Chemical Engineering October 2014 - 7

Letters
Combustion control
It is good to see your magazine publishing articles
highlighting the tangible benefits that process control
is capable of bringing to the chemical process
industries [Advanced Control Methods for Combustion,
Chem. Eng. September, 2014, pp. 34-39].
While the authors seem to be up-to-date as far as
new sensors in the combustion systems is concerned,
they don't seem to be aware of several issues when it
comes to control and advanced control. Even though
all the processes are nonlinear, linear feedback is
capable of successfully dealing with keeping the process
variable close to the setpoint at all times (provided
that the field devices are working properly).
MPC [model predictive control] is also a linear
technology. MPC is not a match to the PID [proportional,
integral derivative] controller, the model
predictive role in process control is optimization
of a process that needs to be steady-state stable
to begin with. MPC is a technique that inherently
optimizes setpoint tracking, and in consequence it
has very poor performance when it comes to load
disturbance rejection.
Sigifredo Nino, Process Control Consultant
Summa Control Solutions Inc., Montreal, Canada
We are very pleased about the professional reflection
our paper has initiated!
Control theory is of course a rather diversified
area, as it is reflected by the internal structure of
the paper. Although this topic covers the majority
of it, it was still not possible to touch a number of
rather important issues, some of which the commenter
missed. This is a good opportunity for clarifying
some issues that seem to be the most important
ones as follows:
Yes, it is an established fact that disturbance
rejection capabilities of MPC do not belong to its
strengths, it very often falls behind a PID controller.
However, this is true only for unmeasured disturbances!
One of the great strengths of MPC is its
capability for excellent, model-based compensation of
measured disturbances, regarding which it definitely
outperforms many other control methods. And the
second well-known weak point of MPC, which the
commenter did not mention, is its higher sensitivity
against model inaccuracies.
Yes, the basic MPC algorithm is a linear one, however,
significant process nonlinearities can be handled
by means of several nonlinear MPC versions, or
some other approaches like " multimode control " or
" gain scheduling " as mentioned later in the article.
Pál Szentannai
Budapest University of Technology and Economics
Maximilian Lackner
Vienna University of Technology
Editor's note: The above are excerpts of letters received on this topic.
For the full text, and to take part in the discussion, search the article
title, Advanced Control Methods for Combustion, on our website at www.
chemengonline.com
Cutting-edge technology
for level measurement.
What more could you want?
http://www.chemengonline.com
Chemical Engineering October 2014

Table of Contents for the Digital Edition of Chemical Engineering October 2014
