PROCESS CONTROL SYSTEMS 3B (PCSB301)

TOPIC 1 NOTES - PROCESS DYNAMICS AND IDENTIFICATION

N.B Prescribed/Recommended Books
Instrument Engineers Handbook (Vol 2): Process Control and Optimization
Programmable Logic Controllers W. Bolton
Programmable Logic Controllers F. Petruzella
Control Guru - Practical Process Control
PCSB NOTES

TOPIC 1 - PROCESS DYNAMICS AND IDENTIFICATION

Refer to Chapter 2.22
 TOPIC 1 NOTES - PROCESS DYNAMICS AND IDENTIFICATION
• Refer to Chapter 2.22.
• Additional resource website (https://controlguru.com/)

WHY PROCESS CONTROL IS IMPORTANT

Process control lies on safety and profit, reductional wastage and working with operational constraints
so that quality control measures are achieved.
➢ Safety - environment, man and equipment, the safety of the plant personnel and the people in the
community are the highest priority, thus human safety must be a priority.
❖ In an industrial facility there may be hundreds or thousands of controllers operating at the plant and,
these controllers must operate in a safe manner which is a manner controlled within certain constraints
so that the environment is not impacted negatively, the personnel operating the equipment or the
equipment itself.
❖ Good process control safety is of primary importance when designing a control system all the elements
surrounding that control system have to be looked at, the hardware and that is where the process control
instrumentation come into play.
❖ The process instrumentation is designed and figures out how all the systems are put together, so that it
can operate as expected to the desired set point, and this gives a safe operating control system.
➢ Profit – Meeting final product specifications, minimizing waste product, minimizing environmental
impact, minimizing energy use and improve efficiencies.
WHAT DO YOU NEED TO KNOW ABOUT PROCESS MONITORING AND CONTROL FROM A PRACTICAL
PERSPECTIVE
1. Different Types of Processes – Linear type process, Non-Linear Type process, 1st Order, 2nd Order,
Integrating Non-Minimal Phase type process.
2. How To Model the Process Signatures or Dynamics – transfer functions are used to explain process
model signatures or the dynamics of it; and the transient or changing nature of dynamics of the system.
3. PID Tuning Methodologies
4. Evaluating The Control Performance of The Closed Loop Control – this is done by first order model
dynamics; the control is designed and that is the tuning part. The final part is evaluating the performance
of the controller, how well the controller is performing.

TERMINOLOGY OF CONTROL
• Final Control Element (FCE) – Such as valve, pump or compressor.
• Control Output (CO) – Controller output signal from a sensor
• Set point (SP) – is a target value that a control system aims to maintain.
• Process Variable (PV) - is the current measured value of a particular part of a process which is being
monitored or controlled.
• Disturbances (D) - any unwanted inputs or external factors that affect the system’s output and
performance
• Manipulated Variable (MV) - This a variable that we can control in the process and directly affects the
output of the process.

Page | 1
PROCESS DYNAMICS AND IDENTIFICATION
➢ Plot below shows a poorly controlled process can exhibit large variability in a measured process variable.
➢ Suppose in this example the PV must not exceed a maximum value. So, the closer we can run to this
operating constraint, the greater our profit.
➢ To ensure that our operating constraint is not exceeded, the operator-specified set point, must be set
far from the constraint to ensure it is never violated.
➢ This is a process reaction, which is a steady state reaction; the set point has not changed.
➢ Operating constraint (Red Line), the process must not go beyond the operating constraint.
Example
- if this is a tank level shown below, there is a set point of 50%, the level must not go beyond 60%, if it
goes beyond 60% the system must turn on an alarm or go through some process, maybe turn on an
interlock as a safety mechanism built into the controller so that it does not continue to violet these
constraints.
This is an example of a poor control system, there is oscillation occurring this can be due to several factors.

