Feature Report

pH Measurement

And Control

Leached
layer

When measured correctly, pH can be an invaluable

tool for both product and process control
Greg McMillan and Rich Baril
Emerson Process Management

he mechanical design and material science of pH-sensor technology have evolved dramatically
over the last century. Currentday uses include a wide range of applications from high-purity pharmaceutical processing to wastewater
treatment. Some of these applications
pose high challenges to pH measurement, including high-temperature and
otherwise harsh measuring environments, stability (particularly in pharmaceutical processes), low conductivity measurement and more-stringent
pH control in processes. This article
outlines the basics of pH measurement and control, with focus on some
of these applications, and explores recent advances such as the development of embedded memory chips and
transmitters with smarter diagnostics
and wireless communication that
have been developed to offer predictive maintenance, portability and
greater stability.

pH Basics

Soren Sorensen of the Carlsberg Laboratory of Copenhagen defined the hydrogen exponent as the negative logarithm of hydrogen-ion concentration
equated to ion activity. This introduced
the concept of pH and started the era
of its measurement. The science behind the pH-glass formulation, reference-electrode designs, electrolytes
and buffer solutions rapidly developed
[1, 2] and prompted the appearance
of the first laboratory-type pH meters. Soon afterwards, industrial pH
sensors were produced and environmental mandates prompted the use
32

of pH electrodes in effluent streams

throughout the chemical process
industries (CPI). The recognition
of the benefits of online pH control
to improve product quality and efficiency created a demand for specialty sensors that could withstand
harsh process conditions and sterilization. The need to reduce costs
and improve reliability resulted in a
number of improvements in design
and intelligence of electrodes and
transmitters. A summary of the current state-of-the-art follows.

Glass electrodes

Process
liquid

Reference
electrolyte

Figure 1. A typical hemi-bulb of the pHsensitive glass of the industrial pH sensor

is shown at the top and a schematic representation of the leached layer formation
on the both sides of the pH-glass is shown
below it. If the pH of the reference electrolyte
and process solution are equal, then there is
no potential (E) generated across the glass,
E = 0. Typically, the reference electrolyte is
buffered to pH 7, so no voltage is generated
if the process liquid has the same acidity as
the buffer. If the process liquid is acidic, then
E < 0, and negative potential is generated. If
the process liquid is alkaline, then E > 0, and
positive potential is generated

It is the glass in a glass pH sensor that

responds to the acidity level of the solution. The pH-glass formulation is
the best-kept secret of any pH-sensor
manufacturer. However, it is not a secret that alkali metals render silicate
glass pH-sensitive. The leached layer
is formed on the very surface of the
glass membrane (semi-bulb) once its
hydrated (Figure 1). This phenomenon
gives us a hint on how to store and
maintain the life of the glass electrode
simply keep the tip in an aqueous
solution. Once the glass is dehydrated,
because the bulb is not immersed or is
exposed to non-aqueous chemicals, the
leached layer disappears and the sensor stops working or develops an erratic signal. The leached layer is very
thin, only about 510 nm and about
100 nm for the transition layer. However, this layer contributes the most
toward the conductivity of the pH
glass. The electrical resistance of the
surface layer increases as glass ages
[3]. Most modern analyzers can measure glass impedance of the pH sensor.
If the effect of process temperature on
glass impedance can be addressed, the

Chemical Engineering www.che.com August 2010

Glass
bulk

health of the glass can be monitored

and the replacement of the electrode
scheduled as necessary.
Aging and electrode life. The impedance of the glass is proportional to
the age of the glass. The simple rule of
how to predict the life of the glass electrode is to monitor the glass impedance. A positive trend would indicate
that the glass is aging. Interestingly,
a process coating of the glass would
create a similar effect. To distinguish
between the two cases one needs to
clean the sensor and re-measure the
glass impedance in buffer solution. If
the impedance drops to the original
value of the glass, then the cause is a
coating. An appropriate cleaning procedure should be selected and scheduled to insure a good service life. If the
glass impedance value does not change
after cleaning, then the cause is aging.
Both coatings and aging can exhibit a
sluggish response [4]. If the test time is
long enough to see an entire response,
aging may also show up as a decrease
in the final change in the indicated pH.
This loss in electrode efficiency or span

