back offs gives a modified operating point whose economics
may be computed. Bahri et al. developed a dynamic operability
framework for operability assessment and process
synthesis based on the back-off optimization formulation for
both linear and nonlinear of steady-state and open-loop
dynamic processes.150–152 Both uncertainties and disturbances
are considered in this method. The objective is to maximize
the process economy, subject to the feasible regulatory
dynamics. Therefore, an economic penalty is determined by
the distance between the steady-state optimum and the
dynamic operating point, which are calculated based on nonlinear
steady-state and nonlinear dynamics models, respectively.
The ideals were further developed by Figueroa
et al.153 where a recovery factor was defined as the ratio of
the amount of penalty recovered with control to the penalty
with no control. The back-off approach was then applied to
a variable structure control case.154 One feature of the backoff
approaches is that they determine the cost increase associated
with moving to the back-off position due to uncertainties
and disturbance. The limitation of this approach is that it
leads to conservative design, because the framework considers
the worst-case uncertainty scenario, even though the
probability of the worst-case uncertainty may not be high.
Ekawati and Bahri155 presented the integration of the output
controllability index156 within the dynamic operability
framework to facilitate controllability and economic assessment
of process system design for regulatory cases. This
framework utilizes a geometric representation of the feasible
operating region. The approach is made simpler by replacing
multiple maximization problems in the inner level and the
inequalities in the outer-level with a single geometric operation
and equality constraint, respectively.
Optimization-Based Control Structure for Improving
Controllability. Perkins was the first to directly evaluate the
effect of process dynamics on process economic performances
within an optimization framework.157 An optimistic
bound on the disturbance rejection performance was provided
by the approach of Walsh and Perkins,158 who
replaced operating variables with an idealized controller to
assess the effect of time delays and bounded parametric
uncertainty on disturbance rejection capabilities when
employing optimal control. To deal with the controllability
issues on an economic level, Narraway and Perkins presented
a method for selecting the optimal control structure
of a process without designing the process controller after
determining the optimal steady-state.148,149 Perfect disturbances
are rejected by the control system and a linear dynamic
model of the process to formulate a mixed-integer linear programming
technique, where the integer variables indicate the
pairings between the manipulated and controlled variables.
This approach was subsequently modified by Kookos and
Perkins,159 whereby the control objectives were posed in
terms of economic penalties associated with the effect of disturbances
on key process variables, aiming to identify optimal
control structure selection for static output feedback
controllers. Seferlis and Grievink developed a method for
assessing alternative process designs and control structures
based on the economic potential and static controllability
characteristics and depicted the advantages of this method
by multiple reactive and separation steps with recycle.160 To
stabilize the open loop unstable process with the minimum
control effect, a new method for control structure screening
based on improving branch and bound optimization was presented
by Cao and Saha,161 using the controllability index
(Hankel singular value, HSV). According to these methods,
the variations of each variable were used to estimate the
required back off for ensuring the feasibility, as well as to
estimate the change in process economics. The economic
analysis was carried out at the expected disturbance frequencies
and amplitudes, although stochastic disturbances were
not involved.
It is clear that sufficient attention has been given to the
complete and combined approaches of rigorous and systematic
screening of alternative process design with embedded
control structure characteristics based on control and economic
performance. The full count of all possible combinations
between potential manipulated and controlled variables
may become large, especially for plant-wide control system
design. Thus, the complete enumeration of all possible sets
of control structures for a number of disturbances incorporating
the dynamic behavior of the system within an optimization
framework would require great computational effort.
Optimization-Based Design and Control Simultaneously.
Modern chemical processes operate in a dynamic
environment, and are expected to handle variations in ambient
conditions and managers’ imposed demand on production.
The conventional design of first obtaining a plant configuration
and initial design based on steady-state economic
calculations and then using over-design factors to account
for variability based on controllability measures may prove
to be inadequate in today’s process design activities. So
simultaneously optimizing the process design and process
control strategy is a very active research area in the academic
world. Over the last decades, important efforts have
been aimed at providing methodologies for tackling process
design and control in an integrated framework, the control
configuration and controller parameters are optimized together
with the plant design parameters to determine the
optimal design and operating conditions of a process in this
integrated framework. A number of methodologies have
been proposed for solving integrated process design and controller
design (IPDC) problems.162,163 In these methodologies,
a mixed-integer nonlinear optimization problem
(MINLP) is formulated and solved with standard MINLP
solvers. When solving this optimization problem, the reconciling
conflicting design and control objectives will be
required. When a MINLP problem represents an IPDC, the
process model considers only steady state conditions.
Although a mixed-integer dynamic optimization (MIDO)
problem represents an IPDC where steady state as well as
dynamic behavior are considered. A substantial algorithm
have been developed to solve the MIDO problem,164–174
from an optimization point of view, these approaches can be
divided into simultaneous and sequential methods. Due to
the computational complexity associated with the resulting
nonlinear dynamic optimization problems, applying these
methodologies to large processes is restricted. In order to alleviate
some of the intensive computational burden associated
with dynamic optimization, in recent years, several
novel approaches have been proposed. A novel decomposition
method to solve the IPDC formulated as a mathematical
programming problem is presented, and the optimization
back offs gives a modified operating point whose economics
may be computed. Bahri et al. developed a dynamic operability
framework for operability assessment and process
synthesis based on the back-off optimization formulation for
both linear and nonlinear of steady-state and open-loop
dynamic processes.150–152 Both uncertainties and disturbances
are considered in this method. The objective is to maximize
the process economy, subject to the feasible regulatory
dynamics. Therefore, an economic penalty is determined by
the distance between the steady-state optimum and the
dynamic operating point, which are calculated based on nonlinear
steady-state and nonlinear dynamics models, respectively.
