Page 280 - Design of Simple and Robust Process Plants

266 Chapter 7 Optimization of an Integrated Complex of Process Plants
be optimized versus the loss of production due to failure of the upstream and down-
stream processes. The production loss of the downstream plant is what counts, as
that material is not available for further processing or sale. The example is used to
illustrate the calculation method, but it can be extended to any storage provision in a
supply chain.

7.6.1
Plant Failures
The first steps that must be taken are the quantification of the failures; this is also
expressed as the unavailability of both processes, upstream and down stream (Shorr
and Griffin, 1991). The availability of plants can be divided into mechanical availabil-
ity and process availability, including planned stops.
The mechanical availability of the plant is affected by its components with their
own typical failure rates and distribution, and can be quantified with a reliability
engineering technique. It can be simulated with a software flowsheeter, which is
provided with a random generator (Malchi Science, Clockwork Design). The plant is
divided into components or a set of components of which the failure rates and their
distribution must be collected. There are different sources for these data, there are
several public data banks (e.g., CCPS, Oreda), and data banks of equipment suppli-
ers. Processing companies may also have their own data banks. Besides failure data,
repair times and process recovery times (start and stop times), together with their
distribution, are also required. The last data are collected for the process under
study from maintenance and plant data. The utilities and supplies of the process
also are analyzed for their failure rates. The individual components or set of compo-
nents are implemented in a flowsheeter and simulated as a set of serial and parallel
components with their specific properties. The simulation with its random genera-
tor runs for an extended time (called the mission time), and for a number of runs.
After a sufficient number of simulations, the results will converge to a set, with lim-
ited variation. The results may be expressed in several ways as: the total number of
stops per year; the average duration of these stops and its distribution; the down-
time per year; the number of stops per type of components and its spread. The over-
all results give the plant mechanical failure rates and its distribution. The technique
can be used to compare several process alternatives that affect the reliability of the
design, and there will be a trade-off between production increase and capital. The
best economic alternative will be a preferable choice as bases for design. For details
on the reliability methodology for process design, see Chapter 6.
The plant mechanical failure rates must be complemented with the process fail-
ure rates. Process failures might be caused by fouling, catalyst degradation, and
operational failures. The process failures are sometimes covered in planned stops
(in which case it is not seen as a failure), but it will be included in the overall avail-
ability. The data for process failure rates can only be collected from careful analysis
of process data over the past. These data become more accurate if more historic data
are available. Licensees of the technology sometimes collect these data for their spe-
cific processes. The overall plant failure of both process and mechanical failures