7 Dynamics Tutorial Ans
7 Dynamics Tutorial Ans
7 Dynamics Tutorial Ans
1) What is the difference between lumped parameter and distributed parameter dynamical
systems? Give an example of each.
Answer
Lumped Parameter Systems
Described by ODEs
Described by PDEs
d 2 y (t )
dy (t )
+5
+ y (t ) = 2u (t )
2
dt
dt
a) What is the Laplace transfer function between the output y(t) and the input u(t)?
b) What is the time domain solution when the input is a unit step?
Answer
a) 10s 2 + 5s + 1 Y ( s ) = 2U ( s )
b) From (a) Y ( s ) =
2
U ( s)
10s + 5s + 1
2
1
10 s + 5s + 1 s
A quick inspection of the transfer functions denominator should reveal that the system has
complex poles and therefore can be written in the standard form:
2n
. From the Laplace Transform table provided, the time domain solution will be:
s 2 + 2n s + n2
2
exp( t )
2
1 1
n
y (t ) = 2 1
Sin n 1 .t + ; = tan
1 2
Compare terms to get the values of the natural frequency and damping factor:
1
5 / 10
0 .5
n =
= 0.316 and =
=
= 0.791 and substitute into above solution.
10
2 n 2 0.316
3) Using an appropriate diagram, highlight the major features of the response of an oscillatory
system to a unit step input.
Answer
A typical response of an oscillatory system is given in the diagram below:
d) dead-time
b) inverse response
e) process poles
c) time constant
f) decay ratio
Answer:
a) Damping Factor
The damping factor determines the degree of damping in a potentially oscillatory system.
With a second order system, is represented by the symbol in the following transfer
function:
n2
G ( s) = 2
s + 2 n s + 2n
A plot of the output of this system to a unit step input for various values of shows the
impact it has on output behaviour.
=0.5
=1
=2
b) Inverse Response
An inverse response is one where the output first moves in one direction before settling out to
an equilibrium at another. It is caused by opposing dynamics, and systems with inverse
responses have zeros that are positive real. An example of an inverse response is shown
below
c) Time constant
The time-constant of a dynamical process determines the speed with which the system will
respond to an input. A system with a smaller time constant will react faster than one with a
larger time constant, as shown in the following diagram
1
1
1 <2
d) Dead-time
The dead-time or time-delay is a measure of the time inertia of a dynamical system. It is
defined as the time period that elapses between the introduction of an input and the time a
change occurs in the output. A pure time-delay does not alter the magnitude or shape of the
input. The figure below compares the outputs of a system with and without a time-delay.
without delay
with delay
time-delay
e) Process poles
Process poles are the roots of the denominator of the process transfer function. They
determine whether the output of the system will oscillate, and whether the system is stable.
The presence of complex poles will indicate that the system is an oscillatory one, while poles
with positive real parts indicates that the system is unstable.
f) Decay Ratio
The decay ratio is a measure of how quickly oscillations will die out and is calculated as the
ratio (C/A) where A is the magnitude of the first overshoot and C is the magnitude of the
second overshoot as shown in the diagram below.
5) Sketch the responses of the following systems to a unit step change in input:
a)
1
1 + 10s
b)
2
2
s 2s 2
2 e 5 s
c)
1 + 15s
d)
1
2s
e)
1 s
(1 + 2s)(1 + 3s))
f)
1
s + 0.7 s + 1
2
2
s 2s 2
(unstable - exponential)
2
2 e 5 s
1 + 15s
(first-order plus time
delay)
1
(ramping)
2s
1 s
(1 + 2s)(1 + 3s))
(inverse response)
1
(oscillatory)
s + 0.7 s + 1
2
R(s)
(1 + s)(1 + 3s) )
Y(s)
a) Determine the transfer function between Y(s) and R(s), where Y(s) = L{y(t)} and
R(s) = L{r(t)}
b) Find the value of K that will give the fastest non-oscillatory response in Y(s) to a step change
in R(s)?
c) What is the corresponding ODE that relates r(t) to y(t)?
Answer:
a)
Y ( s)
K
= 2
R( s) 3s + 4s + (1 + K )
b) For fastest non oscillatory response, poles must be real and equal. Therefore find the value of
K that sets 16-4*3*(1+K) = 0, i.e. K=1/3.
c) 3
d 2 y( t )
dy(t )
+4
+ (1 + K ) y(t ) = Kr (t )
2
dt
dt
7) Step response curves are often used to develop approximate first-order plus time-delay
transfer function models of systems. On having determined the gain and time-delay values,
the time constant is estimated as the time taken for the response to reach 63.2% of the final
change in output. What is the rationale for this?
