2.

6 Application of Temperature Control in Unit Operation

Temperature is one of the most significant measurements in the process industry

due to its ability to be an indicator of process stream composition and product quality.
Temperature can be defined as the measure of the activity inside the internal molecular of
the material. The temperature of the material increase proportionally to the molecular
activity. When there are increasing in molecular activity, the temperature will also increase.
[Advanced Temperature Measurement and Control]. The performance of process
equipment such as a chemical reactor, distillation column, and the heat exchanger highly
depends on the temperature controller to make sure it achieved high efficiency.

2.6.1 Reactor’s Cooling and Heating Control System

Temperature is one of the main process variables inside the reactor beside
pressure, flow, and level. The reaction rate and the selectivity of the reactor highly depend
on the temperature of the reaction. Therefore, it is important to have the heating and
cooling control system to maintain a specific temperature inside the reactor. The
exothermic reaction releases heat into the environment meanwhile endothermic reaction
needs heat energy to be put into the reaction [MCAT Review: Pearls of wisdom]. Hence,
heating is necessary for endothermic reactions while cooling is essential for an exothermic
reaction. Figure and Figure are the examples for jacket temperature control.

Figure 2.4: Jacket inlet temperature control

 Cooling and heating can be used directly to the reactor by heat transfer surfaces
in the vessel exterior or vessel interior. The example for vessel exterior is jacket system,
while for vessel interior is coil system. Coil temperature control give the system with faster
heating and cooling process, but for jacket temperature control it helps the heat transfer
distribute more evenly.

Figure 2.5: Jacket outlet temperature control

2.6.2 Distillation Column

Temperature control inside the distillation column is a method and system for
controlling the operation of a distillation column by adding temperature measured at points
within the column selected to accurately reflect the column’s performance and using the
temperature thus determined to properly adjust control variable. The proper selection of
temperature to be added allows maintenance of product qualities despite changes in feed
composition. Typically, in a single feed, two product distillation column temperature will be
measured 20% of the distance from the top to the feed tray.

The basic function of a simple distillation column is to separate a feed stream into
two or more product streams. The typical distillation column is characterized by a higher
temperature at the lower vessel and lower temperature at upper vessel with constant
pressure. The temperature is the reflection of the composition inside the vessel, therefore
the installation of temperature control is a must at various points throughout the column.

However, the temperature is not a genuine measurement in determining the

composition of multicomponent mixtures but is a right measurement for a binary mixture.
There is the limit usage of temperature in the multicomponent mixture which it must within
a narrow range of operating condition. Although a number of different compositions could
provide the same average boiling point, the temperature at any point within a column may
be expected to indicate the composition actually present. If the temperature is nearly
normal, a temperature change will indicate a changing composition which will require
correction, usually by changing the reflux rate. Usually, the temperature is measured at a
certain point in the vessel and the rate flow of reboiler and reflux is manipulated to adjust
the temperature.

The location of temperature measurement is important and complex in multi-

component distillation. Even the operation condition changes with time, the product quality
must be the same. To measure the composition inside the column, the temperature control
must be placed at the upper and bottom of the column.

2.6.2.1 Feed Temperature System

Figure 2. Block diagram of a simple feedback system.

 The uncertainties of temperature in the column can be a major factor in interfering
with good composition control. One source of heat disturbance is the column feed.
Temperature control in the feed system of the distillation column is necessary to make
sure the equipment operate at minimum energy consumption and to ensure the feed is
sent to the right feed tray. Figure …. Shows a system with a temperature control that helps
the feed temperature remain constant by throttling steam flow. This will fix the feed
enthalpy if the feed is no partially vaporized

2.6.2.2 Composition Measurement

The temperature controller is the most common technique to estimate the

composition on a single tray in the column. Theoretically, the temperature at the end of
the column should be controlled in a binary constant pressure system to maintain constant
product composition. However, the temperature change at the ends of the column is quite
small in moderate to high purity columns.

Therefore, a small change in pressure or the presence of other lighter or heavier

component can affect temperature more than the composition of the key component. The
relationship between tray temperature and the manipulated variable can be quite non-
linear if a tray near the end of the column is used. For good response, the temperature
control should be installed in the active part of a tray.

2.6.2.3 Heat exchanger

Many industrial process and operations heat exchanger is one of the simplest and
important unit for the transfer of thermal energy. There are different types of heat
exchanger used in industries, the shell and tube exchanger system being the common.
The main purpose of heat exchanger is to maintain specific temperature condition, which
is achieved by controlling the exit temperature of one of the fluids in response to variations
of the operating conditions. The temperature control of heat exchanger is nonlinear, time
varying and time delay system. For these situation, nonlinear control strategies can
provide significant improvements over PID control. Heat exchanger need to respond to
highly nonlinear features and work well under different operating point.

The control of temperature in a shell and tube heat exchanger is demonstrated in

figure 1. With cold water flowing on the tube side and steam on the shell side where steam
condenses and heat the water in the tubes. The controlled variable is the tube-side outlet
temperature, and the manipulated variable is the steam flow rate on the shell tube.

The dynamics of the process are complex because nonlinearities into the system.
The installed valve characteristic of the steam may not be linear. Dead time depends on
the steam and water flow rates, the location and the method of installation of the
temperature measuring device. To take into account the non-linearity and the dead time
(temperature control of shell and tube heat exchanger by using intelligent controller -case
study)

Figure

General heat transfer and exchanger system is used in chemical as well as

process, brewellary and different types of industry. Because it can sustain variety ranges
of temperature and pressure. The main purpose of a heat exchanger is to transfer heat
from a hot fluid to a cooler fluid, so temperature control of outlet fluid is of prime
importance.

2.6.2.4 Catalytic cracking unit

The catalytic cracking unit is mainly composed of one furnace, one riser reactor,
one settler, one regenerator and one fractionating tower. The temperature of crude oil is
the most significant feature in the cracking oil since it affects the thermal balance and the
depth of reaction. The preheated temperature is always varying in time, it will produce an
instability to the process condition, therefore it necessary to controlled it probably. The
control target of this temperature control system is to make sure the temperature of crude
oil in the preheated process is always at the set point.

In figure __ , it show the overall details of the preheated system. Basically, the
crude oil is flows into the pipe and then mix with the recycle oil. Then the product of the
mixture will go through the preheated process in the furnace to meet the desired
temperature. Consequently, the product from furnace go to the bottom of the riser to mix
with the catalyst that present in the regenerator.

Figure

2.6.2.5 Furnace Temperature Control System

In metallurgical industries furnace is the most important thermal equipment used.

