2.

23 Ratio Control
F. G. SHINSKEY (1970, 1985) B. G. LIPTÁK (1995) J. E. JAMISON, A. ROHR (2005)

INTRODUCTION because of this divider. The differentiation of Equation

2.23(1) explains why this is true:
Ratio control systems maintain a relationship between two vari-
ables to provide regulation of a third variable. Ratio systems dR 1 R
= = 2.23(3)
are used primarily for blending ingredients into a product or as dB A B
feed controls to chemical reactors. An example would be the
Equation 2.23(3) shows that the loop gain varies both
addition of a gasoline additive under ratio control in order to
with the ratio R and the flow B. In most applications, the
maintain the required octane number of the product, which
ratio R would not often be subject to change, but the flow B
number may or may not be measured.
would. Because the loop gain varies inversely with flow B,
Ratio systems portray an elementary form of feedforward
this can cause instability at low rates. Therefore the use of
control (Section 2.9). In the above example, if the load input
an equal-percentage valve characteristics is essential to over-
to the system (gasoline flow) changed, it would cause a
come this danger. If the ratio were inverted,
variation in the controlled variable (octane number). This
variation can be eliminated by an appropriate adjustment of R = A/B 2.23(4)
the manipulated variable (additive flow rate).
then
The load, or wild flow, as it is called, may be uncontrolled,
controlled independently, or manipulated by other controllers
dR A R
that respond to the variables, such as pressure or level. =− 2 =− 2.23(5)
dB B B
and the result is essentially the same.
FLOW RATIO CONTROL The square-root extractors in Figures 2.23c are shown in
broken lines to indicate that the control system will also
Ratio control is applied almost exclusively to flows. Consider function without them, as it can use the flow-squared signals.
maintaining a certain ratio R between ingredients A and B: If that is the case, the controlled variable is
R = B/A 2.23(1)
R 2 = B 2/A2 2.23(6)
There are two ways to accomplish this. The more com-
mon method is to calculate and manipulate the set point of
a flow loop as follows:
Set Point = B = RA 2.23(2) Wild stream (A)

This system is shown in Figure 2.23a. The set point for FT

1
the flow ratio controller (FFC-2) is calculated by an adjust- Set point(R)
able-gain device, which is known as the ratio station. If the
Flow fraction FFC
ratio station (FY in Figure 2.23b) is outside the control loop, controller 2
it does not interfere with the secondary loop’s response. (ratio)
The second method is to calculate the ratio R from the FT
individual measurements of flows A and B (Equation 2
2.23[1]) and use this calculated ratio as the measurement Controlled stream (B)
input (controlled variable) into a manually set ratio control- Ratio-Flow transmitter FT-1 sets flow controller FT-2
ler (RIC). Such a scheme is shown in Figure 2.23c. The
principal disadvantage of this configuration is that it places FIG. 2.23a
a divider inside a closed loop. If flow B responds linearly Controlled stream is manipulated to maintain a constant ratio (R)
to the opening of valve B, the gain of the loop will vary of controlled flow (B) to wild flow (A).

305

© 2006 by Béla Lipták

306 Control Theory

Primary In this case too, for the configuration shown in Figure 2.23c,
flow (Fp) the loop gain varies inversely with flow.
√
FT
FV Advantages and Disadvantages
FIC
× The principal advantage of using the ratio control system is
Set FV
FY that the controlled variable — flow ratio — can be kept constant
FIC and can be directly recorded to verify control performance
Secondary √ FV (Figure 2.23c). In contrast, using the loop configurations in
FT
flow (Fs) Figures 2.23a or 2.23b would require two records and their
evaluation for ratio verification.
Fp
A The disadvantage of using analog transmitters in ratio
control is that it has no memory. For example, in a blending
R d system, if the controlled flow for any reason cannot match
r Gd
Comparator + e m + + Fs the required value, even temporarily, the resulting incorrect
Gc Gv Gp
– b composition cannot be automatically corrected because there
Gh is no memory of the amount of past error.
A = Input element. Gh = Feedback sensor transfer One way to try to correct this situation is to add a totalizer
b = Feedback variable. function. per stream to verify the overall correctness of total blended
d = Disturbance or load Gp = Process transfer
Variable.
product. If such information is available, subsequent operator
function.
e = Error (deviation) signal. Gv = Control valve transfer
intervention can bring the average composition of the blend
Fp = Primary flow. function. to the desired value. Naturally, this assumes that a correct
Fs = Secondary flow. m = Manipulated variable. mean ratio is acceptable for the purposes of the process,
Gc = Controller transfer r = Reference (set-point) rather than continuously maintaining a correct ratio.
function. Input.
Gd = Disturbance or load While head-type flow meters were used in the earlier exam-
R = Desired ratio.
transfer function. ples, the general comments that were made are equally valid
when positive displacement (PD) or turbine-type flow detectors
FIG. 2.23b
are used. When the plant is controlled by microprocessor-based
If the ratio calculation is made outside the secondary flow control
loop, its setting does not change the dynamics or response of the loop. digital controllers or DCS systems, the reconfiguration of such
ratio control loops requires much less effort. If the flow mea-
As a consequence, the scale of the ratio controller or ratio surement signals are transmitted utilizing bus-based technolo-
station has to be nonlinear. Differentiating Equation 2.23(6): gies, the associated dead time should be considered in the tuning
of the flow ratio control loops.
d ( R2 ) 2 B 2 R 2 R2
= 2 = = 2.23(7)
dB A A B
RATIO STATIONS