7.00 pH

Temperature
compensation

Glass
electrode

Reference
electrode

Reference
junction

Liquid junctions

Orifice-junction

Helical pathway
reference

Process
liquid

Figure 2. The glass electrode creates the potential proportional

to the pH of the process liquid, while the reference electrode completes the electrical circuit and provides a small and stable potential as a reference for the pH signal. Electrode temperature compensation is not absolutely necessary for the pH measurements,
however is highly desirable, especially at elevated temperatures

occurs when the supply of alkali ions

is exhausted and the sensor looses its
sensitivity and the slope (mV versus
pH) of the electrode decreases.
Interestingly, a decreasing electrode temperature has a similar effect, so it is important to monitor the
temperature of the process and use
the electrode temperature compensation for the change in millivolts generated per pH unit as defined in the
Nernst Equation [1, 2]. Most pH sensors have built-in electrode temperature compensation. The solution pH
is also temperature dependant. The
temperature dependence of the water
dissociation constant causes the pH to
decrease with temperature by approximately 0.03 pH/C above neutrality
[2]. The temperature coefficient for
an actual process can be different because of the temperature dependence
of weak-acid and weak-base dissociation constants. The coefficient can be
found by running temperature tests of
the process liquid and monitoring the
pH response. Solution temperature
compensation, unlike electrode temperature compensation, is up to the
user to identify and configure.
Operating temperature. The operating temperature is also important in
sensor selection. Usually low impedance glasses (<10 M) are used for low
operating temperatures, mid-range
impedance glasses (10100 M) are
employed for ambient temperatures,
and high impedance glasses (>100
M) are reserved for high temperature applications. It is important to
remember that glass impedence approximately doubles per every 10C
as temperature decreases [2, 3]. Even

Figure 3. A cross-sectional view of a triple-junction

reference electrode is shown on the left. Multiplejunction
solutions are usually used for combating poisoning agents.
The orifice-junction solution, shown in the middle (magnification 250X) is used in low-conductivity applications. A
helical-pathway design (right) is used to prevent contamination (poisoning) of the internals
7.00 pHof the pH sensor

though the choice and performance of tion. The choice of the materials for
the glass is critical, the performance the reference junction can vary from
Reference
of the pH sensor also depends upon Temperature
Teflon to ceramic,
to wood, to just a
compensation
electrode
the design and integrity of the refer- tiny hole, as shown
in Figure 3. The
ence electrode.
purpose of the reference junction is
to provide electrical continuity with
the
process liquid.
It also serves as a
Reference electrodes
Reference
Glass
to prevent junction
the process liquid
The purpose of the reference electrode guard
electrode
is to create a stable reference poten- from penetrating and contaminating
tial against which the pH signal can or poisoning the sensor.
Process
Each liquid
and every component of the
be measured. Having a constant reference potential is essential for high reference electrode contributes to
impedance measurement. After all, the reference potential. This is why
a pH sensor is essentially a high im- it presents an engineering challenge
pedance voltmeter (Figure 2). The to create a stable design. Typically,
stability of the reference potential the reference impedance is an order
can be achieved through a variety of of magnitude less than the glass immechanical designs, such as single, pedance. The reference impedance can
double and triple-junction; differential also be monitored by the analyzer if
design, which can be described as hav- the pH sensor has a solution ground.
ing a pH sensor inside the pH sensor; A gradual increase in the reference
application-appropriate choice of the impedance is indicative of a coating
junction materials and electrolytes of the reference junction. The sensor
needs to be cleaned regularly to have
(Figure 3).
The typical reference in commercial an extended service life. However, an
pH sensors uses a potassium chloride increase in reference impedance may
electrolyte to satisfy all of the require- also be symptomatic of a plugged juncments for the reference fill solution tion and depletion or poisoning of the
[2]. The consistency of the electrolyte reference electrolyte. If the pH sensor
can vary from liquid to gel to solid in is rebuildable instead of disposable,
order to slow down the migration of its possible to prolong service life of
process ions inside the reference that the sensor by replacing the outer junccome in through the reference junc- tion and refilling the electrolyte. The
tion. Depending on the application, cost of ownership of the rebuildable
the chemistry of the electrolyte can type is usually less than that of a disbe tweaked or even changed to satisfy posable sensor.
specific requirements of the chemical
process. The highly competitive mar- Installation and wiring
ketplace offers rebuildable sensors The process-installation procedure is
with a variety of fill solutions.
an important factor for the service life
The reference junction is usually a of a sensor. Improper installation can
part of a rebuild-kit and is pre-soaked cause incorrect readings and even dein the corresponding electrolyte solu- stroy the sensor. Most glass electrodes
Chemical Engineering www.che.com August 2010