Some basic method has been recognized for automatic control of furnace temperature.
For example, constant control of furnace process variable, cascade control of temperature
in the furnace and flow of the fuel, the control of fuel and combustion air mixture ratio and
also control of oxygen content in flue gas. So far, the control focus has moved to
performance optimization. The gas flow control primarily regulates the furnace
temperature. The dynamic quality of the entire system will be improved by pre-regulating
the interference which influences the intermediates variable. The effectiveness of control
of cascade control system is more efficient than single loop control system. And it has
advantages over single loop in anti-jamming capability, rapidity, flexibility and quality
control.

The constituent of a furnace temperature control system is shown in figure 1.

Whereas T1C is primary controller, T2C is a secondary controller. T1T represents
measure temperature for the exports of raw material, and T2t represents measure
temperature of the furnace heart. The basic operation of the furnace temperature control
system is as follow

Figure:

The output of the primary controller is given as set point to the secondary controller.
Control the fuel flow. In the heating process, material which is place in the crucible is
heated up to a specified temperature from the entrance to the exit. From fuel combustion
chamber to the raw export, there are three capacity components in system temperature.
They are furnace, hearth and the heated raw materials. Disturbances of load and in raw
material side are the system disturbance in one side and on the other side are the
disturbances in the burning side.

In the furnace temperature control system, the primary object is the raw material
and the secondary object is the hearth in the furnace. The main controlled variable is the
temperature of raw material in the export as sub co variable is the temperature of the heath
in the furnace. The primary disturbance is flow of the raw material and temperature of
material at the inlet. The secondary disturbances are the pressure of the fuel before the
valve, the heat value of fuel-air mixture supplied to furnace hearth.

2.6.2.6 Temperature control of CSTR by manipulation of fresh feed.

Reactor capacity is limited by heat removal, and the most recommended way to
tackle the issue is by maximize the coolant flow, and manipulate the feed flow rate to
control reactor temperature. If the feed temperature is lower than the reactor temperature,
the transfer function between the temperature ant feed flow rate contains a positive error,
which can degrade dynamic performance.

Most of the CSTR control studies have considered system in which heat is
transferred to or from the reacting mass in the reactor vessel to some cooling or heating
medium. Jacket cooling, internal coil cooling. The flow rate of the cooling or heating
medium is usually the manipulated variable that is changed by a reactor temperature
controller. Either directly or through reactor or jacket temperature cascade structure.

A less commonly encountered control structure is to use feed flow rate to control
reactor temperature. This kind of control structure can maximize the production rate when
the reactor capacity is limited by heat transfer. The temperature of the feed to a CSTR is
believed have a minor impact on the steady state design because it only affects the
sensible heat of the component which is insignificant compared to the heat produce from
the heat of reaction. Since the application of feed temperature of CSTR is insignificant,
therefore the temperature of feed CSTR is set up same as the stream supply temperature

However, if consider from dynamic point of view, temperature at the feed can significantly
affect the controllability of the reactor. For instance, if the feed has low temperature
compared to the operating condition inside the reactor, by increasing the flow rate it can
decrease the reactor temperature large. However, this method is only applicable for
reactor that designed for high levels of conversion and also have low hat of reaction.

This method is exceptional for reactor that designed for high levels of conversion.
When the conversion is increase, the available heat-transfer area also will be increase.
Consequently, the temperature of the jacket will be higher than the reactor.

Higher temperature of the jacket, will reduce the differences value between jacket
temperature and reactor temperature and then it will help the system to compensate the
disturbance. This means better dynamic control. Figure ___ show the closed loop
response for reactor with difference value of conversion. From the figure, it can be
observed that lower conversion give very small controller gains and large peak deviation,
and longer settling time.

2.6.2.7 Temperature control for reactive distillation column

Reactive distillation column usually used in chemical reaction that involved two
type reactants and two type products, which there is no excess of reactant in the reaction.
Usually to measure the composition inside the column, composition analyser is used.
However due some constrains such as the expensive price, and high maintenance, it is
preferably to use temperature measurement to predict the composition inside the column.

There is various type of control structure being used in chemical industry in order
to control the reactive distillation columns. One of the conventional method is by controlling
two tray temperature in the column by manipulating two feed stream. Figure 1 is the
example of control system which used two temperature controllers to manipulate the feed
flow rate. By manipulating the feed flow rate, it can help the system to maintain the
temperature inside the reactive distillation column.

Figure

2.6.2.8 Temperature Control of the BTX Divided-Wall Column

The mechanism to control a divided-wall column is more complex than

conventional separation especially when it related with separation of three type of mixture.
This is because the divided-wall column has more interaction among control loops. The
divided-wall column is a crucial instance of process intensification since it has the ability
to compensate higher energy consumption and reduce the capital cost for some systems.
 One of the features divided-wall column is, it has four control degree of freedom
which is reflux R, vapor boilup V, side stream S and liquid split BL. All four of this degree
of freedom is used to control the purities of three product stream and control the energy
consumption in the column.

Figure

There are two common method to control the purities of ternary system inside the
divided-wall column which is by using temperature control measurement and composition
analyser. However, composition analyser is not practical for small industry since it required
huge capital cost to purchase. Temperature measurement seems more applicable since
it low cost, reliable and provide fast responses. Unfortunately, the temperature
measurement only has an ability to estimate the composition
 A fluidized catalyticcrackingunit(FCCU)isoneofthemost complexprocessunitsinrefinery [1,2].
IntheFCCU,thepreheated temperatureofcrudeoilisavitalindependentvariablebecauseit affects
thethermalbalanceandthedepthofthereaction.The preheatedtemperature fluctuation
isassociatedwiththein- stability oftheoperatingconditionsandthusshouldbeprecisely controlled.
Moreover,thedynamicsofthistemperatureloopis
complexwithlargetimedelayanduncertainties,whichmaypose difficulty forPIDcontrol [3–5].
WhenthePIDcontrollerisusedto control thepreheatedtemperature,itisdifficult toachievedesir- able
performance [6–8]. As pointedoutbyseveralauthors,traditionalPIDcontrollers may
notbeeffectiveindealingwithprocesswithlargedeadtime to achievethedesirableperformance [9–13].
MajhproposedaPI- PD controlmethodtodealwiththisissue [14]. Infact,thiscon- troller
istheimprovementoftheconventionalPIDcontroller.The specific
wayofimprovementisthataPDcontrollerintheinner feedback alterstheopen-
looppolesandaPIcontrollerintheouter loop guaranteesanoverallsystemresponse [15]. Someresults
haveshownthatPI-PDcontrolcanindeedimprovecontrolper- formance comparedwithPIDcontrol [16].
Tanproposedagra- phical methodtocomputetheparametersofPI-PDcontroller [17].
However,thismethodcanonlygettheestimatedvaluesofPI-PD controllerparametersandtheparametersare
fixed,whicharenot. adaptivetothechangeofsystemoperationconditions.Tsaiand
TsaiputforwardanewapproachtoadjusttheparametersofPI-PD controller [18],
however,itmustobtaintheparametersofPID controller firstly andregulateanadditional β1 to get the para-
meters of PI-PD controller, which are complex to be calculated. Bettou and Charef [19] came up with a
−αPIPD controller [20], however,itposestheissuesthat five parametersshouldbecal- culatedforgettingthePI-
PDcontrolparameters.Theabovefacts show thatthePI-PDcontrollerhasbroadprospectbutislimitedby the
difficulty inparametertuning. EveniftheparametersofPI-PDcontrollerscanbeacquired
throughtheabovemethods,thelargetimedelayanduncertainties of
chemicalprocessesarestillthemajorproblemsforPI-PD [21,22].
Predictivefunctionalcontrol(PFC)issuitableforcoping with largetimedelaysanduncertainties [23–25].
Moreover,itdoes not requirehighcomputationalcapabilityandcanbeim- plemented inreal-
timeconditionsusinglow-costhardware [26– 29]. Thus,combiningPFCalgorithmwithPI-PDmethodmaybea
goodchoice. In thisstudy,anewPI-PDcontroldesignusingpredictive functional
control(PFC)isproposed.Theproposedmethodnot
onlyinheritstheadvantagesofPFC,butalsohasthesamestruc- ture asPI-
PDcontroller.Thepreheatedtemperaturecontrolof crude oilisconsideredasacasestudy. This
paperisorganizedasfollows.In Section 2, typicalPIDand PI-PD controllertuningmethodsaredescribed.In
Section 3, the proposedPI-PDmethodisintroduced.In Section 4, thesuperiority