Flow A No matter what portion of the ratio control loop is imple-

mented in hardware or in software, a computing element must
FT
Ratio A be used whose scaling requires some consideration. The ratio
calculator station (FY in Figure 2.23b) normally has a gain range of
%
about 0.3 to 3.0. The primary flow signal Fp, in percent of
FY FY FY scale, is multiplied by the gain setting to produce a set point
for the secondary flow controller (Fs), in percent of scale.
SP The true flow ratio must take into account the scales of the
FIC Ratio two flowmeters. The setting of the ratio station is related to
controller
the true flow ratio by
FT
B FCV R = (true flow ratio)(Scale of Fp)/(Scale of Fs) 2.23(8)
B
As an example, assume that the true flow ratio of the
Flow B
additive to gasoline is to be 2.0 cc/gal (0.53 cc/ ). If the
FIG. 2.23c
additive flow scale is 0 to 1200 cc/min, and the gasoline flow
Here the ratio of flows A and B is calculated by a divider within scale is 0 to 500 gal/min (0 to 1890 /m), then:
the closed ratio control loop (FIC). The disadvantage of this
approach is that the loop gain changes every time the ratio setting 500 gal/min
R = 2.0 cc/gal = 0.833 2.23(9)
of flow B changes. 1200 cc/min

© 2006 by Béla Lipták

 2.23 Ratio Control 307

SETTING THE RATIO REMOTELY

TABLE 2.23d
Comparing the Actual Ratio Settings Required for Linear and
When a divider is used to calculate the flow ratio (Figure 2.23c),
Head-Type Flowmeters to Obtain the Same Flow Ratio
the set point for this ratio can come from other, related segments
Actual Ratio (Gain) Setting Required of the plant’s process. In order to do this correctly, a multiplier
to Achieve the Desired Ratio must replace the ratio station (Figure 2.23b). If the flow mea-
If Flow-Squared surement signals are linear, the usual choice of scaling factor
Flow Ratio If Signals Signals Are Used and for such a multiplier is 2.0:
Desired Are Linear the Scale Is Linear
B = 2RA 2.23(12)
0.6 0.6 0.36
In this way, a ratio of 1.0 appears at mid-scale of the ratio
0.8 0.8 0.64
input (R = 50%), when A = B. If squared flow signals are used,
1.0 1.0 1.0 a scaling factor of 3.0 provides a ratio range of 0 to 1.73:
1.2 1.2 1.44
B = 3R A
2 2 2
1.4 1.4 1.96
2.23(13)
1.6 1.6 2.56 One reason for using a multiplier to set the flow ratio
(Figure 2.23b) is the availability (even in hardware) of very
narrow ratio ranges for those applications in which the need
When head-type flowmeter signals are used and the scale for precision, not rangeability, is paramount. Ratio ranges of
is linear, such scale should show the square root of R in order 0.9 to 1.1, 0.8 to 1.2, etc. are possible.
to be meaningful. Table 2.23d compares the gain of a ratio A typical application would be the accurate proportioning
station (corresponding to linear flow ratio) with the ratio of ammonia and air to an oxidation reactor for the production
setting for head-type flowmeters. As shown by Table 2.23d, of nitric acid. If the compositions of the individual feeds are
the available range of ratio settings is seriously limited when constant, only a very fine adjustment of the ratio is required.
using squared flow signals.
A ratio controller (FFC-2 in Figure 2.23a) combines the
ratio function and the controller in one unit. If implemented RATIO CONTROL APPLICATIONS
in hardware, it saves cost and panel space. Such algorithms
are also available in digitally based control systems. Blending
Since ratio stations are used with remote-set controllers, Figure 2.23e shows a combination of cascade and ratio con-
some means must be available for setting flow locally during trols of blending fluids A and B and sending the mix into a
start-up, or during abnormal operation. An auto-manual station blend tank. In this process, the liquid level in the blend tank
is sometimes provided for this purpose. With the ratio controller, is affected by total flow, hence the liquid level controller sets
this feature requires two scales on the set point mechanism — flow A, which in turn sets flow B proportionately. (Whenever
one reading in ratio for remote-set operation, the other reading adjusting the set point of a head-type flow controller in a
in flow units for local-set operation.
When using a divider with linear flow signals as the ratio
calculator, the scale factor for the divider should be 1/2. This Flow A
places a ratio of 1.0 at midscale: FT
A
Set
FY FIC
point LIC
1B
R= 2.23(10) A A
2A Ratio
FY
set point AIC
Equal flow signals will produce a 50% output from the
divider, and the full ratio range is then 0 to 2.0, linear. If flow-
squared signals are used, the divider should have a scale factor FY FIC
of 1/3 to provide a full range of 0 to 1.73, with a square root B B
scale. This places a ratio of 1.0 (A = B) at 0.58 on the square
root scale. FT
B