33

Reagent Titration Curves

14

Feature Report

12
10

For a strong acid and base the

pKa are off-scale and the slope
continually changes by a factor
of ten for each pH unit deviation
from neutrality (7 pH at 25 C)

8
pH
6

have an air gap inside the sensor to

allow for thermal expansion. If this
type of sensor is installed upsidedown, an air bubble becomes trapped
right against the pH glass and causes
an erratic signal or loss of measurement. The sensor tip should point
down about 20 deg from horizontal.
Non-glass pH sensors are insensitive to this issue. Installing a sensor
too close to an agitator or a pump can
cause premature failure, noisy readings and electromagnetic interference.
Sensors with integral pre-amplifiers
allow longer cables and have less noise
in the signal. However, long cables increase the installation, maintenance,
and replacement cost. Quick disconnect cables offer some ease of use but
do not eliminate the expense. Having
a pre-amp on the sensor side dramatically improves signal reliability
and signal-to-noise ratio. A typical pHcurrent signal is on the order of tenths
of a nano-amp. It is only reasonable to
amplify a signal that small before it is
transmitted through meters of wires.

Choosing the right sensor

Typically, process and service engineers perceive the pH sensor as a tool

or a black box that has to do the job.
Its important, however, to understand
that there is no one universal pH
sensor for each and every application.
Many specific application and maintenance challenges can be addressed
by technologically advanced sensors.
However, the increasing complexity
and the cost of these specialty sensors
may not be desirable for general-purpose applications. General-purpose applications imply relatively clean water,
no coating, no poisoning ions, good
conductivity, moderate temperature
and pressure range and no sanitary
requirements. In contrast, specialty
sensors are designed to withstand
high temperature and pressure conditions, combat fouling and coating, and
perform well in low conductivity water
as well as in poisoning applications.

High-temperature performance

High temperature processes (for example, ethanol production, geothermal plants, and breweries) put a lot
of stress on a glass electrode. Premature aging and the decrease in the
34

Reagent Titration Curves

2
0

11
10

Reagent / Influent ratio

Despite appearances, there are no straight

lines in a titration curve (zoom in reveals
another curve if there are enough data points,
a big if in neutral region)

8
pH 7
6
5

Yet titration curves are essential for every aspect

of pH system design but you must get numerical
values and avoid mistakes such as insufficient
data points in the area around the set point

4
3

Reagent / Influent ratio

Figure 4. While titration curves are essential for every aspect of pH-system design, you must get numerical values for the spectrum of operating temperatures and
feed conditions and avoid mistakes, such as insufficient data points in the control
region near the setpoint

Table 1. Trending data and follow-up actions

Diagnostic trend

Sensor status

Actions

Glass impedance
increases over time

Glass is aging or
gets coated

Schedule maintenance/
inspection

Slope (mV versus pH)

decreases over time

Glass is aging or
gets dehydrated

Schedule maintenance/
inspection

Reference impedance
increases over time

Coating or
poisoning

Schedule cleaning and

calibration check

Reference offset increases

over time

Coating or
poisoning

Schedule cleaning and

calibration check

speed and magnitude of the response

starts at temperatures above 50C.
The need to retain pH sensitivity and
withstand thermal shocks has led to
more-advanced pH-glass formulations and construction [5]. Most competitive high-temperature sensors
have ratings up to 155C at 400 psig,
but their performance and life expectancies vary. The best way to extend
the life of a sensor at high temperatures is to select high-temperature
glass based on test results, use electrode temperature compensation and
monitor glass impedance.

