THE COPPERBELT UNIVERSITY

School of mines and minerals sciences

CE 560 / MT 580
Process control

Lecture 08
Proportional-only control

C. Botha (Mr.)
Contacts: Chemical Engineering Department
Email: [email protected]
Alt. Email: [email protected]

January, 2022
 Design elements cont’d…controller design

o Once the appropriate control configuration is selected, the final step in designing a
control system is controller design.

o Several options are available: proportional-only, integral, derivative and a combination

of proportional with one or both of the latter.

o Controller is the active element that receives information from the measurements.

o Uses this information to take appropriate control action that results in adjustment of the
manipulated variable(s).

o Suitable control law for the controller can be arrived at by asking question (s) like:

o “How can measurement information be used to adjust values of the manipulated

variables?”
 Controller design cont’d…proportional (P-only) control

o The answer to the question above

Control valve
results in a suitable control law to be Mixture of A and B Pure A
used by the controller. X1 X2=1
w1 w2
o As an example, consider the blending
process previously discussed in figure
7.1. h

o Control objective is still the same, i.e.,

ensure tight control of the exit X
w
composition. Fig.7.1 blending system

o This is in spite of the changes in feed

flow-rate and composition.
 Controller design cont’d…proportional control

o Species A concentration in the feed stream can change at any time leading to exit
composition change.

o Despite changes in inlet composition, control of exit composition within a certain range
is a primary concern.

o Whenever a disturbance occurs in a process, it is natural to think of changing the

manipulated variable proportional to some error amount.

o For the case in Fig. 7.1, the exit composition can be measured and how this
measurement is used by the control is expressed by eq. 7.0.

o That is, compute deviation of the measurement from desired value then adjust the
manipulated variable proportionally to the deviation amount.
 Controller design cont’d…proportional control

o That is;
w2 (t ) = w2 + K c  xsp − x(t )  ........................................(7.0)

o The difference, 𝝐 = 𝒙𝒔𝒑 − 𝒙(𝒕) denotes the error or the amount by which the
measurement deviates from the set-point.
o Note that 𝒙𝒔𝒑 is the set point or desired value of the outlet composition and 𝒙(𝒕) is the
measured exit composition at time 𝒕.
o Eq. 7.0 implies that flow-rate 𝒘𝟐 must be varied proportionally to the error to keep
outlet composition within a desired range.
o When manipulated variable is adjusted in this manner, the control mode is referred to as
proportional (or -only) control.
 Controller design cont’d…proportional control

o Kc is the proportional control constant often called controller gain.

o Note that manipulated variable is adjusted from steady-state value hence presence of
ഥ2 in eq. 7.0
𝑤
o From eq. 7.0 again, a large deviation from set point produces a corresponding large
corrective action.
o Conversely, a small deviation from set point produces a small corrective action.

o The controller gain 𝑲𝒄 must be positive since when 𝒙 increases, 𝒘𝟐 must decrease.

o Or conversely, when 𝒙 decreases, 𝒘𝟐 must increase.

 Controller design cont’d…proportional control

o Proportional control is usually used in Fi (ft3/min), Ti (oF)

conjunction with feedback.

o As another example, a stirred tank heater T

h
discussed in previously is consider. Q
F (ft3/min), T (oF)

o Control objective is still the same, i.e., it is Condensate

desired to control the temperature 𝑻 of the Fst (Ib/min)

Steam

liquid in the tank as 𝑻𝒊 changes.

Fig.7.2 Stirred tank heater system

o Assuming that the liquid hold-up in the tank
remains constant.
 Controller design cont’d…proportional control

o That is, inlet flow-rate must be equal to outlet flow-rate for hold-up to remain const.

o For this process, the measurement will be the liquid temperature 𝑻 and the manipulated
variable will the heat input 𝑸 (𝒐𝒓 𝑭𝒔𝒕 ) provided by steam.

o An appropriate control mode will determine exactly how the heat input from the steam
should be varied in order to keep 𝑻 constant when 𝑻𝒊 changes.