A fluidized catalyticcrackingunit(FCCU)isoneofthemost complexprocessunitsinrefinery [1,2].

IntheFCCU,thepreheated temperatureofcrudeoilisavitalindependentvariablebecauseit affects
thethermalbalanceandthedepthofthereaction.The preheatedtemperature fluctuation
isassociatedwiththein- stability oftheoperatingconditionsandthusshouldbeprecisely controlled.
Moreover,thedynamicsofthistemperatureloopis
complexwithlargetimedelayanduncertainties,whichmaypose difficulty forPIDcontrol [3–5].
WhenthePIDcontrollerisusedto control thepreheatedtemperature,itisdifficult toachievedesir- able
performance [6–8]. As pointedoutbyseveralauthors,traditionalPIDcontrollers may
notbeeffectiveindealingwithprocesswithlargedeadtime to achievethedesirableperformance [9–13].
MajhproposedaPI- PD controlmethodtodealwiththisissue [14]. Infact,thiscon- troller
istheimprovementoftheconventionalPIDcontroller.The specific
wayofimprovementisthataPDcontrollerintheinner feedback alterstheopen-
looppolesandaPIcontrollerintheouter loop guaranteesanoverallsystemresponse [15]. Someresults
haveshownthatPI-PDcontrolcanindeedimprovecontrolper- formance comparedwithPIDcontrol [16].
Tanproposedagra- phical methodtocomputetheparametersofPI-PDcontroller [17].
However,thismethodcanonlygettheestimatedvaluesofPI-PD controllerparametersandtheparametersare
fixed,whicharenot. adaptivetothechangeofsystemoperationconditions.Tsaiand
TsaiputforwardanewapproachtoadjusttheparametersofPI-PD controller [18],
however,itmustobtaintheparametersofPID controller firstly andregulateanadditional β1 to get the para-
meters of PI-PD controller, which are complex to be calculated. Bettou and Charef [19] came up with a
−αPIPD controller [20], however,itposestheissuesthat five parametersshouldbecal- culatedforgettingthePI-
PDcontrolparameters.Theabovefacts show thatthePI-PDcontrollerhasbroadprospectbutislimitedby the
difficulty inparametertuning. EveniftheparametersofPI-PDcontrollerscanbeacquired
throughtheabovemethods,thelargetimedelayanduncertainties of
chemicalprocessesarestillthemajorproblemsforPI-PD [21,22].
Predictivefunctionalcontrol(PFC)issuitableforcoping with largetimedelaysanduncertainties [23–25].
Moreover,itdoes not requirehighcomputationalcapabilityandcanbeim- plemented inreal-
timeconditionsusinglow-costhardware [26– 29]. Thus,combiningPFCalgorithmwithPI-PDmethodmaybea
goodchoice. In thisstudy,anewPI-PDcontroldesignusingpredictive functional
control(PFC)isproposed.Theproposedmethodnot
onlyinheritstheadvantagesofPFC,butalsohasthesamestruc- ture asPI-
PDcontroller.Thepreheatedtemperaturecontrolof crude oilisconsideredasacasestudy. This
paperisorganizedasfollows.In Section 2, typicalPIDand PI-PD controllertuningmethodsaredescribed.In
Section 3, the proposedPI-PDmethodisintroduced.In Section 4, thesuperiority

A fluidized catalyticcrackingunit(FCCU)isoneofthemost complexprocessunitsinrefinery [1,2].

IntheFCCU,thepreheated temperatureofcrudeoilisavitalindependentvariablebecauseit affects
thethermalbalanceandthedepthofthereaction.The preheatedtemperature fluctuation
isassociatedwiththein- stability oftheoperatingconditionsandthusshouldbeprecisely controlled.
Moreover,thedynamicsofthistemperatureloopis
complexwithlargetimedelayanduncertainties,whichmaypose difficulty forPIDcontrol [3–5].
WhenthePIDcontrollerisusedto control thepreheatedtemperature,itisdifficult toachievedesir- able
performance [6–8]. As pointedoutbyseveralauthors,traditionalPIDcontrollers may
notbeeffectiveindealingwithprocesswithlargedeadtime to achievethedesirableperformance [9–13].
MajhproposedaPI- PD controlmethodtodealwiththisissue [14]. Infact,thiscon- troller
istheimprovementoftheconventionalPIDcontroller.The specific
wayofimprovementisthataPDcontrollerintheinner feedback alterstheopen-
looppolesandaPIcontrollerintheouter loop guaranteesanoverallsystemresponse [15]. Someresults
haveshownthatPI-PDcontrolcanindeedimprovecontrolper- formance comparedwithPIDcontrol [16].
Tanproposedagra- phical methodtocomputetheparametersofPI-PDcontroller [17].
However,thismethodcanonlygettheestimatedvaluesofPI-PD controllerparametersandtheparametersare
fixed,whicharenot. adaptivetothechangeofsystemoperationconditions.Tsaiand
TsaiputforwardanewapproachtoadjusttheparametersofPI-PD controller [18],
however,itmustobtaintheparametersofPID controller firstly andregulateanadditional β1 to get the para-
meters of PI-PD controller, which are complex to be calculated. Bettou and Charef [19] came up with a
−αPIPD controller [20], however,itposestheissuesthat five parametersshouldbecal- culatedforgettingthePI-
PDcontrolparameters.Theabovefacts show thatthePI-PDcontrollerhasbroadprospectbutislimitedby the
difficulty inparametertuning. EveniftheparametersofPI-PDcontrollerscanbeacquired
throughtheabovemethods,thelargetimedelayanduncertainties of
chemicalprocessesarestillthemajorproblemsforPI-PD [21,22].
Predictivefunctionalcontrol(PFC)issuitableforcoping with largetimedelaysanduncertainties [23–25].
Moreover,itdoes not requirehighcomputationalcapabilityandcanbeim- plemented inreal-
timeconditionsusinglow-costhardware [26– 29]. Thus,combiningPFCalgorithmwithPI-PDmethodmaybea
goodchoice. In thisstudy,anewPI-PDcontroldesignusingpredictive functional
control(PFC)isproposed.Theproposedmethodnot
onlyinheritstheadvantagesofPFC,butalsohasthesamestruc- ture asPI-
PDcontroller.Thepreheatedtemperaturecontrolof crude oilisconsideredasacasestudy. This
paperisorganizedasfollows.In Section 2, typicalPIDand PI-PD controllertuningmethodsaredescribed.In
Section 3, the proposedPI-PDmethodisintroduced.In Section 4, thesuperiority