High-purity processes

Pharmaceutical processes. Repetitive thermal and chemical shocks, such

as those in steam-in-place (SIP) and
clean-in-place (CIP) procedures, create a challenging environment for pH
sensors. Both SIP and CIP capabilities
are usual requirements in the pharmaceutical industry. A pH sensor that
retains its calibration after steam cycles or autoclaving is one of the biggest
challenges for pH-sensor manufacturers. No where else is more at stake,
in terms of pH performance. A faulty

Chemical Engineering www.che.com August 2010

pH sensor, for example, can cause the

loss of a batch worth ten million dollars that takes 10 to 20 days to process.
As with electrodes designed for high
temperature, recent improvements in
sensor structure and glass formulation have significantly improved the
performance and life of electrodes subjected to sterilization. The drift after
sterilization has been significantly decreased and the number of steam and
autoclave cycles that can be tolerated
has been greatly increased.
Noise and accuracy in low-conductivity applications. High-purity
(HP) applications with low conductivity (<10 S/cm) present another
challenge for pH-sensor design. A
typical example is in the power industry for measurements in boiler
water, return condensate and feedwater. The pH of the feedwater must
be controlled within a very narrow
range to minimize corrosion of the
boiler hardware. The challenge of
measuring pH in low-conductivity
water manifests itself in the fact that
there is not enough ionic strength to
provide a pH signal. This means that
additional charge carriers need to be

pH

Set point

B
Er = 100% Fimax
Frmax

Ss = 0.5 Er

Where:
A
=
B
=
Er
=
Frmax =
Fimax =
Ss
=

Control band

Influent pH

Frmax = 1.25 A Fimax

B
Er = 80%
A

Figure 5. The titration curve is used to

design the control valve and see the pH
oscillations from valves, disturbances,
loops, and non-ideal mixing in flow or
concentration on the x-axis by numerical
translation to the y-axis
B

Reagent flow
Influent flow

The effect of concentration fluctuations or

gradients (noise) on measurement error
and control valve resolution is less on flat
portions of the titration curve but high acid
or base concentrations may attack glass
and materials of construction and affect
reference junction potential

Controller tuning

distance of center of reagent error band on abscissa from origin

width of allowable reagent error band on abscissa for control band
allowable reagent error (%)
maximum reagent valve capacity (kg per minute)
maximum influent flow (kg per minute)
allowable stick-slip (resolution limit) (%)

introduced for stable readings.

A flowing junction type of sensor is
typically employed for these applications because the small flow of electrolyte through the reference junction
into the process maintains a smaller
and more-consistent reference junction potential. Also, the sample flow
past the sensor is kept low and constant by the electrode holder (cell) design, reducing erratic readings from
streaming potentials. Having the
controlled low-flow cell is extremely
advantageous for good accuracy and a
good signal-to-noise ratio. Usually this
type of measurement is conducted in
a side stream to avoid contamination
of the process. In tightly controlled
conditions, the achievable level of accuracy is about 0.05 pH with a noise
level less than 0.02 pH. The best way
to maximize the sensor life in lowionic-strength applications is to know
the stream composition (presence of
contaminants), take measurements
under controlled conditions, monitor
the diagnostics, and have a conductivity measurement of the same stream.

pH-control challenges

The challenges of pH control are actually a separate subject from pH measurements. However, measurement is
an integral part of the pH-loop performance for control. If you consider the
scale of 014 pH, the glass electrode
can measure hydrogen ion concentrations of 1 to 1014 normality. The
extraordinary range and sensitivity
of the pH measurement creates exceptional challenges for pH control in

terms of control-valve and equipment

selection, mixing uniformity, reagent
injection, and controller tuning [6, 7].
The most oscillatory loops in a plant
are often pH loops.

Graphical deception

pKa (where pKa = pKw pKb). The effect of the associated acid or base concentration and process temperature
on the slope is the greatest at these
points on the curve [6].

