5.

a) The overall gain of G is G s =0

K1 K2
= ⋅ = K1 K 2
τ1 × 0 + 1 τ 2 × 0 + 1

b) If the equivalent time constant is equal to τ1 + τ2 = 5 + 3 = 8, then

y(t = 8)/KM = 0.632 for a first-order process.

5e −8 / 5 − 3e −8 / 3
y(t = 8)/KM = 1 − = 0.599 ≠ 0.632
5−3

Therefore, the equivalent time constant is not equal to τ1 + τ2

c) The roots of the denominator of G are

-1/τ1 and -1/τ2

which are negative real numbers. Therefore the process transfer function
G cannot exhibit oscillations when the input is a step function.
5.12

a) 1y1 + Ky1 + 4 y = x

Assuming y(0) = y1 (0) = 0

Y (s) 1 0.25
= 2 =
X ( s ) s + Ks + 4 0.25s + 0.25 Ks + 1
2

b) Characteristic equation is

s2 + Ks + 4 = 0

− K ± K 2 − 16
The roots are s =
2

-10 ≤ K < -4 Roots : positive real, distinct

Response : A + B e t / τ1 + C et / τ 2

K = -4 Roots : positive real, repeated

Response : A + Bet/τ + C et/τ

-4 < K < 0 Roots: complex with positive real part.

t t
Response: A + eζt/τ (B cos 1 − ζ 2 + C sin 1 − ζ 2 )
τ τ

K=0 Roots: imaginary, zero real part.

Response: A + B cos t/τ + C sin t/τ

0<K<4 Roots: complex with negative real part.

t t
Response: A + e-ζt/τ (B cos 1 − ζ 2 + C sin 1 − ζ 2 )
τ τ

K=4 Roots: negative real, repeated.

Response: A + Be-t/τ + C t e-t/τ

4 < K ≤ 10 Roots: negative real, distinct

Response: A + B e −t / τ1 + C e −t / τ 2

Response will converge in region 0 < K ≤ 10, and will not converge in
region –10 ≤ K ≤ 0
5.13

a) The solution of a critically-damped second-order process to a step change

of magnitude M is given by Eq. 5-50 in text.

  t 
y(t) = KM 1 − 1 + e −t / τ 
  τ 

Rearranging

y  t
= 1 − 1 + e − t / τ
KM  τ

 t  −t / τ y
1 + e = 1−
 τ KM

When y/KM = 0.95, the response is 0.05 KM below the steady-state value.

KM

0.95KM

0 ts time

 t s  −t / τ
1 + e = 1 − 0.95 = 0.05
 τ

 t  t
ln1 + s  − s = ln(0.05) = −3.00
 τ τ

 t  t
Let E = ln1 + s  − s + 3
 τ τ
 ts
and find value of that makes E ≈ 0 by trial-and-error.
τ

ts/τ E
4 0.6094
5 -0.2082
4.5 0.2047
4.75 -0.0008

∴ a value of t = 4.75τ is ts, the settling time.

Ka a a a a4
b) Y(s) = = 1 + 22 + 3 +
s (τs + 1)
2 2
s s τs + 1 (τs + 1) 2

We know that the a3 and a4 terms are exponentials that go to zero for large
values of time, leaving a linear response.

Ka
a2 = lim = Ka
s →0 (τs + 1) 2

Ka
Define Q(s) =
(τs + 1) 2

dQ − 2 Kaτ
=
ds (τs + 1) 3

1  − 2 Kaτ 
Then a1 = lim  
1! s →0  (τs + 1) 3 

(from Eq. 3-62)

a1 = − 2 Kaτ

∴ the long-time response (after transients have died out) is

y 2 (t ) = Kat − 2 Kaτ = Ka(t − 2τ)

= a (t − 2τ) for K = 1

and we see that the output lags the input by a time equal to 2τ.
 2τ

y x=at yl =a(t-2τ)