A fluidized catalyticcrackingunit(FCCU)isoneofthemost complexprocessunitsinrefinery [1,2].

IntheFCCU,thepreheated temperatureofcrudeoilisavitalindependentvariablebecauseit affects
thethermalbalanceandthedepthofthereaction.The preheatedtemperature fluctuation
isassociatedwiththein- stability oftheoperatingconditionsandthusshouldbeprecisely controlled.
Moreover,thedynamicsofthistemperatureloopis
complexwithlargetimedelayanduncertainties,whichmaypose difficulty forPIDcontrol [3–5].
WhenthePIDcontrollerisusedto control thepreheatedtemperature,itisdifficult toachievedesir- able
performance [6–8]. As pointedoutbyseveralauthors,traditionalPIDcontrollers may
notbeeffectiveindealingwithprocesswithlargedeadtime to achievethedesirableperformance [9–13].
MajhproposedaPI- PD controlmethodtodealwiththisissue [14]. Infact,thiscon- troller
istheimprovementoftheconventionalPIDcontroller.The specific
wayofimprovementisthataPDcontrollerintheinner feedback alterstheopen-
looppolesandaPIcontrollerintheouter loop guaranteesanoverallsystemresponse [15]. Someresults
haveshownthatPI-PDcontrolcanindeedimprovecontrolper- formance comparedwithPIDcontrol [16].
Tanproposedagra- phical methodtocomputetheparametersofPI-PDcontroller [17].
However,thismethodcanonlygettheestimatedvaluesofPI-PD controllerparametersandtheparametersare
fixed,whicharenot. adaptivetothechangeofsystemoperationconditions.Tsaiand
TsaiputforwardanewapproachtoadjusttheparametersofPI-PD controller [18],
however,itmustobtaintheparametersofPID controller firstly andregulateanadditional β1 to get the para-
meters of PI-PD controller, which are complex to be calculated. Bettou and Charef [19] came up with a
−αPIPD controller [20], however,itposestheissuesthat five parametersshouldbecal- culatedforgettingthePI-
PDcontrolparameters.Theabovefacts show thatthePI-PDcontrollerhasbroadprospectbutislimitedby the
difficulty inparametertuning. EveniftheparametersofPI-PDcontrollerscanbeacquired
throughtheabovemethods,thelargetimedelayanduncertainties of
chemicalprocessesarestillthemajorproblemsforPI-PD [21,22].
Predictivefunctionalcontrol(PFC)issuitableforcoping with largetimedelaysanduncertainties [23–25].
Moreover,itdoes not requirehighcomputationalcapabilityandcanbeim- plemented inreal-
timeconditionsusinglow-costhardware [26– 29]. Thus,combiningPFCalgorithmwithPI-PDmethodmaybea
goodchoice. In thisstudy,anewPI-PDcontroldesignusingpredictive functional
control(PFC)isproposed.Theproposedmethodnot
onlyinheritstheadvantagesofPFC,butalsohasthesamestruc- ture asPI-
PDcontroller.Thepreheatedtemperaturecontrolof crude oilisconsideredasacasestudy. This
paperisorganizedasfollows.In Section 2, typicalPIDand PI-PD controllertuningmethodsaredescribed.In
Section 3, the proposedPI-PDmethodisintroduced.In Section 4, thesuperiority

A fluidized catalyticcrackingunit(FCCU)isoneofthemost complexprocessunitsinrefinery [1,2].

IntheFCCU,thepreheated temperatureofcrudeoilisavitalindependentvariablebecauseit affects
thethermalbalanceandthedepthofthereaction.The preheatedtemperature fluctuation
isassociatedwiththein- stability oftheoperatingconditionsandthusshouldbeprecisely controlled.
Moreover,thedynamicsofthistemperatureloopis
complexwithlargetimedelayanduncertainties,whichmaypose difficulty forPIDcontrol [3–5].
WhenthePIDcontrollerisusedto control thepreheatedtemperature,itisdifficult toachievedesir- able
performance [6–8]. As pointedoutbyseveralauthors,traditionalPIDcontrollers may
notbeeffectiveindealingwithprocesswithlargedeadtime to achievethedesirableperformance [9–13].
MajhproposedaPI- PD controlmethodtodealwiththisissue [14]. Infact,thiscon- troller
istheimprovementoftheconventionalPIDcontroller.The specific
wayofimprovementisthataPDcontrollerintheinner feedback alterstheopen-
looppolesandaPIcontrollerintheouter loop guaranteesanoverallsystemresponse [15]. Someresults
haveshownthatPI-PDcontrolcanindeedimprovecontrolper- formance comparedwithPIDcontrol [16].
Tanproposedagra- phical methodtocomputetheparametersofPI-PDcontroller [17].
However,thismethodcanonlygettheestimatedvaluesofPI-PD controllerparametersandtheparametersare
fixed,whicharenot. adaptivetothechangeofsystemoperationconditions.Tsaiand
TsaiputforwardanewapproachtoadjusttheparametersofPI-PD controller [18],
however,itmustobtaintheparametersofPID controller firstly andregulateanadditional β1 to get the para-
meters of PI-PD controller, which are complex to be calculated. Bettou and Charef [19] came up with a
−αPIPD controller [20], however,itposestheissuesthat five parametersshouldbecal- culatedforgettingthePI-
PDcontrolparameters.Theabovefacts show thatthePI-PDcontrollerhasbroadprospectbutislimitedby the
difficulty inparametertuning. EveniftheparametersofPI-PDcontrollerscanbeacquired
throughtheabovemethods,thelargetimedelayanduncertainties of
chemicalprocessesarestillthemajorproblemsforPI-PD [21,22].
Predictivefunctionalcontrol(PFC)issuitableforcoping with largetimedelaysanduncertainties [23–25].
Moreover,itdoes not requirehighcomputationalcapabilityandcanbeim- plemented inreal-
timeconditionsusinglow-costhardware [26– 29]. Thus,combiningPFCalgorithmwithPI-PDmethodmaybea
goodchoice. In thisstudy,anewPI-PDcontroldesignusingpredictive functional
control(PFC)isproposed.Theproposedmethodnot
onlyinheritstheadvantagesofPFC,butalsohasthesamestruc- ture asPI-
PDcontroller.Thepreheatedtemperaturecontrolof crude oilisconsideredasacasestudy. This
paperisorganizedasfollows.In Section 2, typicalPIDand PI-PD controllertuningmethodsaredescribed.In
Section 3, the proposedPI-PDmethodisintroduced.In Section 4, thesuperiority