Nearly all of the implications for process control can be analyzed in terms
of the slope of the titration curve,
which is representative of the process
gain (change in pH for a change in reagent addition). Systems with strong
acids and bases are the most difficult
to control because the slope changes
by a factor of 10 for every pH-unit deviation from the neutral point (7 pH at
25C), which translates into a process
gain at neutrality that is potentially
107 times larger than the process gain
at the ends of the pH scale. Most laboratory titration curves show a vertical
straight line in the neutral region (4
to 10 pH), which gives the illusion that
the process gain is constant and the
process is linear. However, a zoom centered on the neutral point would reveal another S-shaped titration curve
if there were sufficient data points
(Figure 4). Successive zooms on what
appears to be a straight line would
reveal further S-shaped curves. Furthermore, the slope depends upon process temperature via the dependence
of the dissociation constants for water
(pKw), acids (pKa), and bases (pKb) on
temperature. The presence of a weak
acid or base will cause a moderation
of the slope centered at a pH equal to
the pKa. To spot this effect, the pKb of
a base is converted to the equivalent

The controller gain is inversely proportional to the process gain. Consequently, the change in controller tuning due to the nonlinearity of pH, is
way beyond what you experience in
other process control loops unless you
are lucky enough to have high concentrations of weak acids or weak bases
and are operating close to their pKa.
Tuning settings are either scheduled
based on pH, or the controlled variable
is converted from pH (y-axis) to percent reagent demand (x-axis) based on
a representative titration curve to reduce the nonlinearity. The scheduling
and linearization is not perfect due to
changes in process temperature and
feed concentrations, but is better than
no embedded process knowledge.

Control-valve design

The smallest change possible in a control valve position multiplied by the

process gain at the setpoint must be
less than the allowable deviation in
pH from the setpoint. The smallest
change in position is the resolution
of the control valve expressed as a
percent of the signal span and hence
valve capacity. The typical source of a
valve resolution limit is stick-slip. Integral action in the controller creates
a sustained equal-amplitude oscillation (limit cycle) even if there are no
disturbances, as the controller output
ramps back and forth across the resolution limit. The best pneumatically
actuated control valves with digital
positioners have a resolution of 0.1%
of span. Since the resolution is ultimately a percent of valve capacity, the
resolution required depends on the
distance of the incoming pH from the
setpoint on the titration curve. Figure
5 shows how a reagent flow oscillation is amplified by the process gain
and how the valve resolution required
is proportional to the width of the allowable control deviation (B) and is
inversely proportional to the distance

Chemical Engineering www.che.com August 2010

35

Feature Report
(A) traversed on the titration curve
[6]. It is important that data are used
rather than the plot from a laboratory titrator, that there be at least five
data points in the control region, and
finally that the x-axis be converted to
a ratio of reagent to feed flow.
Concentration and feed disturbances are amplified by the process
gain just as the disturbance from valve
resolution was in Figure 5. Similarly,
the variability in reagent concentration from non-ideal mixing is amplified. The smoothing of oscillations and
fluctuations is inversely proportional
to the residence time of a well mixed
volume. The attenuated amplitude per
Equation (1) must be translated from
the x-axis (ratio of reagent to feed flow)
to an amplitude on the y-axis (pH) of
the titration curve to include the effect
of the nonlinear process gain [6].
 Ao  Ai s

To
2sP sTp

(1)

Where:
Ai is the amplitude of input oscillation
into volume (reagent to influent ratio)
Ao is the amplitude of output oscillation from volume (reagent to influent
ratio)
To is the period of oscillation, minutes
p is the process time constant from
mixing (residence time), minutes

Mixing and injection delays

A controller cannot compensate for a

disturbance until it sees the change
in pH and can get the reagent into
the process at the same point as the
disturbance. The total loop deadtime
involved in the observation and compensation can be approximated as the
sum of delays and small lags in the
loop. This sum should include sample
transport delay, electrode lag, transmitter lag and update delay, DCS
execution time delay, DCS filter lag,
control valve delay and lag, injection
delay, and mixing delay and lag. The
injection delay that occurs when a
reagent valve opens or closes is often
the largest and most insidious source
of loop deadtime [6]. This injection
delay can be approximated as the
dip tube and piping volume between
the injection point and the reagent
valve divided by the reagent flow.
The error from a process disturbance
36

Figure 6. A software screenshot shows the health status of the pH-sensor in a

process. These variables can be historized to see longterm trends

after control action is proportional to

the total loop deadtime. An extremely
effective strategy to minimize the
amount of process nonlinearity is to
minimize the loop deadtime [6, 7]. In
other words, loops with minimal deadtime and well-tuned controllers keep
the pH close to the setpoint. On the
other hand, poor pH-system designs
exhibit larger excursions on the titration curve and changes in process gain
that make the problem worse.

