Page 67 - Design of Simple and Robust Process Plants

3.3 Design Philosophies 51
. Control design will be based on fundamental control strategy design for the
selection of the best loop pairing, to maximize disturbance rejection and
minimize interactions, in particular to facilitate operation at the basic control
level if the higher levels of hierarchy are not operable.
. Optimal control loop design is performed based on dynamic simulations,
and incorporates the selection of instruments and its location in process,
valve selection and sizing, vessel dimensions and controller design.
The design of the best pairing of control loops can be performed by making use of
controllability parameters such as the relative gain array to show the steady-state
interaction, and dynamic parameters (e.g., disturbance condition numbers) (Seider
et al., 1999 and control text books).

3.3.7.3 Model-based control
Model-based control (MBC) is also known as model-predictive control, as it may
anticipate predicted conditions (Skogestad and Postlethwaite, 1996). The present
advanced technique for the development of robust, model-based controllers is based
on input/output models which are developed based on a model identification tech-
nique. However, the current development of these controllers has the following dis-
advantages:
. They are designed after the plant has begun to operate.
. They are linear, with a limited operating window.
. The development of an input/output model is time-consuming, and must be
repeated after each process modification or where there are unidentified
operation conditions, for example another feed composition.
. The development of input/output models is inaccurate due to the limited
availability of process measurements.
In future, model-based controllers will also be designed, based on fundamental
dynamic models. The development of these will be less time-consuming as they will
be derived from static (equation-based) models, as part of the life-cycle models. In
the near future, it is likely that model based controllers will become nonlinear and
applicable in a much larger operating window.
The objective will be to automate operation and achieve robust hands-off control
in order to minimize operator intervention, though plant ªtripsº will occur more
often compared with keeping a plant on-flight with minor operator adjustments. As
a result of this higher level of automation, operators will acquire much less hands-
on experience, and will be trained using dynamic plant simulation models. Basic
operator training will focus on plant behavior, and basic response on upsets. The
operator will be trained in situations where the MBC is out of operation, and only
basic control is functional.

Summary
. Human beings possess characteristics that make them less capable consis-
tently to execute the same tasks.