➢ Below we see the same process with improved control. There is a significantly less variability in the
measured PV, as a result, the SP can be moved closer to the operating constraint.
➢ Good control permits SP near the constraints, this generates more profit.
➢ There is less oscillation on this control, this is a tighter control, behind the process control the response
is tighter around the set point.
➢ When designing a control feedback system be it a feedback or feedforward, split range it must control
through the desired setpoint.
➢ On this system, if there is a violation of the operating constraint there is loss in the profit, possibly there
may be spillage, depending on the nature of the process in hand.
➢ The wide range of the process variable must be kept within the bandwidth around the setpoint.
➢ The set point in this case is Static, as it is standing across the time frame.
➢ In some systems the setpoint will change.

Page | 2
PROCESS DYNAMICS AND IDENTIFICATION
 Page 8

UNDERSTANDING THE COMPONENTS OF A CONTROL LOOP (CLASSICAL NEGATIVE

FEEDBACK CONTROL SCHEME)

▪ Output at the controller goes to the Final Control Element, the Final Control Element is coupled to a
pipeline, which forms a part of the process.
▪ FCE and the Process are lumped together
▪ Disturbance signal is acting on the output of the process
▪ If there is no disturbance, there is no need for Automated Control.
▪ The valve can be set in one position, it will remain in that position through out.
▪ Feedback negates the disturbance, as it is measuring the output of the process as well as the disturbance
then it corrects via the controller.
▪ The disturbance can be reduced by correctly designing a PID and tune it.

I. Cruise Control and Measuring Our PV (Control Guru)

Cruise control in a car is a reasonably common intermediate value control system.
The control system is enabled with a button on the car instrument panel. Once on the open road and at
the desired cruising speed, a second button is pressed, and it switches the controller from manual mode
(where car speed is adjusted by the foot) to automatic mode (where car speed is adjusted by the
controller).

Page | 3
PROCESS DYNAMICS AND IDENTIFICATION
 The speed of the car at the moment the loop is closed and switch from manual to automatic mode
becomes the set point. The controller then continuously computes and transmits corrective actions to
the gas pedal (throttle) to maintain measured speed at set point.
Cruise control serves to illustrate this idea. Actual car speed is challenging to measure. But transmission
rotational speed can be measured reliably and inexpensively. The transmission connects the engine to
the wheels, so as it spins faster or slower, the car speed directly increases or decreases.
Thus, a small magnet is attached to the rotating output shaft of the car transmission and a magnetic field
detector (loops of wire and a simple circuit) to the body of the car above the magnet. With each rotation,
the magnet passes by the detector and the event is registered by the circuitry as a “click.” As the drive
shaft spins faster or slower, the click rates and car speed increase or decrease proportionally.

So, a cruise control system really adjusts fuel flow rate to maintain click rate at the set point value. With
this knowledge, the cruise control can be organized into the essential design elements:
▪ Set Point (SP): The desired speed of the car is set by the driver. This is the target speed that the system
aims to maintain.
▪ Controller Error: The error signal is calculated as the difference between the Set Point (SP) and the
Process Variable (PV), which is the measured car speed.
▪ Cruise Controller: The cruise controller receives the error signal and processes it to generate a Control
Output (CO). This output is a signal sent to the gas pedal control element to adjust the car's speed.
▪ Gas Pedal Control Element: This element receives the Control Output from the Cruise Controller and
adjusts the gas flow rate to the car engine accordingly.
▪ Car Speed Process: The manipulated gas flow rate affects the car's engine, which in turn influences the
car's speed. The actual car speed is the Process Variable (PV).
▪ Disturbances: External factors such as hills, wind, etc., can affect the car's speed. These disturbances are
considered in the system to ensure the cruise control can compensate for them.
▪ Magnet and Coil Sensor: This sensor measures the actual car speed and sends the measured Process
Variable (PV) signal back to the controller to close the feedback loop.