Control solutions

Traditional solutions. The classical solution was to have a large, wellmixed vessel for every 2 pH units of
the feed pH from the setpoint. For a
feed around 2 pH or 12 pH, three wellmixed vessels in series were used.
For the 2 pH case, the setpoints of
the vessels were 4, 6, and 7 pH. The
first vessel typically had the largest
volume to cross neutralize the influent and the largest control valve to
provide a coarse adjustment to move
the pH most of the distance along the
flatter part of the titration curve. The
last vessel had the smallest volume
and smallest control valve to provide
a trim adjustment. The volumes were
designed to be significantly different
in size to prevent the resonance of oscillations since the natural frequency
of the loop was thought to be mostly
dependent on volume. These vessels
had a large capital cost and footprint,
and were prone to large and slow oscillations from the limit cycle of the
large valve on the first volume that
were not effectively smoothed by the
last volume.

Chemical Engineering www.che.com August 2010

Alternative system solutions. Today,

inline pH-control systems followed
by a moderately mixed volume are
used to reduce cost and space needs
[68]. These inline control systems use
pumps, piping fittings and static mixers to provide coarse mixing. Inline
control systems have a much faster response and period of oscillation than a
vessel control system. The downstream
volume provides the needed attenuation per Equation (1). Signal characterization that translates the loops
controlled variable from pH to reagent
demand (x-axis of the titration curve)
is used to suppress reaction to mixing
noise and to help the controller recognize the true distance of the incoming
pH to the setpoint in terms of reagent
addition [68].

Maintenance

As we can see, the overall pH performance is a fine balance of many

variables, and maintenance is one of
them. The burden of maintenance is a
challenge that should not be underestimated. The pH sensors have a finite
life both storage and service-life. The
sensor performance depends on the
process conditions as well as on proper
maintenance and installation. Routine
maintenance usually boils down to
cleaning with water, and a two-buffer
calibration check. If the coating is acidic,
the use of a weak caustic solution (<4%
NaOH) is recommended. If its alkaline
or a scale-coating, then vinegar or ~5%
HCl can help. Oil, grease and organic
compounds can be taken away by using
off-the-shelf detergents or organic solvents that are compatible with the sen-

Figure 7. Actual trend recording of a single-use-bioreactor batch shows that a

wireless transmitter did not show the noise spike seen by the wired transmitter, and
the wireless control of the batch is within 0.001 pH despite a reduction in the number
of communications by 60% to increase battery life

sor body materials. Monitoring the pHsensor diagnostic parameters is very

helpful for predictive maintenance. For
instance, glass impedance can indicate
if the glass is broken, coated or spent.
The reference offset and reference impedance can indicate coating and or
plugging. The pH span and offset can
reflect the sensitivity and number of
the standardizations. The low level of
the pH signal makes it susceptible to
noise. Installation of the sensor in close
proximity to air conditioners, pumps,
agitators, mixers and motors can result
in electromagnetic interference. Piping
material can make a difference in pH
measurements. Its well known that
metal pipes have far less problems with
grounding than plastic ones or metal
pipes with a plastic lining. Correct wiring and proper cabling also fall under
the category of maintenance. Interestingly, some locations or installations for
pH monitoring can be cost-prohibitive
due only to the wiring. The cost to run
the cables all over the plant or to a remote location can be excessive.

Recent advances

Recent developments in sensor and

transmitter intelligence and communication are an indication of the future trend of pH measurements to be
smarter and more portable.

Embedded memory chips

Looking forward, its becomes apparent that sensors with an embedded

memory chip are gaining more popularity. Advanced diagnostics offer undisputable advantages to the plant operation. The pH sensor that can carry
the factory calibration, serial number
and several diagnostics data can ease
the maintenance burden. First of all,
this smart sensor does not need buffer
calibration at the installation point. It
means that the service engineer does
not need to carry the buffers, beakers
and rinse water to the place where
the sensor is installed. It really makes
sense if you are in unfriendly environmental conditions, such as rain, snow
or blazing sun. The calibration can be
done in the laboratory and can be remotely or locally digitally accessible
after installation. Since the sensor
has an embedded factory calibration,
the instrument can always be reset to
the default calibration, without any
special efforts. Furthermore, several
diagnostic data sets and time stamps
stored on the memory chip make it
easy to see the trend of the diagnostics data (Table 1). Until recently, this
historization was possible only by
manual data logging.

