Laplace Transforms

Chapter 70
70.1 Definition
70.2 Use of Laplace Transforms
70.3 Partial Fractions

Laplace transformation is used extensively in con- to have been transformed in which case it is de-
trol engineering. It is a powerful technique for noted as such by putting (s) after it. Thus f(s) is the
describing and solving sets of differential equa- Laplace transform of f(t).
tions. When first encountered Laplace transforms The integration of Equation 70.1 involves the
may appear daunting although,in practice,they are Laplace operator “s” which is a complex variable
fairly easy to use. It is best to learn how to use the that is, by definition, independent of time. Because
technique first without fully understanding how the integration is carried out over the range 0 to
it works. Having mastered the technique a deeper ∞ the resultant transform is an algebraic expres-
understanding will follow. This process is analo- sion in s and is independent of time. This is best
gous to learning to drive. You don’t need to know illustrated by some examples:
how an engine works to learn to drive, but being
able to drive does give an insight into how it works. Example 1. Consider a step change in flow of mag-
For a deeper understanding the reader is referred nitude A at time t = 0 as depicted by Figure 70.1.
to classical control texts such as those by Dutton Substituting in Equation 70.1,assuming zero initial
(1997) and Wilkie (2002). flow, gives:
 ∞  −st ∞
e A
f(s) = A.e−st .dt = A. =
0 −s 0 s
70.1 Definition f(t)
The Laplace transform is defined by the following A
equation:
 ∞ t
f(s) = L{f(t)} = f(t).e−st .dt (70.1) 0
0
Fig. 70.1 A step change of magnitude A
It is normal for the variable being transformed to
be an analogue signal that is some function f(t) of
Example 2. Consider an exponential decay c =
time but it could, for example, be a function f(x)
e−at in concentration as depicted by Figure 70.2.
of distance. If the function f(t) is known explicitly,
Substituting in Equation 70.1, again assuming zero
then its transform can be found by the integra-
initial conditions, gives:
tion defined in Equation 70.1. If the function f(t)
 ∞  −(s+a)t ∞
is not known explicitly then its transform cannot e 1
be found. However, the variable may be considered c(s) = e−at .e−st .dt = =
0 −(s + a) 0 s+a
 576 70 Laplace Transforms

c(t) r(t)
1

slope m

t t
0
0
Fig. 70.2 The unit exponential decay
Fig. 70.3 A ramp signal of slope m
Example 3. Consider a ramping signal as depicted
Fortunately, it is not normally necessary to have
in Figure 70.3. Assume that the ramp has a slope of
to work out the transform of a function from first
m. Integrating by parts gives:
principles. Extensive tables of Laplace transforms
 ∞ exist.Table 70.1 lists the more commonly used ones
r(s) = mt.e−st .dt in process control. Table 70.2 lists the more im-
0
 −st ∞  ∞ −st portant properties of Laplace transforms. These
e e m properties, which are all proved in texts such as
= m. t. − m. dt = 2
−s 0 0 −s s Coughanowr (1991) and Ogata (1990),will be made
use of in subsequent chapters.

Table 70.1 Table of common Laplace transforms

Table 70.2 Important properties of Laplace transforms
f(t)t≥0 f(s)
(f(t)t<0 = 0) f(t)t≥0 f(s)
(f(t)t<0 = 0)
A
A
s
f1 (t) + f2 (t) f1 (s) + f2 (s)
1
t
s2 k.f(t) k.f(s)
n!
tn d
sn+1 f(t) s.f(s) − f(0)
dt
1  
e−at d2 d
s+a f(t) s2 .f(s) − s.f(0) − f(t)
dt 2 dt t=0
T t
e−t/T f(s)
Ts + 1 f(t)dt
1 s
1 − e−t/T 0
s.(Ts + 1)
T2 f(t − L) e−Ls .f(s)
t.e−t/T
(Ts + 1)2
ı0 (t) 1
!
sin(!t) 1 − e−Ls
s2 + ! 2 ıL (t)
s Ls
cos(!t)
s2 + ! 2 f(∞) Lim s.f(s)
! s→0
e −at
. sin(!t)
(s + a)2 + !2 f(0) Lim s.f(s)
s→∞
s+a
e −at
. cos(!t)
(s + a)2 + !2
 70.2 Use of Laplace Transforms 577