Page | 4
PROCESS DYNAMICS AND IDENTIFICATION
II. Home Heating Control Loop.
The desired temperature is set by the user.
The Controller calculates the Error, which is the difference between the Set Point (SP) and The Actual
Temperature (PV). The Controller then sends a signal to adjust the Furnace Fuel Valve.
The Furnace Fuel Valve controls the amount of fuel going into the furnace, affecting the heat produced.
The furnace heats the home, and The Temperature Sensor measures The Actual Temperature (PV). This
value is fed back to the controller to adjust as needed.
The system continuously adjusts to maintain the desired temperature despite disturbances like heat loss.

III. Water Level Control System

The user sets the desired water level (setpoint).
The sensor measures the current water level and sends this information to the controller.
The controller calculates the error and adjusts the valve to correct the water level.
The system continuously monitors and adjusts the water level to match the setpoint, compensating for
any disturbances.
➢ This control loop ensures that the water level remains stable and at the desired setpoint despite any
external or internal disturbances

1. Automatic Controller:
▪ On/off, PID, model based, Fuzzy logic.
▪ Many other types of control algorithms

2. Process Dynamic Behaviour

▪ Self-Regulating Systems – system of F.C.E and process, these systems are systems which when a
controller sends a signal to FCE, the process will respond, and the response comes to a steady state on
its own.
❖ The system is upset by changing the position of the control valve, and the process will respond to those
changes after some period of time it will reach steady state again; it will go to a transient state then to
the steady state.
❖ This type of strategy in terms of feedback is only applicable to systems where the disturbance is not
dominant; the disturbance acts now and again, but it is not a major influence on the control strategy.
❖ Systems affected predominantly by disturbances need feedforward strategy to negate the disturbances
before it imparts itself on the control loop.

Page | 5
PROCESS DYNAMICS AND IDENTIFICATION
❖ This is a Single Input, Single Output (SISO) system, there is one setpoint and one output.
a) First Order Plus Dead Time (FOPDT) – Self Regulating
b) Second Order Plus Dead Time (SOPDT) - Self Regulating
▪ Integrating (1st Or 2nd Order With/Without Dead Time) – these systems are not self-regulating which
means U(s) is changed, the process increases indefinitely and does not come to rest.

3. Sensor:
▪ Temperature, pressure, level, flow, etc.

4. Disturbance:
▪ External Load – some disturbances are deterministic (can be step, ramp, or sinusoidal), load disturbance
can be a change in setpoint.
❖ Anything affecting the controller is a disturbance, when there is a change in setpoint, the controller takes
that as a disturbance, as it will impact the controller output.
❖ Anything imparting itself externally like setpoint or disturbance at Y(s) can be in the form of a step change
(changes all of a sudden), ramp (slowly increases, then stops) or in the form of sinusoidal.
▪ Noise (stochastic, wide range of frequencies) – Stochastic is a randomized noise, it imparts itself on the
output signal, it can come from the process.
Example
- If there is a tank, with a level transmitter on the side the pipe feeding the water, the water will splash,
the level will be noisy.

MODELLING SYSTEM DYNAMICS FROM 1ST PRINCIPLES.

▪ This is done by analysing the process from 1st principles.
▪ This can be done by having a complete theoretical understanding of the energy balance surrounding the
equipment or distillation column or boiler drum.
▪ In order to model the tank level, the derivative of this tank level is a function of the area of the tank, the
amount of liquid flowing in minus the amount of liquid going out, depending on the area of the pipes
(inlet and outlet).
▪ To work out the system dynamics; equations have to be obtained from 1 st principles.

Page | 6
PROCESS DYNAMICS AND IDENTIFICATION
GRAPHICAL METHOD
▪ This is the easiest method; the graphical method entails having the loop in open mode.
▪ A signal is sent to the valve, the valve is pumped or stepped up.
▪ The process will react, the level will rise, the signal is captured.
▪ Transfer function is output over input.
▪ Depending on the nature of the output function relative to input function, the dynamic nature of the
system can be worked out.