Wireless transmitters

Wireless signal transmission is taking pH sensors even further. Having

advanced diagnostics stored inside
the sensor and being able to transmit
them wirelessly from the analyzer is a
big step forward. It eliminates the need
of wiring the transmitters to the main

control room. The pH measurement

can be wirelessly transmitted from
the most difficult and remote location
to a main computer. Sensor designs
can be tested, and data for in-process
samples can be communicated to the
control system without any wiring.
Sensors proven in laboratory tests can
be installed in the process without any
wiring. The sensors can also be readily moved, based on field results, to a
location that minimizes fluctuations
from imperfect mixing and processtransportation time delay. The sensor
health can be monitored remotely, and
the maintenance trips can be scheduled in accordance with diagnostics records, as shown in Figure 6. It can save
many dollars to the end user because
needless maintenance can be avoided
and catastrophic failure prevented.
Wireless control. A simple enhancement to the PID control algorithm
has been developed, which combined
with the WirelessHART communication rules can reduce the number of
communications by an order of magnitude, thereby dramatically increasing battery life [9]. The communication rules provide a transmission of
signal in the schedule time slot set
up by the network manager, only if
the difference between the new measured value and the last transmitted
value exceeds a sensitivity limit, or
the elapsed time since the last transmission exceeds a refresh time. The
transmitter continues to measure the
process at a fast rate but will transmit faster than the periodic reporting
set by the refresh time on an exception basis set by the sensitivity limit.
The optimum sensitivity setting
should be larger than the maximum
amplitude of measurement noise and
larger than the resolution of the control valve multiplied by the process
gain, but less than the minimum allowable deviation from setpoint [9].
The enhanced PID algorithm keeps
the controller execution fast, to provide an immediate response to setpoint changes and fast disturbances
via proportional feedback and feedforward control. The integral mode
is only executed when there is a new
value of the measurement. The integral contribution uses the external
reset-mechanism and dynamic-reset

Chemical Engineering www.che.com August 2010

37

Feature Report
limiting that has been found to be important for cascade control. The reset
time is set equal to the process time
constant, and the contribution to the
reset mode is based on the elapsed
time since the last updated measurement to provide a correction that
matches the time response of the process. The derivative mode only makes
a change in the controller output if
there is an update and computes the
rate of change based on the elapsed
time from the last update rather than
the execution time of the controller.
Thus, integral and derivative action
only occurs when there is a measurement update. Figure 7 shows how the
new communication rules and enhanced PID provided tight control of
a single-use-bioreactor batch, while
increasing battery life and ignoring a
spike from ground noise [9].
In control studies, the PID enhancements improved the stability for any
control system whenever there was a

Figure 8. The enhanced, wireless PID performance (bottom row) has value beyond wireless operation by providing tighter and more stable control of processes
with significant measurement delays from at-line analyzers. The Lambda factor is
used for computing controller gain, and is inversely proportional to it

significant measurement-time delay.

Figure 8 for a generic self-regulating
(continuous) process, dominated by a
large time delay, shows that the PID
enhancement eliminates oscillations.
This technological advancement in the
PID for all loops is a fall out of maxi-

mizing the utility of the new wireless

technology [9].The wireless communication and advanced diagnostics are
taking us into the future of not only
pH measurement and maintenance,
but also the overall approach of running the plant.

Ultrameter II 6Psi

bluDock Enabled
TM

NEW! bluDock Wireless

Data Transfer (optional)

NEW! LSI and Hardness

Calculator

Circle 19 on p. 50 or go to adlinks.che.com/29253-19
38

Chemical Engineering www.che.com August 2010

www.myronl.com

Authors

Embedded process models

control and inferential measurement of

pH as a third signal for middle signal
selection. An application of an adapted
embedded model for a RCRA (Resource
Conservation and Recovery Act) waste
neutralization system in a chemical
intermediates plant demonstrated the
concept and provided reagent savings of
$100,000/yr [10]. In both cases, the key
to model fidelity was matching the process gain (titration curve slope) of the
model to the plant.


