反應/分離迴流程序之設計與控制(1/3)
Date Issued
2004-05-27
Date
2004-05-27
Author(s)
余政靖
DOI
922214E002021
Abstract
In this work we explore dynamics and control of recycle
plants with multiple feed streams under different process
designs. Without loss of generality, a ternary system with
second order irreversible reaction, A+B=2C, is used to
illustrate the dynamics and control of such multiple-feed
systems. Given design parameters, process transfer function
matrices can be derived analytically (design parameters are
expressed in terms of reactor composition, i.e., zA, zB, and zC
in the triangular composition space). Therefore, we are able
to evaluate recycle dynamics for different designs (e.g.,
different conversion, zC, and reactant distribution, zA/zB).
The results show that, irrespective of design parameters,
recycle plants with multiple feeds are internally unstable
system (a pole at the origin). The recycle plant can be
stabilized only if the feed flows are exactly the same at
steady-state. This proves Luyben’s conjecture that we have
to balance the reactants down to the last molecule. This also
implies feed-ratio control scheme will not work in practice
and one of the feed flows should be adjusted via feedback.
After eliminating unworkable control structures, remaining
candidate control structures are evaluated next. Globally, for
a given control structure, the bifurcation diagram is used to
identify unstable region (if exist) in the composition space
and, subsequently, a stability boundary can be drawn. Locally, the speed of recycle flows (pole location from
transfer function matrix) is used select the preferable one.
Finally, dynamical compensation to the feed flow is devised
according to the transfer function matrix such that almost
perfect production rate control can be achieved. A nonlinear
recycle process with one reactor and two columns
(programmed in FORTRAN) is used to illustrate the
usefulness of the proposed analysis. Simulation results
clearly indicate that correct control structure is selected and
almost perfect production rate changes can be achieved.
plants with multiple feed streams under different process
designs. Without loss of generality, a ternary system with
second order irreversible reaction, A+B=2C, is used to
illustrate the dynamics and control of such multiple-feed
systems. Given design parameters, process transfer function
matrices can be derived analytically (design parameters are
expressed in terms of reactor composition, i.e., zA, zB, and zC
in the triangular composition space). Therefore, we are able
to evaluate recycle dynamics for different designs (e.g.,
different conversion, zC, and reactant distribution, zA/zB).
The results show that, irrespective of design parameters,
recycle plants with multiple feeds are internally unstable
system (a pole at the origin). The recycle plant can be
stabilized only if the feed flows are exactly the same at
steady-state. This proves Luyben’s conjecture that we have
to balance the reactants down to the last molecule. This also
implies feed-ratio control scheme will not work in practice
and one of the feed flows should be adjusted via feedback.
After eliminating unworkable control structures, remaining
candidate control structures are evaluated next. Globally, for
a given control structure, the bifurcation diagram is used to
identify unstable region (if exist) in the composition space
and, subsequently, a stability boundary can be drawn. Locally, the speed of recycle flows (pole location from
transfer function matrix) is used select the preferable one.
Finally, dynamical compensation to the feed flow is devised
according to the transfer function matrix such that almost
perfect production rate control can be achieved. A nonlinear
recycle process with one reactor and two columns
(programmed in FORTRAN) is used to illustrate the
usefulness of the proposed analysis. Simulation results
clearly indicate that correct control structure is selected and
almost perfect production rate changes can be achieved.
Publisher
臺北市:國立臺灣大學化學工程學系暨研究所
Type
report
File(s)![Thumbnail Image]()
Loading...
Name
922214E002021.pdf
Size
905.42 KB
Format
Adobe PDF
Checksum
(MD5):25beb60608c8a124b0660bdcecb6529b