Page 429 - Design of Simple and Robust Process Plants

416 Chapter 10 The Efficient Design and Continuous Improvement of High-quality Process Plants
A non-validated model is useless for optimization. A good method for validation of
off-line models is to compare the simulated results with the profit meter results.
Adjustments of process parameters or model updating are the degrees of freedoms
to achieve a closer match between plant data and simulated data. The use of off-line
models is limited to situations that are not subject to frequent variations, less than
once per day while process parameters do not change during operation. Closed loop
optimizations must be provided with an on-line profit meter and an integrated data
reconciliation and parameter estimation module to keep the models valid (see
Chapter 9).
Operation optimization pushes a plant against its constraints, and shows oppor-
tunities for improvements. The capacity constraints at environmental conditions are
challenges to operate with maximum variable throughput by taking advantage of
the day and night temperature cycle for capacity maximization.
Summarizing, production should in order to optimize production:

. Install a performance meter on the process, and utilize optimization models
for operation.
. Validate and maintain the optimization models.
. Monitor the process for operational improvements with as reference the per-
formance meter.
. Identify constraints and search for ways to eliminate these.

10.2.5
Opportunities for Design Improvements

The results of the above activities already should give the production and process
engineers/managers some insight into the potential(s) for design improvements.
Opportunities for capacity improvement might be concluded from the process capa-
city performance. Analysis of the data and simulation results will show restrictions
on capacity, such as valves, heat exchangers at environmental extreme conditions, or
others.
Operation optimization results push the process conditions against constraints.
This information is also valuable as de-bottle-necking information. Plant designs are
often based on arbitrary decisions on specifications for recycle streams, while opera-
tion optimizations show the direction that these must take. Design optimizations
should provide the ultimate answer.
Reliability and availability studies are other valuable input data for improvements.
The identification of reliable components, or improvements in less reliable compo-
nents, might lead to single components in a revamp, or to a new facility. On one
occasion, a situation was noticed where a compressor was spared and the equipment
was regularly utilized for back-up. The challenge was to analyze the compressor fail-
ures in detail and solve these to remove the spare compressor. In case supervision
does not take the decision to remove a spare unit, the remaining one will never
achieve the required reliability.