Steps Involved:
• Put the automatic controller in manual mode. Ensure controlled variable is not in transient state - Must
be open loop not a closed loop because this will prevent the feedback signal to affect the controller; only
the output of the controller is desired to affect the process or capture the dynamics of the process
• Step the output (force change on the F.C.E) - the change must be enough to excite the system.
• Controller output (u) and process output (y) is recorded -
• Determine the model parameters.

THE RECIPE FOR SUCCESS IS SHORT:

1. Establish the design level of operation (DLO), defined as the expected value for set point and major
disturbances during normal operation.
2. Bump the process and collect controller output (CO) to process variable (PV) dynamic process data
around this design level.
3. Approximate the process data behaviour with a first order plus dead time (FOPDT) dynamic model.
4. Use the model parameters from step 3 in rules and correlations to complete the controller design and
tuning
𝑲𝑷 : Process steady state gain.

𝝉𝑷 : Process time constant.

𝜽𝑷 : Process dead time or lag time.
Fit the parameters to a FOPDT model (Laplace Transform).

Page | 7
PROCESS DYNAMICS AND IDENTIFICATION
 Finally, the respective controller is tuned based on the model for either disturbance rejection or setpoint
tracking.
This only applies to self-regulating system; the model will change for other systems and second order
systems.
The model above is a first order plus deadtime, it is a function of the process, which is a summarized
form of the output over the input.
This is known as a 3-term transfer function, where there are 3 variables.
There is a onetime constant, that is why it is a 1st order.

- The 3 tanks will have 1st order dynamics, if they are all lumped together there
will be 3-time constants, which represent a 3-order system.
- This 3rd order system can be represented as one tank.
- The idea is to take a transient response and fit it into the model shown
above.

The controller can be tuned to track a setpoint as shown in Here, however the setpoint is changing.
The controller can be used for the disturbance rejection, for Load Disturbance Rejection the setpoint is
kept constant; in disturbance rejection the disturbances are accounted for in the controller.
In Setpoint Tracking the setpoint is not constant, the focus is not on the disturbances.

FOPDT Parameters
➢ FOPDT parameters tell us important information about the measured process variable (PV) behavior
whenever there is a change in the controller output (CO) signal:
𝑲𝒑 Process gain, tells the direction and how far the PV will travel
𝝉𝒑 Process time constant, tells how fast PV moves after it begins its response.
𝜽𝒑 Process dead time, tells how much delay before PV first begins to respond

IMPORTANCE OF THESE PARAMETERS

▪ Tuning – They can be plugged into a proved correlation to directly compute P-only, PI, PID, and PID with
CO filter tuning values.
▪ Controller Action – Determine if the controller is reverse acting or direct acting or other used terms like
up-up or up-down (as CO goes up, the PV goes down)
▪ Loop sample time
▪ Dead time problems
▪ Model Based Control

Page | 8
PROCESS DYNAMICS AND IDENTIFICATION
IMPORTANT PRACTICAL TIPS ON THE FOPDT MODEL:
➢ sign (+ or -) of KP indicates “reverse acting” or “direct acting”, example of tank level control.
➢ Example: there may be some systems where the valve is at the top, when a positive signal is put on the
valve, the level will rise.

➢ Then there are systems where the valve is at the bottom, when an increasing signal is put on the valve,
the level will decrease.

𝜽𝑷 the process deadtime is when a step change is made and there is time that elapses before the process
respond, this has a direct impact on the controllability ratio which is the deadtime over the time constant.
𝜽𝑷
ratio (controllability ratio) indicates the level of control difficulty. If the ratio is greater than 1, then
𝝉𝑷
the process is generally considered dead time dominant.
Deadtime dominant processes are hard to achieve good control, PID controller cannot be used.
Deadtime compensating techniques can be employed like the smith predictor, which is designed for
deadtime dominant systems.
When the deadtime is equal to the time constant, it will be 1 and it will be difficult to control because
the deadtime is large and impacts the controller.
If the deadtime is greater than the time constant (which relates to when the signal is put on the system),
it will take time for the level to respond and increase.
Time constant is the time taken for the system to reach a steady state.
𝝉𝑷 gives an indication of the loop sample time.