Answwer
Y ( s)
K
=
U ( s ) 1 + s
The time domain solution depends on the form of the input. If the the input is a unit step change
in u(t), then:
L{u(t)} = L{1} = U(s) = 1/s
1
(1 / )
K
= K
(1 + s ) s
s (1 / ) + s
or
Y (t ) = K 1 e t / + Yss
Thus time tends to infinity, the exponential term will decay to zero and hence y(t) will tend
towards the gain of the process. Further, at time t = , y(t) is:
y( t ) = K 1 e 1 = 0. 632 K
The time constant, , is therefore the time taken for the process to reach 63.2% of its final
change in value.
8) What is the transfer function between the input U(s) and the output Y(s) of the following
system:
1
1 + 5s
+
-
U(s)
1
1+ s
Y(s)
+
1
1 + 5s
(1 3s )
Y (s)
=
U ( s ) (1 + 5s )(1 + s )
9) Estimate the parameters of a 1st order-plus-time-delay model for the system that has the
following step response.
15.0
14.0
Output
13.0
12.0
11.0
10.0
9.0
8.0
0
50
100
Time (secs)
10
150
200
7.5
7.0
Input
6.5
6.0
5.5
5.0
4.5
4.0
0
50
100
150
200
Time (secs)
Answer
Sketch a smooth curve to approximate the noisy output response, then apply the process reaction
curve method to the smoothed response, the estimated transfer function should be:
Y ( s ) 2 exp(20s )
=
U (s)
1 + 20s
+ + C A (t ) = C A ,0 (t )
dt
11
3
2.5
2
1.5
1
0.5
0
20
40
60
80
100
120
Time (secs)
From the plot, the gain of the process is approximately 2.5, while the time constant is
approximately 20 secs.
Comparing this to the above ODE, we have
1
= 20 and
= 2.5 , ie. 2 equations with 2 unknowns, which can be solved to give
1 +
1 +
= 8 and = -0.075
11)
The schematic on the left shows a header tank.
Qin and Qout are volumetric flows, and h is the
level of liquid in the tank.
Qin
Qout
12
The schematic does not provide sufficient information for a mass balance. The flows are
volumetric flows. So, we need a density term () to convert from volume units to mass units. We
also need to know the cross-sectional area (A) of the tank, so that we can determine the volume
holdup in the tank, and hence the mass holdup.
Given these information, the mass balance of the tank can be written as:
d (Ah)
= Qin Qout
dt
Since there is no heating effects, density can be assumed constant. Also, since the tank is
cylindrical, it has constant cross-sectional area. Therefore, the ODE becomes:
Adh
Adh
= Qin Qout and hence
= Qin Qout
dt
dt
The above equation is correct as a mass balance, but is not really in the right form for solution.
Note that the flow out, Qout , is determined by the pressure exerted by the liquid, and is given by:
Qout = k h
12)
qi
qo
qi and qo are volumetric flow rates. The cylindrical tank is open to atmosphere and the outlet
flow rate can be assumed to be proportional to liquid level.
a)
Develop a dynamic mass balance for the system, focusing on the relationship between the
level, h, and input flowrate, qi.
b)
Determine the Laplace transfer function relating changes in the level, h, to changes in input
flowrate, qi.
13
c)
From the transfer function determined in (b), calculate the final value of the level, h, when
the input flow rate undergoes a unit step-change in magnitude.
d)
Find the time domain solution of the transfer function determined in (b).
Answer
a) Given these information, the mass balance of the tank can be written as:
d (Ah)
= qi qo where is the density of the liquid and A is the cross sectional area
dt
of the tank.
Since there is no heating effects, density can be assumed constant. Also, since the tank is
cylindrical, it has constant cross-sectional area. Therefore, the ODE becomes:
Adh
= qi qo
dt
The flow out, q o , is determined by the pressure exerted by the liquid, and can be approximated
linearly by:
qo = h / R where R is the resistance to flow
dh
+ qo = Rqi
dt
b) Taking Laplace Transforms of the ODE above, the resulting transfer function between level
H (s)
R
and flowrate in is:
=
Qo ( s ) 1 + ARs
c) The final value of level to a unit step change in flowrate in is R.
d) The time domain solution to the transfer function in (b) when the input is a unit step is:
h(t ) = R[1 exp( t / AR )]
14
13)
F0,1
F2
F0,2
TANK 1
TANK 2
O1
O2
TANK 3
O3
The feed, at a flowrate of F0, is split into 2 equal streams and fed to two separate tanks. A
substance is added to tank 1 at a flowrate of F1, while another substance is added to tank 2 at a
flowrate of F2. The mixtures from both tanks flow into tank 3, with flowrates O1 and O2
respectively. Further mixing takes place in tank 3 and the product leaves tank 3 with a flowrate
of O3.