o The simplest law, proportional-only can be used, expressed as

Q = Qs + K c (Ts − T )...................................................(7.1)
 Controller design cont’d…proportional control

o Effectiveness of this control mode in attenuating the error (deviation from set point) can
be analysed in the following way.
o Assuming that the stirred tank has been operating at steady state, that is, temperature of
the liquid in the tank 𝑻 = 𝑻𝒔 and liquid volume 𝑽 = 𝒄𝒐𝒏𝒔𝒕.
o Under these conditions, the steady-state energy balance on the stirred tank heater
system yields
Heat absorbed by liquid in the tank = heat supplied by steam
F  C p (Ts − Ti ,s ) = Qs ............................................................(7.1)
Rearranging: 0 = F  C p (Ti ,s − Ts ) + Qs ........................................................(7.2)

o In eq. 7.1, 𝑭, 𝝆 and 𝑪𝒑 are the flow-rates of the inlet or outlet streams, liquid density and
the liquid’s specific heat capacity.
 Controller design cont’d…proportional control

o 𝑻𝒊,𝒔 , 𝑻𝒔 and 𝑸𝒔 are the corresponding steady state values of the inlet temperature, liquid
temperature and steam heat supply.
o Process will be disturbed if any of the input variables 𝑻𝒊 𝒐𝒓 𝑭𝒊 changes.
o Suppose inlet temperature 𝑻𝒊 suddenly increases as a step change shown in Fig. 7.3.
o Liquid temperature 𝑻 in the tank will likewise begin to increase provided that nothing is
done on the heat supply 𝑸 i.e. no control.
o The transient energy balance around the tank can be used to determine how the liquid
temperature 𝑻 changes with time 𝒕 . That is;
Rate of energy  Rate of  Rate of  Rate of heat 
       
accumulation  = heat inflow  − heat outflow  + addition  ......(7.3)
 in the tank   with the feed   with the feed   by steam 
       
 Controller design cont’d…proportional control

dT
V C p = F  C p (Ti − T ) + Q...............................................(7.4)
dt

Kc= 0 (No control)

Error
(T-Ts )
Ti
Kc= 1

Kc= 2
Ti,s
Off-set

to Time 0 Time
Fig.7.4 Temperature response under proportional
Fig.7.3 Inlet temperature step change
feedback control
 Controller design cont’d…proportional control

o Expressing the transient energy balance in terms of deviation variables i.e. subtracting
equation 7.2 from equation 7.4 yields
d (T − Ts )
V C p = F  C p (Ti − Ti ,s ) − (T − Ts )  + ( Q − Qs ) ........(7.5)
dt
d (T − Ts ) dT
o Note that, = since Ts = const (steady-state).
dt dt

o Also, the difference 𝝐 = 𝑻 − 𝑻𝒔 is the error or the deviation of the liquid’s temperature
from the desired value at time 𝒕.

o Ideally, it is desirable to drive the value of this error to zero by manipulating appropriately
the value of the heat input 𝑸.
 Controller design cont’d…proportional control

o Using the suggested proportional control mode, after rearranging

Q = − K c (T − Ts ) + Qs ...................................................(7.6)

o Eq. 7.6 the parameter 𝑲𝒄 is the proportional gain.

o Rewriting eq. 7.5 in terms of the proportional gain by substituting eq. 7.6 into eq. 7.5
gives
d (T − Ts )
V C p = F  C p (Ti − Ti ,s ) − (T − Ts )  − K c (T − Ts )........(7.7)
dt
o Eq. 7.7 can be solved for (𝑻 − 𝑻𝒔 ) for various values of the controller gain 𝑲𝒄 .
o Solution gives the temperature response under proportional-only feedback control and is
shown in Fig. 7.4.
 Controller design cont’d… proportional control

o Quality of the control is better for large values of Kc (Fig. 7.4 )

o That is, less error or deviation of temperature from nominal value as Kc increases.
o Although, for all values of Kc none of the response is desirable, since 𝑻 − 𝑻𝒔 ≠ 𝟎.
o Proportional control depends on continuous existence of the error signal.
o Hence, a sustained or persistent error must remain for continued effective control
action.
o This means, for a sustained change in load (disturbance), controlled variable never
return completely to the desired value.
o But rather it attains a new equilibrium value known as the control point.
 Controller design cont’d… proportional control