When doing graphical analysis to determine transfer function, the loop sample has to be considered.
Sampling time (𝝉𝒔 ) must not be less than or equal 0.1 which is the 1st time constant.
There must be two samples to capture the dynamics of the system, this is Nyquist Theorem.
Rule of thumb can be used to prevent anti-aliasing.

Page | 9
PROCESS DYNAMICS AND IDENTIFICATION
Example 1 Of Graphical Process Modelling (Gravity Drain Tank)

▪ The liquid exits the outlet on its own cord.

▪ Valve on the top opens and close based on the user input, then the level is
recorded.
▪ The controller output goes from 50% to 60%
▪ Level recorded is given in meters (1.93) to (2.88).

➢ To extract the 3 parameters for the 3-term 1st order plus deadtime, change the input (∆u) and change
the output (∆y).
➢ This is self-regulating system.
Process Gain: Change In Output Over Change In Input

➢ 0.095 m/% implies that when there is a sizable input to the valve the level will not change that much.
➢ Large gain will have a small input change (∆u) which will give large level derivation.
Process Time Constant
𝒕𝟔𝟑.𝟐 is the time at 63.2%
𝒕𝒚𝒔𝒕𝒂𝒓𝒕 is the starting time

➢ Process time constant is from where the process starts changing to where it ends in terms of reaching its
steady state, the equilibrium point.
➢ This is broken into 4-time constants, but only the first-time constant is used, which is 63.2% of the total
deviation.

Page | 10
PROCESS DYNAMICS AND IDENTIFICATION
 ▪ The time when the process reached 63.2%.
▪ This is done by determining the 63.2 in terms of process variable which is the
level at 63.2%.

1.93 m starting point which is 63.2% of the change in y.

Process Dead Time
➢ Process deadtime is determined when the process started minus the time when the input signal took
place.
➢ The process started to react at 9.6 minutes and the input signal took place at 9.25 minutes.

Process Controllability Ratio

➢ The process is considered to be easy to control since the controllability ratio is less than 1.
➢ It is not deadtime dominant.
Gravity Drained Tank FOPDT Model
➢ Has a very small gain

➢ 0.35 can be represented using deadtime approximation

➢ This gives the sensitivity of the system, the deadtime and how quickly the system responds to input
changes.
➢ Tells the controllability ratio, how well or easy the system can be controlled.

Page | 11
PROCESS DYNAMICS AND IDENTIFICATION
Example Of Graphical Process Modelling (Heat Exchanger).
➢ On this example the process liquid is cooled to suitable temperature.
➢ There will be a decrease in temperature approximately 150 ℃ to 140 ℃.
➢ The gain will be negative; input to the system starts from 25% and goes to 35%.
➢ This is a self-regulating system; it does not continue or go into instability.

Process Gain:

Process Time Constant

➢ This is a stable system, its neither a low gain system or a high gain system.
➢ The system close to 1 will have better controllability ratio in terms of closing or opening the valve.
➢ It will be less susceptible to any changes imparted on the valve.

Page | 12
PROCESS DYNAMICS AND IDENTIFICATION
Process Dead Time:

Process Controllability Ratio

The process is considered easy to control since controllability ratio is less than 1.
Heat Exchanger FOPDT Model:

When the valve is opened the temperature will decrease.

More Examples Here
LIMITATIONS OF THE STEP (BUMP) TEST
▪ Graphical method may be inaccurate if readings are not clear due to process noise or disturbances.
▪ Process must be in a steady state to obtain usable responses.
▪ Sometimes loops are not permitted to be manual mode.
▪ New setpoint may create off-spec products.
▪ Use of alternative output signals, i.e. PRBS, Pulse and doublet. (Each has its own distinct advantages and
disadvantages).
▪ A pulse can be sent directly to the output.

