Department of Chemical Engineering
Indian Institute of Technology Guwahati
CL 611:Advanced Process Control
Solution key of Midsem September 23, 2015
1. Digital word 10111011
GR (z)
2. (a)
1 + GH (z)
3. sampling period 0.5
4. Final value 1
z + 2z 2 + z 3
5. (a) physically realizable D (z) =
1 + 2z − 3z 2 + 4z 3 + z 4
6. (b) unstable
1
7. (a) ringing pole G (z) =
1 − 0.81z −2
8. PID control law 60 − 65z −1 + 30z −2
1 z −3
9. (c) deadbeat controller D (z) = ×
HGp (z) 1 − z −3
10. (d) no of zeroes
11. The first order hold in time domain is
m(nT ) − m {(n − 1) T }
m(t) = m(nT ) + (t − nT ) (1)
T
Now for one time step delay,
m {(n − 1) T } ≡ e−sT m(nT ) (2)
Page 1 of 4 Please go on to the next page. . .
Solution Key of Midsem CL 611:Advanced Process Control September 23, 2015
R (n+1)T RT
Consider τ = t − nT , then dt ≡ dτ and nT
≡ 0
, hence
Z T
m(s) = m(t)e−sτ dτ
0
Z T
m(nT ) − m {(n − 1) T }
= m(nT ) + (t − nT ) e−sτ dτ
0 T
Z T
1 − e−sT
= m(nT ) 1+ τ e−sτ dτ
0 T
Z T Z T
1 − e−sT
−sτ −sτ
= m(nT ) e dτ + τ e dτ
0 T 0
" ( −sτ T Z T −sτ )#
e−sT − 1 1 − e−sT
τe e
= m(nT ) + − dτ
−s T −s 0 0 −s
1 − e−sT 1 − e−sT −T e−sT 1 e−sT − 1
= m(nT ) + +
s T s s −s
−sT −sT
1−e 1−e 1 −sT
T −sT
= m(nT ) + 1−e − e
s T s2 s
" ( )#
1 − e−sT
1 − e−sT
s T −sT
= m(nT ) 1+ − e
s T s2 s
" ( )#
1 − e−sT
1 − e−sT
= m(nT ) 1+ − e−sT
s sT
1 − e−sT sT + 1 − e−sT − sT e−sT
= m(nT )
s sT
−sT
(1 + sT ) − e−sT (1 + sT )
1−e
= m(nT )
s sT
−sT 2
1−e 1 + sT
= m(nT ) (3)
s T
12. The process
2e−4s
Gp (s) =
20s + 1
The first order hold is
2
1 − e−T s
1 + sT
H (s) =
T s
With sampling interval T = 2, the delay lT = 4 = 2T or l = 2
Page 2 of 4 Please go on to the next page. . .
Solution Key of Midsem CL 611:Advanced Process Control September 23, 2015
The discretization of process
HGp (z) = Z {H (s) Gp (s)}
( 2 )
1 − e−T s 2e−2T s
1 + sT
=Z ×
T s 20s + 1
2 −2 −1 2
1 + 2s
= z 1−z Z
T s2 (20s + 1)
−2 −1 2
1 18 360
=z 1−z Z − +
s2 s 20s + 1
−1
−2 −1 2
2z 18 18
=z 1−z − +
(1 − z −1 )2 1 − z −1 1 − e−0.1 z −1
( )
−1 −0.1 −1 −1 −0.1 −1 −1 2
2 z (1 − e z ) − 9 (1 − z ) (1 − e z ) + 9 (1 − z )
= 2z −2 1 − z −1
(1 − z −1 )2 (1 − e−0.1 z −1 )
= ···
1 − z −1
−3
= −14.2z
1 − 0.9z −1
The dead-beat controller:
u (z) 1 z −(l+1)
= D (z) = ×
ε (z) HGp (z) 1 − z −(l+1)
1 − 0.95z −1 z −3
= ×
−14.2z −3 (1 − z −1 ) 1 − z −3
−0.07 + 0.06z −1
=
1 − z −1 − z −3 + z −4
uk = uk−1 + uk−3 − uk−4 − 0.07εk + 0.06εk−1
E
13. The only nonlinear portion of the model is e− RT cA which can be linearized around
{cAs , Ts } using Taylor’s series expantion upto the first order term in the following
manner
E
− RT E
− RT E
− RT E
− RT E 1
e cA = e s cA s + e s (cA − cAs ) + cAs e s − − 2 (T − Ts )
R Ts
= c0 + c1 (cA − cAs ) + c2 (T − Ts )
Page 3 of 4 Answer 13 continues on the next page. . .
Solution Key of Midsem CL 611:Advanced Process Control September 23, 2015
where c0 , c1 and c2 are relevant constants. Hence the equations can be written as
dcA 1 E
= (cAi − cA ) − k0 e− RT cA
dt τ
1
= (cAi − cA ) − k0 [c0 + c1 (cA − cAs ) + c2 (T − Ts )] (4)
τ
dT 1 E
= (Ti − T ) + Jk0 e− RT cA − Q
dt τ
1
= (Ti − T ) + Jk0 [c0 + c1 (cA − cAs ) + c2 (T − Ts )] − Q (5)
τ
At steady state i.e. at cA = cAs , T = Ts and Q = Qs , the eqs. (4) and (5) can be
written as
dcAs 1
= (cAi − cAs ) − k0 [c0 + c1 (cAs − cAs ) + c2 (Ts − Ts )] (6)
dt τ
dTs 1
= (Ti − Ts ) + Jk0 [c0 + c1 (cAs − cAs ) + c2 (Ts − Ts )] − Qs (7)
dt τ
Now, eq.(4)−eq.(6) and eq.(5)−eq.(7) yield
d (cA − cAs ) 1
= − (cA − cAs ) − k0 [c1 (cA − cAs ) + c2 (T − Ts )]
dt τ
d (T − Ts ) 1
= − (T − Ts ) + Jk0 [c1 (cA − cAs ) + c2 (T − Ts )] − (Q − Qs )
dt τ
or with deviation variables cA = (cA − cAs ) and T = (T − Ts ),
dcA 1 1
= − cA − k0 c1 cA + c2 T = − − k0 c1 cA + (−k0 c2 ) T
dt τ τ
dT 1 1
= − T + Jk0 c1 cA + c2 T − Q = (Jk0 c1 ) cA + − + Jk0 c2 T − Q
dt τ τ
Hence
dcA
− τ1 − k0 c1
dt (−k0 c2 ) cA 0
= + Q
dT
dt
(Jk0 c1 ) − τ1 + Jk0 c2 T −1
cA
cA = 1 0
T
The state space model is represented as above with
" E E #
− τ1 − k0 e− RTs E
− RT c k
2 As 0 e − RT
s
A= E
− RT
s
E
Jk0 e s − τ + RT 2 cAs Jk0 e− RTs
1 E
s
0
B=
−1
C= 1 0
D=0
Page 4 of 4 End of Solution Key
