Page 53 - Design of Simple and Robust Process Plants

3.2 Modeling 37
The migration of the modeling tools in the proposed direction should, according to
CEFIC, lead to excellence in:

. product innovation;
. improved and new unit operations and processes;
. enhanced interaction between chemistry, engineering, operation, business
and product development, and
. optimized process operations.

3.2.1
Process Modeling is the Way to Improve Your Process

ªAny model is better than no modelº. This ideal can be compared with the training
program of a sportsman, who might suggest that ªany training scheme is better
than no schemeº. For the design and economic operation of chemical processes
within safe and environmentally sound constraints, models are essential. The above
statements all reflect vision inclusively the vision of CEFIC that process knowledge
resides in models, and we should realize that a complete range of modeling environ-
ments (simulators) and models are available commercially.
The following differentiation's may be made among models:

. Linear versus nonlinear models (LP versus NLP)
. Static versus dynamic models
. Sequential versus equation-based models
. Continuous versus mixed integer model environment for optimization (NLP
versus MINLP and LP versus MILP)
. Unit models versus flowsheet models
. Empirical models versus fundamental models
. Design versus rating models
. Specific models such as:
± molecular modeling
± flow modeling/mixing
± product property models as a function of reactor conditions, for example
polymer properties
± vent sizing packages
± reliability modeling
± dispersion modeling
± quantitative risk analysis models
All these models have common aspects that require specific attention, with model
validation being one of the most important. The constraints (and their applicability
range) need to be known, and are often based on the basic assumptions and applic-
ability of certain equations.
The accuracy requirements of the model might differ for each application. It is
understandable that any inaccuracy of an operation optimization model results in a
lost opportunity, and this will be discussed in Chapter 9. This in comparison with