1 B2 Flow B
R2 = 2.23(11)
3 A2
FIG. 2.23e
If a scale factor of 1/2 were used, the ratio range would The level controller manipulates flow rate and the composition
be restricted to 0 to 1.41. controller manipulates flow ratio.

© 2006 by Béla Lipták

308 Control Theory

cascade manner, in order to prevent instability at low flows, This new controller (VPC in Figure 2.23f) is usually
the square root is removed from the flow measurement signal selected to be the opening of the most-open control valve. If
and linear flow signals are used.) the most-open valve is held nearly fully open while the flows
Conversely, composition is not affected by the absolute are equally distributed, the control system is fully defined
value of either flow but only by their ratio. Therefore, to make and is also efficient because it provides the minimum resis-
a change in composition, the AIC controller must adjust the ratio tance to flow. This approach is applicable to all liquid or gas
set point of the multiplier (FY). To minimize the interaction of distribution controls, and its use is illustrated by the example
the composition controller with blend tank level control (through of distributing the returning cooling tower water among sev-
its manipulation of flow B), flow B should be the smaller of the eral cooling cells.
two streams. The purpose of the control system in Figure 2.23f is not
only to distribute the returning water correctly between cells
Distribution Controls A special case of ratio control is dis- but also to make sure that this is done at minimum cost of
tribution control. When a common supply of some material is operation. The operating cost in this case is pumping cost,
to be distributed among three or more destinations, an addi- and it will be minimal when the pressure drop through the
tional degree of freedom is created. Therefore, in addition to control valves is minimal. This is the function of the valve
setting the individual flow controllers to some percentage of position controller (VPC) in Figure 2.23f. As long as even
the total flow, one can also make sure that the system pressure the most open valve is not nearly fully open, the VPC adds
drop is minimized. Because equal flow distribution can be a positive bias to all the set-point signals of all the flow
achieved with all the valves nearly open, nearly closed, or fraction controllers (FFICs). As a result, all valves open and
anywhere in between, one must introduce an additional control keep opening until the most open valve reaches the desired
variable. 95%. This technique enables the meeting of the dual goals

FT

+
Σ
+

SP SP SP
FFIC FT FFIC FT FFIC FT

Set
FC FC FC at
95%

> VPC

Integral
M1 M2 M3 only
RA
Air Air Air
at at at
Air Air Air
Twb at Twb at Twb at
Twb Twb Twb

Users

FIG. 2.23f
Pumping costs can be minimized automatically by operating water distribution systems at minimum pressure drop and therefore with the
most open control valve nearly fully open.

© 2006 by Béla Lipták

 2.23 Ratio Control 309

of providing the correct flow distribution and doing it at a Users

minimum cost in pumping energy.
When cooling tower cells are manually balanced, it is Compressor
Pd
not unusual to find all balancing valves throttled to some
extent or to see the same water flow when the fan is on or Ratio
off. Both of these conditions are undesirable because they ∆P relay
∆PT ∆PY
will increase operating costs. The savings from automatic Pi
balancing can more than justify the instruments required for m(∆P) + b
implementing it. FCV
SP
h
Surge Control of Compressors FT FIC FO
Surge PI
measurement DA
In compressor surge protection applications, the ratio that is To vent or
controlled is not between flows. The surge line of variable recycle
speed compressors closely approximates a parabola if the Surge valve opens if
pressure rise across the machine (Pd − Pi) is plotted against h < m(∆P) + b
the volumetric inlet flow rate (Figure 2.23g). If the pressure Compressor
rise across the compressor (Pd − Pi) is plotted against the Surge line
speeds
orifice differential (h) of a head-type flow sensor located in