A fluidized catalyticcrackingunit(FCCU)isoneofthemost complexprocessunitsinrefinery [1,2].

IntheFCCU,thepreheated temperatureofcrudeoilisavitalindependentvariablebecauseit affects
thethermalbalanceandthedepthofthereaction.The preheatedtemperature fluctuation
isassociatedwiththein- stability oftheoperatingconditionsandthusshouldbeprecisely controlled.
Moreover,thedynamicsofthistemperatureloopis
complexwithlargetimedelayanduncertainties,whichmaypose difficulty forPIDcontrol [3–5].
WhenthePIDcontrollerisusedto control thepreheatedtemperature,itisdifficult toachievedesir- able
performance [6–8]. As pointedoutbyseveralauthors,traditionalPIDcontrollers may
notbeeffectiveindealingwithprocesswithlargedeadtime to achievethedesirableperformance [9–13].
MajhproposedaPI- PD controlmethodtodealwiththisissue [14]. Infact,thiscon- troller
istheimprovementoftheconventionalPIDcontroller.The specific
wayofimprovementisthataPDcontrollerintheinner feedback alterstheopen-
looppolesandaPIcontrollerintheouter loop guaranteesanoverallsystemresponse [15]. Someresults
haveshownthatPI-PDcontrolcanindeedimprovecontrolper- formance comparedwithPIDcontrol [16].
Tanproposedagra- phical methodtocomputetheparametersofPI-PDcontroller [17].
However,thismethodcanonlygettheestimatedvaluesofPI-PD controllerparametersandtheparametersare
fixed,whicharenot. adaptivetothechangeofsystemoperationconditions.Tsaiand
TsaiputforwardanewapproachtoadjusttheparametersofPI-PD controller [18],
however,itmustobtaintheparametersofPID controller firstly andregulateanadditional β1 to get the para-
meters of PI-PD controller, which are complex to be calculated. Bettou and Charef [19] came up with a
−αPIPD controller [20], however,itposestheissuesthat five parametersshouldbecal- culatedforgettingthePI-
PDcontrolparameters.Theabovefacts show thatthePI-PDcontrollerhasbroadprospectbutislimitedby the
difficulty inparametertuning. EveniftheparametersofPI-PDcontrollerscanbeacquired
throughtheabovemethods,thelargetimedelayanduncertainties of
chemicalprocessesarestillthemajorproblemsforPI-PD [21,22].
Predictivefunctionalcontrol(PFC)issuitableforcoping with largetimedelaysanduncertainties [23–25].
Moreover,itdoes not requirehighcomputationalcapabilityandcanbeim- plemented inreal-
timeconditionsusinglow-costhardware [26– 29]. Thus,combiningPFCalgorithmwithPI-PDmethodmaybea
goodchoice. In thisstudy,anewPI-PDcontroldesignusingpredictive functional
control(PFC)isproposed.Theproposedmethodnot
onlyinheritstheadvantagesofPFC,butalsohasthesamestruc- ture asPI-
PDcontroller.Thepreheatedtemperaturecontrolof crude oilisconsideredasacasestudy. This
paperisorganizedasfollows.In Section 2, typicalPIDand PI-PD controllertuningmethodsaredescribed.In
Section 3, the proposedPI-PDmethodisintroduced.In Section 4, thesuperiority

A fluidized catalyticcrackingunit(FCCU)isoneofthemost complexprocessunitsinrefinery [1,2].

IntheFCCU,thepreheated temperatureofcrudeoilisavitalindependentvariablebecauseit affects
thethermalbalanceandthedepthofthereaction.The preheatedtemperature fluctuation
isassociatedwiththein- stability oftheoperatingconditionsandthusshouldbeprecisely controlled.
Moreover,thedynamicsofthistemperatureloopis
complexwithlargetimedelayanduncertainties,whichmaypose difficulty forPIDcontrol [3–5].
WhenthePIDcontrollerisusedto control thepreheatedtemperature,itisdifficult toachievedesir- able
performance [6–8]. As pointedoutbyseveralauthors,traditionalPIDcontrollers may
notbeeffectiveindealingwithprocesswithlargedeadtime to achievethedesirableperformance [9–13].
MajhproposedaPI- PD controlmethodtodealwiththisissue [14]. Infact,thiscon- troller
istheimprovementoftheconventionalPIDcontroller.The specific
wayofimprovementisthataPDcontrollerintheinner feedback alterstheopen-
looppolesandaPIcontrollerintheouter loop guaranteesanoverallsystemresponse [15]. Someresults
haveshownthatPI-PDcontrolcanindeedimprovecontrolper- formance comparedwithPIDcontrol [16].
Tanproposedagra- phical methodtocomputetheparametersofPI-PDcontroller [17].
However,thismethodcanonlygettheestimatedvaluesofPI-PD controllerparametersandtheparametersare
fixed,whicharenot. adaptivetothechangeofsystemoperationconditions.Tsaiand
TsaiputforwardanewapproachtoadjusttheparametersofPI-PD controller [18],
however,itmustobtaintheparametersofPID controller firstly andregulateanadditional β1 to get the para-
meters of PI-PD controller, which are complex to be calculated. Bettou and Charef [19] came up with a
−αPIPD controller [20], however,itposestheissuesthat five parametersshouldbecal- culatedforgettingthePI-
PDcontrolparameters.Theabovefacts show thatthePI-PDcontrollerhasbroadprospectbutislimitedby the
difficulty inparametertuning. EveniftheparametersofPI-PDcontrollerscanbeacquired
throughtheabovemethods,thelargetimedelayanduncertainties of
chemicalprocessesarestillthemajorproblemsforPI-PD [21,22].
Predictivefunctionalcontrol(PFC)issuitableforcoping with largetimedelaysanduncertainties [23–25].
Moreover,itdoes not requirehighcomputationalcapabilityandcanbeim- plemented inreal-
timeconditionsusinglow-costhardware [26– 29]. Thus,combiningPFCalgorithmwithPI-PDmethodmaybea
goodchoice. In thisstudy,anewPI-PDcontroldesignusingpredictive functional
control(PFC)isproposed.Theproposedmethodnot
onlyinheritstheadvantagesofPFC,butalsohasthesamestruc- ture asPI-
PDcontroller.Thepreheatedtemperaturecontrolof crude oilisconsideredasacasestudy. This
paperisorganizedasfollows.In Section 2, typicalPIDand PI-PD controllertuningmethodsaredescribed.In
Section 3, the proposedPI-PDmethodisintroduced.In Section 4, thesuperiority

A fluidized catalyticcrackingunit(FCCU)isoneofthemost complexprocessunitsinrefinery [1,2].