o The difference between this new equilibrium value (control point) and the set point value
is referred to as offset.
o Reason for offset with proportional-only control can be better understood by considering
Fig. 7.4 again.
o Suppose that inlet temperature decreases, then clearly, the temperature in the tank and
outlet stream will begin to fall.
o In response, controller will open the valve further by an amount proportional to the error.
o Consequently, exit temperature will get closer to desired value.
o As exit temperature gets closer and closer to desired value, the error will be so small
that additional output Kc times error will not be enough to drive controlled variable to
desired value.
 Controller design cont’d…proportional control mode

o The offset can be reduced by increasing Kc. However, to completely drive controlled
variable to setpoint, Kc must be made infinite.
o The fundamental idea behind proportional control is that:
o 1) the controller output is varied proportionally to the error signal. The general expression
for the proportional control law is:
CO = CObias + K c e(t )...................................................(7.8)
e(t ) = error signal = ysp (t ) − ym (t )................................(7.9)
Also written as e(t ) = SP − PV ..........................................................(7.10)
o Where 𝐂𝐎 and 𝐂𝐎𝐛𝐢𝐚𝐬 are the controller output and bias or null value or the steady state
output.
o The bias (null or steady-state value) is the controller output when the SP and PV are the
same so that the error signal is zero.
 Controller design cont’d…proportional control mode

o 𝑲𝒄 measures the change in controller output per unit change in error.

o It indicates strength of the control action in relation to the error signal or how fast the
system responds.
o 𝑲𝒄 (is a tuning parameter) is usually dimensionless when 𝐂𝐎 and 𝝐 𝒕 have the same
units otherwise may have units.
o And as before 𝒚𝒔𝒑 𝒕 and 𝒚𝒎 𝒕 are the set point (or SP) and measurement (or an
equivalent signal from the sensor/transmitter) also called Process variable (PV).
o Set point is usually kept constant for long periods in most industrial processes though
can vary with time.
o Controller gain can be adjusted to make controller output as sensitive as desirable to the
error.
 Controller design cont’d…proportional control mode

o The sign of Kc can be chosen to make the controller output increase or decrease. A case of
positive or negative feedback control.
o For proportional controllers, the bias can be adjusted - a procedure referred to as manual
reset.
o In this case, bias is adjusted so that controller output and manipulated variable are at their
nominal values at which the error is zero.
o Realistic representation of a controller includes controller saturation.
o In this case, controller output reaches a physical limit, that is, either 𝐂𝐎𝒎𝒂𝒙 or 𝐂𝐎𝒎𝒊𝒏 is
reached.
o Controller saturation occurs when the disturbance or set point change is so large that its
beyond the range of a manipulated variable or physical limits of a controller.
 Controller design cont’d…proportional control mode
o Some controllers have a proportional band (PB) setting rather than a controller gain.

100%
PB(in %) ..........................................................(5.13)
Kc

o Where Kc must be dimensionless. A small (narrow) PB corresponds to large Kc and

large (wide) PB corresponds to small Kc.
o PB represents range over which error must change to drive controller actuating signal
over its full range.
o Fig. 7.5 shows effect of the controller gain on process response.
o For controllers with a proportional gain setting, increasing this parameter makes the
response faster and aggressive to a change in the error signal.
o This may cause oscillatory and unstable response with proportional-only control.
 Controller design cont’d…proportional control mode

Kc=18
Kc=8
Kc=1

SP

0 Time

Fig.7.5 Proportional-only control: effect of proportional gain on process response

 Controller design cont’d…proportional control mode

o Conversely, for the controllers with proportional band (usually expressed as percent),
decreasing this parameter results in an increase in proportional gain.
o This again makes the response faster and more aggressive and may lead to instability
due to oscillatory behaviour provided that 𝑲𝒄 is made too large.
o Knowing whether the controller uses Kc or PB is essential to ensure that right
adjustments are made.
o Inherent disadvantages with P-only control is presence of steady-state error.
o Furthermore, it might be tempting to increase 𝑲𝒄 indefinitely to reduce the offset.
o Response will be faster, however, may result in oscillations and unstable behaviour.
 Controller design cont’d…proportional control mode

o This is yet another disadvantage of proportional-only control.

o The usual setting for 𝑲𝒄 is a compromise between set point and degree of oscillatory
behaviour.
o Since proportional-only control is a simple form of control, it is frequently employed on its
own.
o When offset is not an important consideration and in cases where the system is
sufficiently stable so that a fairly high value of 𝑲𝒄 can be tolerated.
o Then P-only control can be used.