Assuming that
-
the flow out of a tank is proportional to the level of liquid in that tank
Answer
dh1 (t )
= F1 (t ) + Fo ,1 (t ) O1 (t )
dt
15
A2
dh2 (t )
= F2 (t ) + Fo , 2 (t ) O2 (t )
dt
A3
dh3 (t )
= O1 (t ) + O2 (t ) O3 (t )
dt
Where Ai and hi are the cross-sectional areas and the liquid levels of tanks 1 to 3.
b)
If F1=F0,1 and F2= F0,2 , then the material balances for tanks 1 and 2 become (because F0 splits
equally to F0,1 and F0,2) :
A1
dh1 (t )
= Fo (t ) O1 (t )
dt
A2
dh2 (t )
= Fo (t ) O2 (t )
dt
Taking Laplace Transforms of all three ODEs, and assuming the using of deviation variables:
A1 sH 1 ( s ) = Fo ( s ) O1 ( s )
A2 sH 2 ( s ) = Fo ( s ) O2 ( s )
A3 H 3 ( s ) = O1 ( s ) + O2 ( s ) O3 ( s )
[1 + A1 R1 s]O1 ( s) = Fo ( s)
Similarly, for tank 2
[1 + A2 R2 s]O2 ( s) = Fo ( s)
While for tank 3
1
1
+
1 + A1 R1 s 1 + A2 R2 s
[1 + A3 R3 s]H 3 ( s) = R3 Fo ( s)
Thus the transfer function between the level in tank 3 and Fo is:
H 3 ( s)
1
1
1
= R3
+
Fo ( s)
1 + A3 R3 s 1 + A1 R1 s 1 + A2 R2 s
16
c) Because all the poles of the transfer are real, the level in tank 3 will not oscillate.
14)
The diagram on the left show a heated stirred tank.
The heating mediums flow rate is S [kgmin-1]. The
temperature of the bulk liquid it T [degC]. Feed
liquid enters the system at Fi [kgmin-1] and a
temperature of Ti [degC]. This flow is used to
maintain the liquid level in the tank. The heated
liquid leaves the tank with a flowrate of at
Fo [kgmin-1] and a temperature of To [degC].
a) Assuming perfect level control, develop a dynamic energy balance for the system [40%]
b) From (a), obtain the transfer function that describes the effects of changes in heating
medium flowrate on outlet stream temperature. [10 %]
c) What modifications will you have make to the balance equation(s) if the level is not
controlled perfectly [50%]
Answer
a) Perfect level control implies that liquid level is constant, which means that:
F = Fi = Fo .
Assume that heat is transferred to the system purely by condensing steam and that the heat
capacity of the liquid is constant. Assume also that the tank is well stirred, so that the
temperature of the output stream is equal to the temperature of the liquid in the tank.
Using the following general dynamic heat balance equation:
Rate of Energy Accumulation = Rate of Energy Input - Rate of Energy Consumption
Rate of Energy Input = FC p (Ti Ta ) + S
17
dTo
where M is the mass of liquid in the tank.
dt
Thus the dynamic model of the stirred tank heating system is:
MC p
dT
= FC p (Ti Ta ) + S FC p (To Ta )
dt
Simplification yields
MC p
dTo
= FC p (Ti To ) + S
dt
b)
Since the contribution of inlet temperature is not considered, and that Laplace variables have
zero initial values, the required transfer function is obtained from:
MC p
dTo
+ FC p To = S or
dt
M dTo
S
+ To =
F dt
Cp
( / C p )
To ( s)
=
S ( s) 1 + ( M / F ) s
c) If the controller does not provide perfect level control, then a mass balance will have to be
written. That is,
dh
= Fi k h / A
dt
where h is the level, and A is the cross sectional area of the tank. Since the flow in is being
manipulated by the controller (proportional controller say), we will need to describe this as well.
Fi = k c (hs h) + Fi ,0
where Fi ,0 is the value of initial input flow rate; hs is the desired level and kc is the gain of the
controller. The energy balance will then have to be modified to:
MC p
dTo
= Fi C p (Ti Ta ) + S Fo C p (To Ta ) , with M = A.h.
dt
MC p
dTo
= Fi C p ,i (Ti Ta ) + Ws Fo C p ,o (To Ta )
dt
18
15)
a) Determine the Laplace transfer function between the output y(t) and the input u(t)
b) Sketch the response of the system when the input is a step of magnitude 2.
c) Determine the time domain solution of the ODE when the input is a step of magnitude 2.
Answer
a)
Y ( s ) 2 exp(5s )
=
U ( s)
1 + 10s
b)
4
3.5
3
Output
2.5
2
1.5
1
0.5
0
10
20
30
Time (units)
40
50
60
16)
The concentration of component A (xA) in the top product stream of a distillation column
is related to changes in the reflux flowrate (Re ) according to the ODE:
20
dx A
+ x A = 2. Re
dt
19
x A ( s)
2
=
Re( s) 1 + 20s
2
1
. , which from tables will
1 + 20s s
c) The time constant of the system is 20 mins. Therefore the change in xA due to a -1 change in
Re after 20 mins. would be -1*2*0.632 = -1.264. Hence the final value of xA at this point in
time will be (95-1.264)=93.736 weight percent
20