.

42
PROIECT I.R.A.

Nume student Grupa

With the designed PI regulator we have:

And for this function we draw the responce to the 0.1/s step comparing it to the responce obtained
to the same input for the function U.

Conclusions:

5.3. COMPUTING THE REGULATOR OF THE TEMPERATURE CONTROL


SYSTEM

The structure of the fixed part has the simplified form from the figure:

im0 ic0 iE Kfeic


-sTc
Kt` e-stm
HRm HRC Tfs+1 Tt`s+1
43
PROIECT I.R.A.

Nume student Grupa

We suppose the interior loop has a PID regulator. The tuning parameters for the interior
loop result directly from the Zigler-Nichols relations:

VRC=0.9 c /Tf Kf =
ic=3.3c=

In order to maximumly simplify the computations we propose the following computations:


HRCKfe-2sc/(Tfs+1) (VRCKf-3.3cs)(1+2.3cs)(1+Tfs)(cs+1).
Hoc(s)= VRCKf(1+2.3cs)/ 3.3cs(1+Tfs)+ VRCKf(1+2.3cs)=
we determine the poles of the system
p1= ; p2= ;
We propose the form:
K
H 'OC ≃
T OC s+1 =

in which:
{ K =V RC K f /B=
T OC =( T 1 +T 2 −2. 3 τ c ) =

Replacing the interior loop with the simplified transfer function (if it is possible) we propose an
„exterior” proportional regulator:

HRm=VRm
and using the logaritmic diagrams we determine
VRmax=
which brings the exterior loop to its stability limit, and the frequency
0=
44
PROIECT I.R.A.

Nume student Grupa

which characterizes the maintained frequency, so



T 0= =
ω0
Using the Ziegler-Nichols tuning relations:

{
V Rm =0 .75 V R max =
τ im=0 . 6 T 0 =
τ im =0 . 1T 0 =

Testing the performances by simulation:

Conlusions:

6. COMPUTING A PREDICTION REGULATOR

Because of the preponderant delay time of the flow rate control loop (fig A8) obtaining
good performsnces requires a model based prediction regulator.

45
PROIECT I.R.A.

Nume student Grupa

6.1 CONSIDERATIOND ON THE FIXED PART

The interior control loop of the revolution is computed using the „modulus” method, so:
H0(s)=
From fig. 5.3 results that the structure of the granulat material control loop (6.1) in which H rp is the
prediction regulator and the fixed part:

l0 KTM*K* e-sm i


HRP (2aps+1)(TTM*s+1)(T*s+1)

(6.2) Hf= KTM*K* e-sm/(2aps+1)(TTM*s+1)(T1*s+1)=Hf`(s) e-sm


Hf=
in which: T1*=T+TG=

6.2. COMPUTING THE PREDICTION REGULATOR

The imposed set of performances is:


- the position stationary error Estp=0 ;
- the overshooting *=0 ;
- responce time tr<tr* sec ;

The computations wiil be done based on the structure in fig. 6.2 following to determine the
expression.
The structure in fig. 6.2

B(s) e46
-sm
PROIECT I.R.A.

Nume student Grupa

i0 im

will satisfy the set of performances and more, for HR2(s)=1, we have:
(6.5) HR1(s)=B(s)1/Hf`(s)=
We try to satisfy this demands if:
(6.6) B(s)=1/TBs(Tmins+1)
In which:
Tmin=min[(2AP+TTM*);(T1*)]=
and TB will have to be determined.
The overshooting =0, requires real poles in B(0)
So let the following be:
(6.8) Bo(s)=1/(Trs+1)2
from which we have that:
(6.8') Bo(s)=1/(TBTmins2+TBs+1)
By identification we have that:
Tr=
TB=
so:
(6.6') B(s)=
and we obtain:

(6.5') HR1={T1*/4[(2ap+TTM*)s+1](2ap+TTM*)s}(T1*s+1)/T1*s =
So we will need a PI regulator with the parameters:

VR= i=T1*=

47
PROIECT I.R.A.

Nume student Grupa

so HR=

As H(s)=HR1(s).Hf'(s) so for HR2=1 we have a H(s)=B(s) and the structure is the one in fig 6.3

IQ0  HR1 Hf(s) iQ

 e-sm H(s)

So:
(6.13) Ho(s) = e-sm/(2Tmins+1)2 =
The imposed testing reffers to set 6.3.
Ho(s) obtained has no overshooting because a double real pole, so ve have an aperiodic
transitory regime
In the case of the double poles we have:
tr’=6Tmin=
If we take into consideration the delay time, then the transitory regime of the characteristic
responce is: tr=6Tmin+m=

6.3. The analysis of the results


We will compare the characteristic responces obtained for the same fixed part
Hf=
using a simple PI regulator, computed by imposing a .>600 phase margin

48
PROIECT I.R.A.

Nume student Grupa

For the same i0(s)=1/s step we will compute iQ(t) in the case of the regulator computed with the
prediction method and iQ(t), coresponding to the H0(s) structure. The simulation will be done in the
discret domain because of the inclusion of the delay time.

49
PROIECT I.R.A.

Nume student Grupa

Conlusions:

50

You might also like