Optimization
The purpose of Optimization is to:
Improve the profitability
of the process units by generating new optimal
setpoints and targets which can be either:
Downloaded to lower level controllers in
real-time, or
Implemented offline
The objective of online closed
loop optimization is to drive the
process unit towards its optimum (maximum
profit) point. The optimizer will calculate
the targets in real time based on the economic
data, process conditions, equipment efficiencies,
and feed stock availability. Essentially,
the business and economic objectives are
converted into operating targets. The solution
is then implemented by the lower level controllers
and other APC schemes.
The critical distinction between
online and offline optimization
is based on the number and nature of the
process and equipment constraints which
can affect optimization results in real-time.
Usually, Multivariable
Predictive Controllers also
have some sort of cost optimization included
in their control algorithm, but this higher
level optimization is process model based,
allows nonlinear optimization for wider
operating process domain, and usually has
a profit maximization objective.
There is a choice in the type of process
models to be used for optimization - closed
models and open equation models. With the
availability of large scale nonlinear optimization
algorithms (over 50,000 nonlinear equations),
open equation models are gaining popularity.
These models can include the required functions
of data reconciliation and steady state
detection within the same model. While benefits
normally increase as the type of optimization
moves towards online open equation model
type optimization, the costs increase very
rapidly. It then becomes a judicious choice
to select the level of optimization and
process model types based on incremental
benefits, overall service factor and anticipated
maintenance requirements. Our approach has
been to select the level of technology which
can realize the major portion of the benefits
with a relatively high level of return on
investment.
Some of our Real Time Optimization applications
have been:
Waste Incinerator Load Maximization:
A Gensym/G2 based program was developed
to infer the possible operating modes from
process conditions. These modes were converted
into mathematical models online and a mixed
integer optimizer embedded in G2 maximized
throughput by allocating the multiple vapor
and liquid streams into multiple incinerators.
Energy Management System:
The Gensym/G2 based I-EMS
program converts an offline MESA utility
system model into objects (e.g. gas turbines,
boilers, etc.) for online optimization.
The I-EMS program validates input data,
checks for steady state and verifies new
setpoints for safe and feasible optimization
results. It compares levels of optimizations
for decision making, monitors equipment
efficiencies and can also be customized
for process advisory purposes.
Real
Time Optimization Study