Note that in the above examples, and for all the and the transform of the variable. Thus the co-
transforms given in Tables 70.1 and 70.2, the time efficient of any term in an ODE has no effect on
t = 0 is both the instant from which the func- the transformation process.
tion exists and from which integration commences. 2. Solve the algebraic equation resulting from Step
Also, the function’s prior values are specified as be- 1 for the transformed variable f(s).
ing zero. This is consistent with the practice of us- 3. Reduce the function f(s) resulting from Step 2
ing deviation variables, as discussed in Chapter 22, into partial fractions that occur on the right
and assuming that the variable was at its normal hand side of Table 70.1. For simple functions
value prior to the period of interest. this step is trivial but for complex ones it can
be very tedious.
4. Find the inverse transform of the equation re-
sulting from Step 3 by using Table 70.1. Again,
70.2 Use of Laplace Transforms make use of the first two properties of Ta-
ble 70.2 which enable the inverse transforms to
be found one term at a time and independently
Laplace transforms are used for solving ordinary of their coefficients.
linear differential (ODE) equations with constant
coefficients of the general form:
Example 4. To illustrate this procedure reconsider
dn f(t) dn−1 f(t) the sensor used for monitoring the temperature of
an n
+ an−1 + ... a process stream, as discussed in Chapter 69 and
dt dtn−1
df(t) depicted in Figure 69.4. The sensor is a first order
+ a1 + a0 f(t) = u(t) (70.2) system and its dynamics are described by the first
dt
order ODE:
Thus Laplace transforms cannot normally be used d0
T + 0 = 1 (70.3)
for solving the following equations on the grounds dt
that they are not ordinary, linear or with constant where T = Mcp /UA as defined in Equation 69.5.
coefficients respectively: Suppose there is a step change of magnitude A
∂f(t, x) in the process stream temperature. Thus, defining
a + f(t, x) = u(t) t = 0 to be the instant when the step change occurs,
∂x
df(t) and assuming 0 to be in deviation form:
a + [f(t)]2 = u(t)
dt d0
df(t) T + 0 = A
a(t) + f(t) = u(t) dt
dt
Step 1. Transform both sides of the equation:
There is essentially a four-step procedure to solv-
ing ODEs using Laplace transforms: A
T(s0 (s) − 0 (0)) + 0 (s) =
1. Transform both sides of the ODE using the s
transforms of Table 70.1. The first two prop- Since 0 is in deviation form, 0 (0) is presumed to
erties listed in Table 70.2 are important here. be zero:
First, the transform of the sum of two or more A
Ts0 (s) + 0 (s) =
variables is the sum of the transforms of the in- s
dividual variables. Thus the terms of the ODE Step 2. Solve for 0 (s):
may be transformed one at a time. And second,
the transform of the product of a coefficient A
and a variable is the product of the coefficient 0 (s) =
s(Ts + 1)
 578 70 Laplace Transforms

Step 3. There is no need to split this into partial Note that the denominator of 0 (s) has three roots,
fractions because 0 (s) is on the right hand side of s2 , s and (Ts + 1), the s root being obscured by the
Table 70.1. s2 root. Any such “hidden” roots must be allowed
Step 4. Inverse transform: for in the expansion. If, in the above example, the
hidden s root did not exist then its coefficient b
0 (t) = A(1 − e−t/T ) (69.2) would be found to be zero.
This step response is as depicted in Figure 69.2.
Approach 1. Equating coefficients.
Example 5. Suppose that the process stream tem- Putting the right-hand-side of Equation 4 over a
perature is ramping upwards with a slope m.Define common denominator gives:
t = 0 to be the instant when the ramp starts and,
again, assume 0 to be in deviation form: m (bT + c)s2 + (aT + b)s + a
=
d0 s2 (Ts + 1) s2 (Ts + 1)
T + 0 = mt
dt The coefficients of the left and right hand numer-
Step 1. Transform both sides of the equation: ators may be equated:
m s2 0 = bT + c
T(s0 (s) − 0 (0)) + 0 (s) =
s2 s1 0 = aT + b
As before, 0 is in deviation form so 0 (0) is zero: s0 m=a
m
Ts0 (s) + 0 (s) = 2 These three equations may be solved for the three
s
unknowns,yielding: a = m,b = −mT and c = mT2 .
Step 2. Solve for 0 (s): Hence:
m
0 (s) = m m mT mT2
s2 (Ts
+ 1) = − +
s2 (Ts + 1) s2 s (Ts + 1)
Step 3. Split into partial fractions:
The equating coefficients approach is realistic for
m mT mT2 simple functions when the number of simultane-
0 (s) = 2 − +
s s (Ts + 1) ous equations to be solved is small. However, for
more complex functions the method of residues is
Step 4. Inverse transform:
more appropriate.
0 (t) = mt − mT + mTe−t/T (69.3)
Approach 2. The method of residues.
This ramp response is as depicted in Figure 69.3.
This method involves isolating the coefficients in
turn by multiplying throughout by the roots as ap-
propriate. The value of the coefficient is the resid-
70.3 Partial Fractions ual when a value for s is substituted such that all
There are two approaches to this. In both cases all the other numerator terms become zero. This is
possible partial fractions must be identified by in- best illustrated by means of the same example.
spection. Consider the function 0 (s) in Example 5 Multiplying Equation 70.4 throughout by s2
which may be expanded as follows: gives:

m a b c m cs2
= + + (70.4) = a + bs + (70.5)
s2 (Ts + 1) s2 s (Ts + 1) (Ts + 1) (Ts + 1)
 70.3 Partial Fractions 579

Since this equation is true for all values of s, it must m m mT mT2

= 2 − +
be true for any particular value. Choose s = 0 and s2 (Ts
+ 1) s s (Ts + 1)
substitute gives a = m.
Now multiply Equation 70.4 throughout by
(Ts + 1):
Option 2. Since s is a variable and all the other
m a(Ts + 1) b(Ts + 1) parameters are constants, Equation 70.5 may be
= + +c
s2 s2 s differentiated with respect to s:
Choose s = −1/T and substitute gives c = mT2 .
The value for b cannot be found in the same
−mT 2cs cTs2
way. Multiplying Equation 70.4 throughout by s = b + −
does not enable b to be isolated because of the (Ts + 1)2 (Ts + 1) (Ts + 1)2
s2 in the denominator. In such circumstances there
are two options. This has now isolated b which can be found by
substituting any convenient value for s, say s = 0,
Option 1. Substitute the values found for a and c which gives b = −mT.
into Equation 70.4 as follows: It is important to understand the processes in-
volved and to be able to handle them manually for
m m b mT2
= + + simple functions.However,for more complex func-
s2 (Ts + 1) s2 s (Ts + 1) tions, any modern maths or control systems design
Again this equation is true for all values of s so it package will have routines for finding the roots and
must be true for any particular value. Choose s = 1 partial fractions of polynomial functions, which
say and substitute gives b = −mT whence: obviously simplifies the process of solving ODEs.