0 actual response time

5.21

Assumptions: 1) Perfectly mixed reactor

2) Constant fluid properties and heat of reaction

a) Component balance for A,

dc A
V = q (c A i − c A ) − Vk (T )c A (1)
dt

Energy balance for the tank,

dT
ρVC = ρqC (Ti − T ) + (− ∆H R )Vk (T )c A (2)
dt

Since a transfer function with respect to cAi is desired, assume the other
inputs, namely q and Ti, are constant. Linearize (1) and (2) and note that

dc A dc ′A dT dT ′
= , = ,
dt dt dt dt

dc ′A 20000
V = qc ′A i − (q + Vk (T ))c ′A − Vc A k (T ) T′ (3)
dt T2

dT ′  20000 
ρVC = − ρqC + ∆H RVc A k (T ) T ′ − ∆H RVk (T )c ′A (4)
dt  T2 

Taking Laplace transforms and rearranging

[Vs + q + Vk (T )]C ′ ( s) = qC ′ (s) − Vc

A Ai A k (T )
20000
T ′( s ) (5)
T2
 20000 
ρVCs + ρqC − (−∆H R )Vc A k (T ) T 2  T ′( s ) = (−∆H R )Vk (T )C A′ ( s ) (6)

Substituting C ′A (s ) from Eq. 5 into Eq. 6 and rearranging,

T ′( s ) (−∆H R )Vk (T )q
=
C A′i ( s )  20000  20000
Vs + q + Vk (T )  ρVCs + ρqC − (−∆H R )Vc A k (T )  + (−∆H R )V 2 c A k 2 (T )
 
2
T T2

(7)
c A is obtained from Eq. 1 at steady state,
 qc Ai
cA = = 0.001155 lb mol/cu.ft.
q + Vk (T )

Substituting the numerical values of T , ρ, C, –∆HR, q, V, c A into Eq. 7

and simplifying,

T ′( s ) 11.38
=
C ′A i ( s ) (0.0722s + 1)(50s + 1)

For step response, C ′Ai ( s ) = 1 / s

11.38
T ′( s ) =
s (0.0722s + 1)(50s + 1)

 1 
T ′(t ) = 11.381 + (0.0722e −t / 0.0722 − 50e −t / 50 ) 
 (50 − 0.0722) 

A first-order approximation of the transfer function is

T ′( s ) 11.38
=
C ′A i ( s ) 50s + 1

For step response, T ′( s ) =

11.38
s (50s + 1)
or T ′(t ) = 11.38 1 − e −t / 50 [ ]
The two step responses are very close to each other hence the
approximation is valid.
12

10

8
T'(t)

Using transfer function

Using first-order approximation
0
0 20 40 60 80 100 120 140 160 180 200
time

Fig S5.21. Step responses for the 2nd order t.f and 1st order approx.
5.22

(τas+1)Y1(s) = K1U1(s) + Kb Y2(s) (1)

(τbs+1)Y2(s) = K2U2(s) + Y1(s) (2)

a) Since the only transfer functions requested involve U1(s), we can let U2(s)
be zero. Then, substituting for Y1(s) from (2)

Y1(s) = (τbs+1)Y2(s) (3)

(τas+1)(τbs+1)Y2(s) =K1U1(s) + KbY2(s) (4)

Rearranging (4)

[(τas+1)(τbs+1) − Kb]Y2(s) =K1U1(s)

Y2 ( s ) K1
∴ = (5)
U 1 ( s ) (τ a s + 1)(τ b s + 1) − K b
Also, since

Y1 ( s )
= τb s + 1 (6)
Y2 ( s )

From (5) and (6)

Y1 ( s ) Y2 ( s ) Y1 ( s ) K 1 (τ b s + 1)
= × = (7)
U 1 ( s ) U 1 ( s ) Y2 ( s ) (τ a s + 1)(τ b s + 1) − K b

b) The gain is the change in y1(or y2) for a unit step change in u1. Using the
FVT with U1(s) = 1/s.