FOPDT MODEL NOT SUITABLE FOR

▪ Under damped process behavior.
▪ Inverse response. (Non-minimum phase)
▪ Integrating systems.
▪ Non-linear behavior

Under Damped Response

There are 3 systems Underdamped, which is oscillating type of system; this is a 2nd order characteristic.
Critically Damped system which is a 1st order response; this system is required to use a graphical method.
Overdamped System, the valve has not travelled sufficiently to excite the system well.
Page | 13
PROCESS DYNAMICS AND IDENTIFICATION
Non-Minimum Phase Response

➢ It goes to the negative region, then to the positive region; this is exhibited by boiler level controller, in
boiler drums the level controllers are there because of the boiler shrink and swell effects.
➢ When the valve to the boiler drums opens, the level will drop, after a certain time it will increase

Integrating Response
➢ There is step change (in blue) the process increases this is an integrating since it does not come to rest;
it is not self-regulating.
➢ To work out this dynamic, change of increase has to be determined.
➢ 1st order model cannot be used here.

Page | 14
PROCESS DYNAMICS AND IDENTIFICATION
Integrating Process Response

➢ When the flowrate coming into the tank is greater than the flow going out of the tank, the level will
increase until it overflows; it does not come to steady state equilibrium point.
➢ It is an integrating process because there is time constant.

Non-Linear Response

➢ The system is changed over a certain range from 140 to 150.

➢ Initially the process is Sluggish, it responds on lower end of the scale, as it gets higher in terms of
temperature, is it excited; thus, it is non-linear behaviour where throughout the entire range of the
process it gives different characteristics.
➢ There will be 3 separate transfer functions, and these transfer functions are different and unique to the
operating points of the heat exchanger.
➢ To get a system to work like this, the 3 transfer functions models have to be determined and define the
controllers to work out these parameters at these systems, this is PID Gain Scheduling.

Page | 15
PROCESS DYNAMICS AND IDENTIFICATION
COMBINATION OF NON-LINEAR AND NON-MINIMUM PHASE

➢ Integrating with second order, combination of non-linear with integrating or non-linear with non-
minimum phase.
➢ The system reacts sluggish at the bottom end of the scale, there is non minimum phase; it is negative
then, positive just after 20 seconds, then it gets excited more in an oscillation after that.

RESPONSE SHAPE OF FIRST ORDER AND SECOND ORDER MODELS

To approximate 2nd order using FOPDT by converting the time constants to a 1st order. The fist order 𝝉𝑭𝑶 is
proportional to the second order 𝝉𝑺𝑶 constant plus 0.5 of which ever is smaller.
Relating 2nd order time constants to fit 1st order time constants can be done by:

Page | 16
PROCESS DYNAMICS AND IDENTIFICATION
Example:

The controller can be worked out using this; the gain and time constant remain the same, only the 2 nd
order or 1st order changed.

PRACTICAL TIPS TO REMEMBER WHEN OBTAINING DATA FOR MODELING DYNAMICS

Eliminating all possible hardware errors from loop – transient process is required which is not hampered
by any sort of noise or any sort of disturbance signals, when doing the test, the hardware must be under
normal operating conditions. Example, sensor faults and noise sources
Have an idea what type of system you are dealing with, i.e. self-regulating, integrating.
Process must be at steady state before capturing data - this is because the starting point cannot be
found.
Test dynamics must dominate process noise (SNR>10) – SNR signal to noise ratio must be greater than
10
Load disturbances must not affect the test – load disturbances are an effect to external disturbances.
Step change must be adequate to response of the system – if it is a nonlinear system a doublet must be
used or PRB signal software.

Page | 17
PROCESS DYNAMICS AND IDENTIFICATION