The ideals were further developed by Figueroa
et al.153 where a recovery factor was defined as the ratio of
the amount of penalty recovered with control to the penalty
with no control. The back-off approach was then applied to
a variable structure control case.154 One feature of the backoff
approaches is that they determine the cost increase associated
with moving to the back-off position due to uncertainties
and disturbance. The limitation of this approach is that it
leads to conservative design, because the framework considers
the worst-case uncertainty scenario, even though the
probability of the worst-case uncertainty may not be high.
Ekawati and Bahri155 presented the integration of the output
controllability index156 within the dynamic operability
framework to facilitate controllability and economic assessment
of process system design for regulatory cases. This
framework utilizes a geometric representation of the feasible
operating region. The approach is made simpler by replacing
multiple maximization problems in the inner level and the
inequalities in the outer-level with a single geometric operation
and equality constraint, respectively.
Optimization-Based Control Structure for Improving
Controllability. Perkins was the first to directly evaluate the
effect of process dynamics on process economic performances
within an optimization framework.157 An optimistic
bound on the disturbance rejection performance was provided
by the approach of Walsh and Perkins,158 who
replaced operating variables with an idealized controller to
assess the effect of time delays and bounded parametric
uncertainty on disturbance rejection capabilities when
employing optimal control. To deal with the controllability
issues on an economic level, Narraway and Perkins presented
a method for selecting the optimal control structure
of a process without designing the process controller after
determining the optimal steady-state.148,149 Perfect disturbances
are rejected by the control system and a linear dynamic
model of the process to formulate a mixed-integer linear programming
technique, where the integer variables indicate the
pairings between the manipulated and controlled variables.
This approach was subsequently modified by Kookos and
Perkins,159 whereby the control objectives were posed in
terms of economic penalties associated with the effect of disturbances
on key process variables, aiming to identify optimal
control structure selection for static output feedback
controllers. Seferlis and Grievink developed a method for
assessing alternative process designs and control structures
based on the economic potential and static controllability
characteristics and depicted the advantages of this method
by multiple reactive and separation steps with recycle.160 To
stabilize the open loop unstable process with the minimum
control effect, a new method for control structure screening
based on improving branch and bound optimization was presented
by Cao and Saha,161 using the controllability index
(Hankel singular value, HSV). According to these methods,
the variations of each variable were used to estimate the
required back off for ensuring the feasibility, as well as to
estimate the change in process economics. The economic
analysis was carried out at the expected disturbance frequencies
and amplitudes, although stochastic disturbances were
not involved.
It is clear that sufficient attention has been given to the
complete and combined approaches of rigorous and systematic
screening of alternative process design with embedded
control structure characteristics based on control and economic
performance. The full count of all possible combinations
between potential manipulated and controlled variables
may become large, especially for plant-wide control system
design. Thus, the complete enumeration of all possible sets
of control structures for a number of disturbances incorporating
the dynamic behavior of the system within an optimization
framework would require great computational effort.
Optimization-Based Design and Control Simultaneously.
Modern chemical processes operate in a dynamic
environment, and are expected to handle variations in ambient
conditions and managers’ imposed demand on production.
The conventional design of first obtaining a plant configuration
and initial design based on steady-state economic
calculations and then using over-design factors to account
for variability based on controllability measures may prove
to be inadequate in today’s process design activities. So
simultaneously optimizing the process design and process
control strategy is a very active research area in the academic
world. Over the last decades, important efforts have
been aimed at providing methodologies for tackling process
design and control in an integrated framework, the control
configuration and controller parameters are optimized together
with the plant design parameters to determine the
optimal design and operating conditions of a process in this
integrated framework. A number of methodologies have
been proposed for solving integrated process design and controller
design (IPDC) problems.162,163 In these methodologies,
a mixed-integer nonlinear optimization problem
(MINLP) is formulated and solved with standard MINLP
solvers. When solving this optimization problem, the reconciling
conflicting design and control objectives will be
required. When a MINLP problem represents an IPDC, the
process model considers only steady state conditions.
Although a mixed-integer dynamic optimization (MIDO)
problem represents an IPDC where steady state as well as
dynamic behavior are considered. A substantial algorithm
have been developed to solve the MIDO problem,164–174
from an optimization point of view, these approaches can be
divided into simultaneous and sequential methods. Due to
the computational complexity associated with the resulting
nonlinear dynamic optimization problems, applying these
methodologies to large processes is restricted. In order to alleviate
some of the intensive computational burden associated
with dynamic optimization, in recent years, several
novel approaches have been proposed. A novel decomposition
method to solve the IPDC formulated as a mathematical
programming problem is presented, and the optimization
การแปล กรุณารอสักครู่..