Pressure increase Pd − Pi , PSI

the inlet to the compressor (Figure 2.23h), the surge line 100%
becomes an almost straight line.
Because surge occurs at low flows, the control system is 90%
so configured as to keep the compressor operations to the
right of the surge line. Therefore, the set point for the surge 80%
protection FIC is the surge line itself (m(∆P)) plus a safety 70%
bias (b). This in effect locates the control line parallel with 60%
and to the right of the surge line. 50%

Differential pressure h, PSI

M

FIG. 2.23h
PDT The parabola shape of the surge line changes to a straight line if
FT Pd − Pi the pressure rise across the compressor (Pd − Pi) is plotted against
the orifice differential (h) of a head-type flow meter.
h

Surge valve The slope m of the surge line is

Characteristic m = (Pd − Pi)/h 2.23(14)
Surge limit
curves for various
compressor speeds where (Pd − Pi) is the head of the compressor and h is the
100% differential pressure across the flow element. A safe set point
Pressure increase Pd − Pi , PSI

F for the flow controller (FIC) is

90%
h = (Pd − Pi)/m + bias 2.23(15)
E
80% The complete surge control loop is shown in Figure 2.23h.
D The figure shows both the surge line and the control line at
70%
C which the surge valve opens.
60%
B 50%
A
Bibliography
Inlet capacity, actual ft3/min.
Balchen, J. G., and Mummè, K. I., Process Control Structures and Applications,
New York: Van Nostrand Reinhold, 1988.
FIG. 2.23g Beard, M. J., “Analog Controllers Develop New Wrinkles,” Instruments and
The surge line of a variable speed centrifugal compressor is para- Control Systems, November 1977.
bolic if the pressure rise across the machine (Pd − Pi) is plotted Buckley, P. S., “Dynamic Design of Pneumatic Control Loops: Parts I and
against inlet volumetric flow rate. II,” InTech, April and June 1975.

© 2006 by Béla Lipták

310 Control Theory

Buckley, P. S., Techniques of Process Control, New York: John Wiley & Murrill, P., Fundamentals of Process Control, 2nd ed., Research Triangle
Sons, 1964. Park, NC: ISA 1991.
Eckman, D. P., Automatic Process Control, New York: John Wiley & Sons, Nachtigal. C. N., Instrumentation and Control, New York: Wiley, 1990.
1958. Phillips, C. L., and Harbor, R. D., Feedback Control Systems, Englewood
Erickson, K. T., and Hedrick, J. L., Plant-Wide Process Control, New York: Cliffs, NJ: Prentice Hall, 1991.
John Wiley & Sons, 1999. Roffel, B., and Chin, P., Computer Control in the Process Industries,
Harriott, P., Process Control, New York: McGraw-Hill, 1964. Chelsea, MI: Lewis Publishers, 1989.
Jones, B. E., Instrumentation, Measurement and Feedback, New York: Rowan, D., “On-Line Expert Systems in Process Industries,” AI/Expert,
McGraw-Hill, 1977. August 1989.
Korn, G., Interactive System Simulation, New York: McGraw-Hill, 1989. Seborg, D., and Mellichamp, D. A., Process Dynamics and Control, 2nd
Luyben, W., Plantwide Process Control, New York: McGraw-Hill, 1998. ed., New York: McGraw-Hill, 2004.
Luyben, W. L., Process Modeling, Simulation and Control for Chemical Shinskey, F. G., “Control Topics,” Publ. no. 413-1 to 413-8, Foxboro
Engineers, 2nd ed., New York: McGraw-Hill, 1990. Company, 1976.
Luyben, W. L., and Luyben, M., Essentials of Process Control, New York: Shinskey, F. G., Distillation Control, 2nd ed., New York: McGraw-Hill, 1984.
McGraw-Hill, 1997. Shinskey, F. G., Process Control Systems, 4th ed., New York: McGraw-Hill,
Marlin, T., Process Control: Designing Processes & Control Systems for 1996.
Dynamic Performance, New York: McGraw-Hill, 2000. Smith, C. A., Automated Continuous Process Control, New York: John Wiley
McAvoy, T. J., Interaction Analysis, Research Triangle Park, NC: ISA, 1983. & Sons, 2002.
Merritt, R., “Electronic Controller Survey,” Instrumentation Technology, Stephanopoulos, G., Chemical Process Control: An Introduction to Theory
June 1977. and Practice, Englewood Cliffs, NJ: Prentice Hall, 1984.
Murrill, P., Application Concepts of Process Control, Research Triangle White, M. H., “Surge Control for Centrifugal Compressors,” Chemical
Park, NC: ISA, 1988. Engineering, Dec. 25, 1972.

© 2006 by Béla Lipták