Chap 05

Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

On-Off Control

Chapter 28
28.1 On-Off Cycling
28.2 On and Off Curves
28.3 Lag Effects
28.4 Worked Example
28.5 Comments
28.6 Nomenclature

On-off control, sometimes referred to as bang- 28.1 On-Off Cycling


bang control, is conceptually the same thing as
proportional control, as described in Chapter 22, It is seldom that a proportional controller per se is
with a high controller gain. It is characterised by used for on-off control. Much more typical is the
use of amplifiers and relays in simple thermostats.
very small, but finite, errors causing the controller
output to switch between maximum and minimum On-off control is surprisingly common for simple,
output according to the sign of the error. The re- non-critical applications. There is more to it than
sponse of a forward acting on-off controller to a meets the eye though.
sawtooth error is depicted in Figure 28.1. Consider a tank containing a liquid whose tem-
perature is thermostatically controlled as depicted
e in Figure 28.2.

Fig. 28.2 Tank with thermostat


Fig. 28.1 Response of on-off controller to sawtooth error
Depending on the temperature of the liquid in the
tank, the power supply is either connected to or
disconnected from the heating element. This re-
sults in the temperature cycling about the set point
R within a narrow band , as depicted in Fig-
192 28 On-Off Control

θ’

heating curve
cooling curve

θP ton toff

tc
Δθ
OFF
ON

Fig. 28.3 Heating and cooling curves for tank contents

ure 28.3. Since  is small, the sections of the heat- W


If t → ∞ then P →  = . Defining the time
ing and cooling curves may be approximated by U.A
M.cp
straight lines. constant TP = yields:
U.A
P =  . 1 − e−t/TP
28.2 On and Off Curves The slope of the heating curve is thus:
Consider the heating curve. If the heater is dP  −t/TP  − P
= .e =
switched on, assuming that the liquid is well mixed dt TP TP
and that there are no heat losses due to evapora- Let  P be the mean value of P within the band .
tion, then an unsteady state heat balance for the Thus, within :
contents of the vessel gives: 
dP   − P
 ≈ (28.1)
dP dt on T P
M.cp = W − U.A.P
dt Now consider the cooling curve. If the heater is
where  is measured relative to ambient temper- switched off, then an unsteady state heat balance
ature. This is a first order system, of the type de- for the contents of the vessel gives:
scribed in Chapter 69, whose response is of the dP
form: M.cp = −U.A.P
⎛ −UA ⎞
dt
W ⎜ t This too is a first-order system, whose response is
P = Mcp ⎟
⎝1 − e ⎠ of the form:
UA
P =  .e−t/TP
28.3 Lag Effects 193

The slope of the cooling curve is thus: The measured temperature M lags behind the tank
temperature by the time constant TM of the mea-
dP − −t/TP −P suring element. Because of mechanical backlash
= .e =
dt TP TP in the switches, hysteresis, stiction, etc., the heater
state will not change from on to off and vice versa
So, for values of P within the band : until there exists a finite error about the set point.
 The corresponding band of measured temperature
dP  − P is called the differential gap M .
 ≈ (28.2)
dt off TP From Figure 28.4 it can be seen that:
 
dp  dp 
 = M + .TM + .TM
28.3 Lag Effects dt on dt off

A section of the sawtooth of Figure 28.3 is repro- Substituting from Equations 28.1 and 28.2 gives:
duced in Figure 28.4. TM
The controller period tC is the sum of the ton  = M + . (28.4)
TP
and toff times. Thus:
The mean tank temperature is half way across :
1 1
tC =  . +  .  M
dP  dP  TM
 P = B + = B + + .
dt on dt off 2 2 2.TP
Assume that the backlash, etc., is symmetrical
Substituting from Equations 28.1 and 28.2 gives: about the set point R :

 .TP . dP  M TM M
tC = (28.3) R = B + .TM + = B + . P +
 P.  −  P dt off 2 TP 2

TM

measured temp’r

liquor temp’r Δθ

ΔθM

TM
ton
toff

Fig. 28.4 On and off sections of sawtooth


194 28 On-Off Control

For on-off control the offset is defined to be the 28.5 Comments


difference  P − R , whence:
  Most on-off systems operate with a relatively large
TM  cycle time and deviation. A long cycle time gives
Offset =  P − R ≡ . − P (28.5)
TP 2 rise to large deviations in the controlled variable,
whereas a short cycle time may cause excessive
wear on the relays, actuators, etc. Thus there is a
trade off between cycle time and deviation.
28.4 Worked Example Equations 28.3–28.5 give the relationships be-
The temperature of the liquid in a tank is con- tween the cycle time, width of the band and its
trolled by an on-off controller. The trend diagram mean, the differential gap, set point, offset and
indicates that if the set point is 65◦ C then the on- time lags. To achieve a narrow band requires min-
time is 249 s, the off-time is 150 s and the differen- imal backlash, etc., and a fast measurement, i.e.
tial gap is 1.5◦ C. If the heater is left switched on the TM  TP .
temperature reaches 95◦ C eventually. What is the There will be no offset when  = 2  P in which
temperature variation in the tank? case:  
dP  dP  
M = 1.5◦ C, ton = 249 s and toff = 150 s. dt on
= =
dt off 2 TP
Assume an ambient temperature of 15◦ C: The on and off times will be equal when the slopes
◦ ◦ of the on and off curves are the same.
R = 65 − 15 = 50 C and  = 95 − 15 = 80 C.
From the slopes of the heating and cooling curves,
Equations 28.1 and 28.2:
28.6 Nomenclature
 − ¯P ¯P
 = .ton = .toff A effective surface area m2
TP TP cP specific heat kJ kg−1 K−1
Substituting values: M mass of contents of vessel kg
t time s
80 − ¯P .249 = ¯P .150 T time constant s
whence ¯P = 49.9248 U overall heat transfer kW m−2 K−1
From the definition of offset, Equation 28.5: coefficient
  W heater power kW
TM   temperature ◦
C
 P − R ≡ . − P
TP 2
TM Subscripts
−0.0752 ≡ . − 9.925
TP
B ase position
whence C controller
TM
= 0.007577 M measurement
TP P process
Substituting into Equation 28.4 gives: R set point
TM
 = M + . = 1.5 + 0.007577 × 80 = 2.106
TP
Hence temperature variation in the tank is approx-
imately 2◦ C.

You might also like