IntheFCCU,thepreheated temperatureofcrudeoilisavitalindependentvariablebecauseit affects
thethermalbalanceandthedepthofthereaction.The preheatedtemperature fluctuation
isassociatedwiththein- stability oftheoperatingconditionsandthusshouldbeprecisely controlled.
Moreover,thedynamicsofthistemperatureloopis
complexwithlargetimedelayanduncertainties,whichmaypose difficulty forPIDcontrol [3–5].
WhenthePIDcontrollerisusedto control thepreheatedtemperature,itisdifficult toachievedesir- able
performance [6–8]. As pointedoutbyseveralauthors,traditionalPIDcontrollers may
notbeeffectiveindealingwithprocesswithlargedeadtime to achievethedesirableperformance [9–13].
MajhproposedaPI- PD controlmethodtodealwiththisissue [14]. Infact,thiscon- troller
istheimprovementoftheconventionalPIDcontroller.The specific
wayofimprovementisthataPDcontrollerintheinner feedback alterstheopen-
looppolesandaPIcontrollerintheouter loop guaranteesanoverallsystemresponse [15]. Someresults
haveshownthatPI-PDcontrolcanindeedimprovecontrolper- formance comparedwithPIDcontrol [16].
Tanproposedagra- phical methodtocomputetheparametersofPI-PDcontroller [17].
However,thismethodcanonlygettheestimatedvaluesofPI-PD controllerparametersandtheparametersare
fixed,whicharenot. adaptivetothechangeofsystemoperationconditions.Tsaiand
TsaiputforwardanewapproachtoadjusttheparametersofPI-PD controller [18],
however,itmustobtaintheparametersofPID controller firstly andregulateanadditional β1 to get the para-
meters of PI-PD controller, which are complex to be calculated. Bettou and Charef [19] came up with a
−αPIPD controller [20], however,itposestheissuesthat five parametersshouldbecal- culatedforgettingthePI-
PDcontrolparameters.Theabovefacts show thatthePI-PDcontrollerhasbroadprospectbutislimitedby the
difficulty inparametertuning. EveniftheparametersofPI-PDcontrollerscanbeacquired
throughtheabovemethods,thelargetimedelayanduncertainties of
chemicalprocessesarestillthemajorproblemsforPI-PD [21,22].
Predictivefunctionalcontrol(PFC)issuitableforcoping with largetimedelaysanduncertainties [23–25].
Moreover,itdoes not requirehighcomputationalcapabilityandcanbeim- plemented inreal-
timeconditionsusinglow-costhardware [26– 29]. Thus,combiningPFCalgorithmwithPI-PDmethodmaybea
goodchoice. In thisstudy,anewPI-PDcontroldesignusingpredictive functional
control(PFC)isproposed.Theproposedmethodnot
onlyinheritstheadvantagesofPFC,butalsohasthesamestruc- ture asPI-
PDcontroller.Thepreheatedtemperaturecontrolof crude oilisconsideredasacasestudy. This
paperisorganizedasfollows.In Section 2, typicalPIDand PI-PD controllertuningmethodsaredescribed.In
Section 3, the proposedPI-PDmethodisintroduced.In Section 4, thesuperiority

A fluidized catalyticcrackingunit(FCCU)isoneofthemost complexprocessunitsinrefinery [1,2].

IntheFCCU,thepreheated temperatureofcrudeoilisavitalindependentvariablebecauseit affects
thethermalbalanceandthedepthofthereaction.The preheatedtemperature fluctuation
isassociatedwiththein- stability oftheoperatingconditionsandthusshouldbeprecisely controlled.
Moreover,thedynamicsofthistemperatureloopis
complexwithlargetimedelayanduncertainties,whichmaypose difficulty forPIDcontrol [3–5].
WhenthePIDcontrollerisusedto control thepreheatedtemperature,itisdifficult toachievedesir- able
performance [6–8]. As pointedoutbyseveralauthors,traditionalPIDcontrollers may
notbeeffectiveindealingwithprocesswithlargedeadtime to achievethedesirableperformance [9–13].
MajhproposedaPI- PD controlmethodtodealwiththisissue [14]. Infact,thiscon- troller
istheimprovementoftheconventionalPIDcontroller.The specific
wayofimprovementisthataPDcontrollerintheinner feedback alterstheopen-
looppolesandaPIcontrollerintheouter loop guaranteesanoverallsystemresponse [15]. Someresults
haveshownthatPI-PDcontrolcanindeedimprovecontrolper- formance comparedwithPIDcontrol [16].
Tanproposedagra- phical methodtocomputetheparametersofPI-PDcontroller [17].
However,thismethodcanonlygettheestimatedvaluesofPI-PD controllerparametersandtheparametersare
fixed,whicharenot. adaptivetothechangeofsystemoperationconditions.Tsaiand
TsaiputforwardanewapproachtoadjusttheparametersofPI-PD controller [18],
however,itmustobtaintheparametersofPID controller firstly andregulateanadditional β1 to get the para-
meters of PI-PD controller, which are complex to be calculated. Bettou and Charef [19] came up with a
−αPIPD controller [20], however,itposestheissuesthat five parametersshouldbecal- culatedforgettingthePI-
PDcontrolparameters.Theabovefacts show thatthePI-PDcontrollerhasbroadprospectbutislimitedby the
difficulty inparametertuning. EveniftheparametersofPI-PDcontrollerscanbeacquired
throughtheabovemethods,thelargetimedelayanduncertainties of
chemicalprocessesarestillthemajorproblemsforPI-PD [21,22].
Predictivefunctionalcontrol(PFC)issuitableforcoping with largetimedelaysanduncertainties [23–25].
Moreover,itdoes not requirehighcomputationalcapabilityandcanbeim- plemented inreal-
timeconditionsusinglow-costhardware [26– 29]. Thus,combiningPFCalgorithmwithPI-PDmethodmaybea
goodchoice. In thisstudy,anewPI-PDcontroldesignusingpredictive functional
control(PFC)isproposed.Theproposedmethodnot
onlyinheritstheadvantagesofPFC,butalsohasthesamestruc- ture asPI-
PDcontroller.Thepreheatedtemperaturecontrolof crude oilisconsideredasacasestudy. This
paperisorganizedasfollows.In Section 2, typicalPIDand PI-PD controllertuningmethodsaredescribed.In
Section 3, the proposedPI-PDmethodisintroduced.In Section 4, thesuperiority

A fluidized catalyticcrackingunit(FCCU)isoneofthemost complexprocessunitsinrefinery [1,2].