Edited by Dorothy Lozowski

References

Processing, April 2008.

6. McMillan, G.K. and Cameron, R.A., Advanced
pH Measurement and Control, 3rd Ed., ISA,
Research Triangle Park, N.C. (2005).
7. Shinskey, F.G., pH and pION Control in Process
and Waste Streams, John Wiley & Sons, 1973.
8. McMillan, G.K., Reedy, R.D., and Moulis, J.P.,
Virtual Plant Provides Real Insights, Chem.
Process., January, 2009.
9. McMillan, G.K., What have I Learned
Unexpected Wireless Benefits, February 9,
2009, http://www.modelingandcontrol.com/
10. McMillan, G. K. and Sowell, M.S., Virtual Control of Real pH, Control, November, 2007.

Dynamic models of the pH system will

be embedded with prototypes of the
control system in the DCS to provide a
more efficient process design. In a recent
example, the capital cost was reduced
by 50% for neutralization of the regeneration effluent from a demineralized
water system [9]. These embedded models will also be put online and adapted
to provide better signal characterization and inferential measurements of
influent composition for feedforward

1. H. Galster, pH Measurement Fundamentals,

Methods, Applications, Instrumentation,
VCH Verlagsgesellschaft mbH, Weinheim,
1991.
2. C. C. Westcott, pH Measurements, Academic Press Inc., 1978.
3. C. D. Feng, Aging mechanism of pH sensing
glass in high temperature environment, Proceedings of the ISA 53rd Analysis Div. Symposium, April 2008.
4. Moore, Ralph, Good pH Measurements in
Bad Process Streams, Instrum. Control Syst.
(I&CS), December, 1990.
5. N. Starostin, New trend in pH-technology,

Stop wasting
money on
rupture discs.

Gregory McMillan is a principal consultant to Emerson

Process Managements Process Systems and Solutions
Group (Research Park Plaza,
Bldg. 3, 12301 Research Blvd.,
Austin, TX, 78759; Email:
Greg.McMillan@Emerson.
com). McMillan is a retired
senior fellow from Solutia/
Monsanto and an ISA Fellow.
He received the ISA Kermit
Fischer Environmental Award for pH control
in 1991, was inducted into the Control Process
Automation Hall of Fame in 2001 and received
the ISA Life Achievement Award in 2010. He has
a B.S. in engineering physics from Kansas University and M.S. in electrical engineering from
Missouri University of Science and Technology.
McMillan is the author of numerous books on
process control, the most recent being Essentials
of Modern Measurements and Final Elements
for the Process Industry. His expertise is summarized on his Website: www.modelingandcontrol.com/
Richard Baril is the product marketing manager for
Emerson Process Management, Rosemount Analytical (2400 Barranca Parkway,
Irvine, CA, 92696; Phone:
(8008548257; Email: [email protected]). He
has been active in the drinking water and wastewater
process industries for more
than 20 years. Baril holds a
B.S. in Chemistry from Long Beach State and
an MBA from the University of La Verne. He is
a current member of the International Ozone
Assn., has spoken at several industry conferences, and has authored numerous articles in a
variety of publications.

Get the right rupture disc

for your application.
Call the experts at Fike.
s#OMPREHENSIVESELECTIONOF!3-%#%AND !CERTIlEDDISCS
s-EETOREXCEEDGLOBALINDUSTRYREQUIREMENTS
s#OMPLETELINEOFPRODUCTSDESIGNEDFOR#)03)0APPLICATIONS
s$EDICATEDTEAMOFENGINEERSANDTECHNICALSUPPORT

Solutions to Fit your Facility and your Budget.

PRV/SRV
PROTECTION

SANITARY
RUPTURE DISCS

FULL-SCALE
TESTING

Proven Quality and Reliability

for over 60 Years
1-866-758-6004

WWW.FIKE.COM

Circle 10 on p. 50 or go to adlinks.che.com/29253-10
Chemical Engineering www.che.com August 2010

39