Page 68 - Design of Simple and Robust Process Plants

52 Chapter 3 Design Philosophies
. Total automation and robust hands-off control can overcome these handicaps,
and will result in considerable savings.
. A common automation strategy needs to be applied for different process
plants operated by the same personnel.
. Control design for basic control as well as model-based control will be based
on fundamental dynamic models, while nonlinear controllers will find appli-
cations.
. The training of operators will need to be adapted to concur with their new
role.

3.3.8
Operation Optimization Makes Money
Operation optimization is a broad area where considerable savings can be achieved.
The optimization might be split into business models and process operational mod-
els, but as both types have a similar basis they will each be mentioned in the follow-
ing section.
The business models include:

. Planning models for products; this is particularly important for batch plants
and continuous plants with campaign operation.
. Supply chain optimization applied to product distribution over several custo-
mers.
. Feedstock evaluation; this is particularly important for plants that handle dif-
ferent feed stocks, for example refineries and ethylene plants.
. Overall operational models (as used for refinery complexes) to optimize a
chain of plants.
The process optimization models referred to include:

. Scheduling for batch and campaign operations
. Scheduling of equipment in relation to batch operations
. Optimization for continuous processes based on actual feedstock, product,
and utility prices. The optimization emphasizes the selection of:
± optimal process conditions
± optimal capacity (if not set by the business)
. Optimization of dynamic operations as batch and transient operations
regarding operational and capacity conditions. Optimization of the run times
of fouling systems will be carried out in combination with the operational
conditions, e.g., catalyst aging or coke formation.
The optimization models for businesses are based mainly on linear programming
(LP) techniques, which are faster to develop and execute. The businesses need to
explore many options ± hence the need for a rapid response. In addition, feedstock
evaluators and overall operational models need to be highly accurate, and are there-
fore derived from the more accurate process optimization models.