IntheFCCU,thepreheated temperatureofcrudeoilisavitalindependentvariablebecauseit affects
thethermalbalanceandthedepthofthereaction.The preheatedtemperature fluctuation
isassociatedwiththein- stability oftheoperatingconditionsandthusshouldbeprecisely controlled.
Moreover,thedynamicsofthistemperatureloopis
complexwithlargetimedelayanduncertainties,whichmaypose difficulty forPIDcontrol [3–5].
WhenthePIDcontrollerisusedto control thepreheatedtemperature,itisdifficult toachievedesir- able
performance [6–8]. As pointedoutbyseveralauthors,traditionalPIDcontrollers may
notbeeffectiveindealingwithprocesswithlargedeadtime to achievethedesirableperformance [9–13].
MajhproposedaPI- PD controlmethodtodealwiththisissue [14]. Infact,thiscon- troller
istheimprovementoftheconventionalPIDcontroller.The specific
wayofimprovementisthataPDcontrollerintheinner feedback alterstheopen-
looppolesandaPIcontrollerintheouter loop guaranteesanoverallsystemresponse [15]. Someresults
haveshownthatPI-PDcontrolcanindeedimprovecontrolper- formance comparedwithPIDcontrol [16].
Tanproposedagra- phical methodtocomputetheparametersofPI-PDcontroller [17].
However,thismethodcanonlygettheestimatedvaluesofPI-PD controllerparametersandtheparametersare
fixed,whicharenot. adaptivetothechangeofsystemoperationconditions.Tsaiand
TsaiputforwardanewapproachtoadjusttheparametersofPI-PD controller [18],
however,itmustobtaintheparametersofPID controller firstly andregulateanadditional β1 to get the para-
meters of PI-PD controller, which are complex to be calculated. Bettou and Charef [19] came up with a
−αPIPD controller [20], however,itposestheissuesthat five parametersshouldbecal- culatedforgettingthePI-
PDcontrolparameters.Theabovefacts show thatthePI-PDcontrollerhasbroadprospectbutislimitedby the
difficulty inparametertuning. EveniftheparametersofPI-PDcontrollerscanbeacquired
throughtheabovemethods,thelargetimedelayanduncertainties of
chemicalprocessesarestillthemajorproblemsforPI-PD [21,22].
Predictivefunctionalcontrol(PFC)issuitableforcoping with largetimedelaysanduncertainties [23–25].
Moreover,itdoes not requirehighcomputationalcapabilityandcanbeim- plemented inreal-
timeconditionsusinglow-costhardware [26– 29]. Thus,combiningPFCalgorithmwithPI-PDmethodmaybea
goodchoice. In thisstudy,anewPI-PDcontroldesignusingpredictive functional
control(PFC)isproposed.Theproposedmethodnot
onlyinheritstheadvantagesofPFC,butalsohasthesamestruc- ture asPI-
PDcontroller.Thepreheatedtemperaturecontrolof crude oilisconsideredasacasestudy. This
paperisorganizedasfollows.In Section 2, typicalPIDand PI-PD controllertuningmethodsaredescribed.In
Section 3, the proposedPI-PDmethodisintroduced.In Section 4, thesuperiority
 A fluidized catalyticcrackingunit(FCCU)isoneofthemost complexprocessunitsinrefinery [1,2].
IntheFCCU,thepreheated temperatureofcrudeoilisavitalindependentvariablebecauseit affects
thethermalbalanceandthedepthofthereaction.The preheatedtemperature fluctuation
isassociatedwiththein- stability oftheoperatingconditionsandthusshouldbeprecisely controlled.
Moreover,thedynamicsofthistemperatureloopis
complexwithlargetimedelayanduncertainties,whichmaypose difficulty forPIDcontrol [3–5].
WhenthePIDcontrollerisusedto control thepreheatedtemperature,itisdifficult toachievedesir- able
performance [6–8]. As pointedoutbyseveralauthors,traditionalPIDcontrollers may
notbeeffectiveindealingwithprocesswithlargedeadtime to achievethedesirableperformance [9–13].
MajhproposedaPI- PD controlmethodtodealwiththisissue [14]. Infact,thiscon- troller
istheimprovementoftheconventionalPIDcontroller.The specific
wayofimprovementisthataPDcontrollerintheinner feedback alterstheopen-
looppolesandaPIcontrollerintheouter loop guaranteesanoverallsystemresponse [15]. Someresults
haveshownthatPI-PDcontrolcanindeedimprovecontrolper- formance comparedwithPIDcontrol [16].
Tanproposedagra- phical methodtocomputetheparametersofPI-PDcontroller [17].
However,thismethodcanonlygettheestimatedvaluesofPI-PD controllerparametersandtheparametersare
fixed,whicharenot. adaptivetothechangeofsystemoperationconditions.Tsaiand
TsaiputforwardanewapproachtoadjusttheparametersofPI-PD controller [18],
however,itmustobtaintheparametersofPID controller firstly andregulateanadditional β1 to get the para-
meters of PI-PD controller, which are complex to be calculated. Bettou and Charef [19] came up with a
−αPIPD controller [20], however,itposestheissuesthat five parametersshouldbecal- culatedforgettingthePI-
PDcontrolparameters.Theabovefacts show thatthePI-PDcontrollerhasbroadprospectbutislimitedby the
difficulty inparametertuning. EveniftheparametersofPI-PDcontrollerscanbeacquired
throughtheabovemethods,thelargetimedelayanduncertainties of
chemicalprocessesarestillthemajorproblemsforPI-PD [21,22].
Predictivefunctionalcontrol(PFC)issuitableforcoping with largetimedelaysanduncertainties [23–25].
Moreover,itdoes not requirehighcomputationalcapabilityandcanbeim- plemented inreal-
timeconditionsusinglow-costhardware [26– 29]. Thus,combiningPFCalgorithmwithPI-PDmethodmaybea
goodchoice. In thisstudy,anewPI-PDcontroldesignusingpredictive functional
control(PFC)isproposed.Theproposedmethodnot
onlyinheritstheadvantagesofPFC,butalsohasthesamestruc- ture asPI-
PDcontroller.Thepreheatedtemperaturecontrolof crude oilisconsideredasacasestudy. This
paperisorganizedasfollows.In Section 2, typicalPIDand PI-PD controllertuningmethodsaredescribed.In
Section 3, the proposedPI-PDmethodisintroduced.In Section 4, thesuperiority
 A fluidized catalyticcrackingunit(FCCU)isoneofthemost complexprocessunitsinrefinery [1,2].
IntheFCCU,thepreheated temperatureofcrudeoilisavitalindependentvariablebecauseit affects
thethermalbalanceandthedepthofthereaction.The preheatedtemperature fluctuation
isassociatedwiththein- stability oftheoperatingconditionsandthusshouldbeprecisely controlled.
Moreover,thedynamicsofthistemperatureloopis
complexwithlargetimedelayanduncertainties,whichmaypose difficulty forPIDcontrol [3–5].
WhenthePIDcontrollerisusedto control thepreheatedtemperature,itisdifficult toachievedesir- able
performance [6–8]. As pointedoutbyseveralauthors,traditionalPIDcontrollers may
notbeeffectiveindealingwithprocesswithlargedeadtime to achievethedesirableperformance [9–13].
MajhproposedaPI- PD controlmethodtodealwiththisissue [14]. Infact,thiscon- troller
istheimprovementoftheconventionalPIDcontroller.The specific
wayofimprovementisthataPDcontrollerintheinner feedback alterstheopen-
looppolesandaPIcontrollerintheouter loop guaranteesanoverallsystemresponse [15]. Someresults
haveshownthatPI-PDcontrolcanindeedimprovecontrolper- formance comparedwithPIDcontrol [16].
Tanproposedagra- phical methodtocomputetheparametersofPI-PDcontroller [17].
However,thismethodcanonlygettheestimatedvaluesofPI-PD controllerparametersandtheparametersare
fixed,whicharenot. adaptivetothechangeofsystemoperationconditions.Tsaiand
TsaiputforwardanewapproachtoadjusttheparametersofPI-PD controller [18],
however,itmustobtaintheparametersofPID controller firstly andregulateanadditional β1 to get the para-
meters of PI-PD controller, which are complex to be calculated. Bettou and Charef [19] came up with a
−αPIPD controller [20], however,itposestheissuesthat five parametersshouldbecal- culatedforgettingthePI-
PDcontrolparameters.Theabovefacts show thatthePI-PDcontrollerhasbroadprospectbutislimitedby the
difficulty inparametertuning. EveniftheparametersofPI-PDcontrollerscanbeacquired
throughtheabovemethods,thelargetimedelayanduncertainties of
chemicalprocessesarestillthemajorproblemsforPI-PD [21,22].
Predictivefunctionalcontrol(PFC)issuitableforcoping with largetimedelaysanduncertainties [23–25].
Moreover,itdoes not requirehighcomputationalcapabilityandcanbeim- plemented inreal-
timeconditionsusinglow-costhardware [26– 29]. Thus,combiningPFCalgorithmwithPI-PDmethodmaybea
goodchoice. In thisstudy,anewPI-PDcontroldesignusingpredictive functional
control(PFC)isproposed.Theproposedmethodnot
onlyinheritstheadvantagesofPFC,butalsohasthesamestruc- ture asPI-
PDcontroller.Thepreheatedtemperaturecontrolof crude oilisconsideredasacasestudy. This
paperisorganizedasfollows.In Section 2, typicalPIDand PI-PD controllertuningmethodsaredescribed.In
Section 3, the proposedPI-PDmethodisintroduced.In Section 4, thesuperiority