 K1 1 K1
y 2 (t → ∞) = lim  s =
s →0
 (τ a s + 1)(τ b s + 1) − K b s  1 − Kb

This is the gain of TF Y2(s)/U1(s).

Alternatively,

 Y (s)   K1  K1
K = lim  2  = lim   =
 1  s →0  (τ a s + 1)(τ b s + 1) − K b  1 − K b
s →0 U ( s )

For Y1(s)/U1(s)
  K 1 (τ b s + 1) 1 K1
y1 (t → ∞) = lim  s =
s →0
 (τ a s + 1)(τ b s + 1) − K b s  1 − Kb
K1
In other words, the gain of each transfer function is
1 − Kb
Y2 ( s ) K1
c) = (5)
U 1 ( s ) (τ a s + 1)(τ b s + 1) − K b

Second-order process but the denominator is not in standard form, i.e.,

τ2s2+2ζτs+1

Put it in that form

Y2 ( s ) K1
= (8)
U 1 ( s) τ a τ b s + (τ a + τ b ) s + 1 − K b
2

Dividing through by 1- Kb

Y2 ( s ) K 1 /(1 − K b )
= (9)
U 1 (s) τ a τ b 2 (τ a + τ b )
s + s +1
1 − Kb 1 − Kb

Now we see that the gain K = K1/(1-Kb), as before

τa τ b τa τb
τ2 = τ= (10)
1 − Kb 1 − Kb

τa + τb
2ζτ = , then
1 − Kb

1 τa + τb 1 − K b  1 τa + τb  1
ζ= =  (11)
2 1 − Kb τa τb  2 τ a τ b  1 − Kb

Investigating Eq. 11 we see that the quantity in brackets is the same as ζ

for an overdamped 2nd-order system (ζOD) [ from Eq. 5-43 in text].

ζ OD
ζ= (12)
1 − Kb
1 τa + τ b
where ζ OD =
2 τa + τb
 Since ζOD>1,

ζ>1, for all 0 < Kb < 1.

In other words, since the quantity in brackets is the value of ζ for an

overdamped system (i.e. for τa ≠ τb is >1) and 1 − K b <1 for any positive
Kb, we can say that this process will be more overdamped (larger ζ) if Kb
is positive and <1.

For negative Kb we can find the value of Kb that makes ζ = 1, i.e., yields a
critically-damped 2nd-order system.

ζ OD
ζ =1= (13)
1 − K b1
ζ
2

or 1 = OD
1 − K b1

1 – Kb1 = ζOD2
Kb1 = 1 − ζOD2 (14)

where
Kb1 < 0 is the value of Kb that yields a critically-damped process.

Summarizing, the system is overdamped for 1 − ζOD2 < Kb < 1.

Regarding the integrator form, note that

Y2 ( s ) K1
= (8)
U 1 ( s) τ a τ b s + (τ a + τ b ) s + 1 − K b
2

For Kb = 1

Y2 ( s ) K1 K1
= =
U 1 ( s ) τ a τ b s + (τ a + τ b ) s s[τ a τ b s + (τ a + τ b )]
2

K 1 /(τ a + τ b )
=
 τ τ 
s  a b s + 1
 τa + τ b 
K 1′
which has the form = ( s indicates presence of integrator)
s (τ′s + 1)
d) Return to Eq. 8

System A:

Y2 ( s ) K1 2K 1
= = 2 1 = 2
U 1 ( s ) (2)(1) s + (2 + 1) s + 1 − 0.5 4s + 6s + 1 4s + 6s + 1
2

τ2 = 4 → τ=2
2ζτ = 6 → ζ = 1.5

System B:
1 1
For system = 2
(2 s + 1)( s + 1) 2 s + 3s + 1

τ22 = 2 → τ2 = 2
3 1.5
2ζ2τ2 = 3 → ζ2 = = ≈ 1.05
2 2 2

Since system A has larger τ (2 vs. 2 ) and larger ζ (1.5 vs 1.05), it will
respond slower. These results correspond to our earlier analysis.