| |
Abstract:
(Invited Talks)
Modern industrial plants are mostly characterized by highly
complex processes and layouts. The complexity stems from the number
of inputs and outputs, the frequent occurrence of delays, the
inherent process non-linearity and from the high degree of
interconnection between the various process units (heat
regeneration, product flow recirculation, etc.) which has steadily
increased over the last 10 years. Complexity management strategies
have largely relied on new technologies able to provide better
control and monitoring functions, eventually proving themselves
able to increase return on existing capital investment in plants.
In this framework, Neural technology has distinguished itself as a
very effective tool to solve modeling and optimization problems in
the continuous process industry. However its application is still
not so widespread as it should be, the main hurdles being much more
related to difficulties in approaching the control room environment
than to actual algorithmic or technological challenges. The talk
discusses how Neural Networks can be introduced as normal, basic
components inside complex, high-performing control strategies. The
key issue is taking care of all the side aspects so that the new
functionalities are perceived as a relief and not as an additional
burden or a complication. Control room personnel is rightly
conservative because he is requested to keep complex, uncertain and
not completely observable plants running in almost any operating
conditions and under an increasing number of constraints
(economical, legal, environmental, etc.). Successfully applications
cannot be dropped by just a management's decision into the control
room: it should be very clear that plant operators are the actual
"end-users" and any project needs their earnestly support and
involvement. The best way to let them fully exploit the advantages
new technologies bring, is to carefully percolate new capabilities
into the familiar control room equipment and diplays. Our
successful experience is based on a tight cooperation between the
American and Italian company units. The proper merging of
technologies, local capabilities and process know-how made
available to the Italian market, is allowing the national process
industry community to really benefit from what high-tech may
offer.
Nunzio Bonavita was born in 1960 at Stazzema (Lucca), Italy. He
graduated in Particle Physics from the University of Pisa in 1986.
After having gained experience in data analysis and modeling in the
experiment for the Top Quark quest at Fermi National Laboratory
(Batavia - Illinois), he joined Elsag Bailey Hartmann & Braun
SpA, Genova, Italy. Nunzio has been working in the field of
Simulation and Advanced Process Control for 12 years. Currently he
is APC Group Leader in the Advanced Application Services
department. In his current position, he is responsible for
coordination of advanced process control implementation for power
plants and process industries. He has published over 30 papers on
these subjects.
|