A fluidized catalyticcrackingunit(FCCU)isoneofthemost complexprocessunitsinrefinery [1,2].

IntheFCCU,thepreheated temperatureofcrudeoilisavitalindependentvariablebecauseit affects
thethermalbalanceandthedepthofthereaction.The preheatedtemperature fluctuation
isassociatedwiththein- stability oftheoperatingconditionsandthusshouldbeprecisely controlled.
Moreover,thedynamicsofthistemperatureloopis
complexwithlargetimedelayanduncertainties,whichmaypose difficulty forPIDcontrol [3–5].
WhenthePIDcontrollerisusedto control thepreheatedtemperature,itisdifficult toachievedesir- able
performance [6–8]. As pointedoutbyseveralauthors,traditionalPIDcontrollers may
notbeeffectiveindealingwithprocesswithlargedeadtime to achievethedesirableperformance [9–13].
MajhproposedaPI- PD controlmethodtodealwiththisissue [14]. Infact,thiscon- troller
istheimprovementoftheconventionalPIDcontroller.The specific
wayofimprovementisthataPDcontrollerintheinner feedback alterstheopen-
looppolesandaPIcontrollerintheouter loop guaranteesanoverallsystemresponse [15]. Someresults
haveshownthatPI-PDcontrolcanindeedimprovecontrolper- formance comparedwithPIDcontrol [16].
Tanproposedagra- phical methodtocomputetheparametersofPI-PDcontroller [17].
However,thismethodcanonlygettheestimatedvaluesofPI-PD controllerparametersandtheparametersare
fixed,whicharenot. adaptivetothechangeofsystemoperationconditions.Tsaiand
TsaiputforwardanewapproachtoadjusttheparametersofPI-PD controller [18],
however,itmustobtaintheparametersofPID controller firstly andregulateanadditional β1 to get the para-
meters of PI-PD controller, which are complex to be calculated. Bettou and Charef [19] came up with a
−αPIPD controller [20], however,itposestheissuesthat five parametersshouldbecal- culatedforgettingthePI-
PDcontrolparameters.Theabovefacts show thatthePI-PDcontrollerhasbroadprospectbutislimitedby the
difficulty inparametertuning. EveniftheparametersofPI-PDcontrollerscanbeacquired
throughtheabovemethods,thelargetimedelayanduncertainties of
chemicalprocessesarestillthemajorproblemsforPI-PD [21,22].
Predictivefunctionalcontrol(PFC)issuitableforcoping with largetimedelaysanduncertainties [23–25].
Moreover,itdoes not requirehighcomputationalcapabilityandcanbeim- plemented inreal-
timeconditionsusinglow-costhardware [26– 29]. Thus,combiningPFCalgorithmwithPI-PDmethodmaybea
goodchoice. In thisstudy,anewPI-PDcontroldesignusingpredictive functional
control(PFC)isproposed.Theproposedmethodnot
onlyinheritstheadvantagesofPFC,butalsohasthesamestruc- ture asPI-
PDcontroller.Thepreheatedtemperaturecontrolof crude oilisconsideredasacasestudy. This
paperisorganizedasfollows.In Section 2, typicalPIDand PI-PD controllertuningmethodsaredescribed.In
Section 3, the proposedPI-PDmethodisintroduced.In Section 4, thesuperiority