Page 251 - Design of Simple and Robust Process Plants

6.3 Methodology of Reliability Engineering Techniques for the Design of Process Plants 237
lected data can be used for other designs. Sources are the public data banks
(e.g. Oreda, 1992; CCPS, 1989). On occasion, we might receive data from
equipment suppliers for complete systems such as compressor stations, gas-
turbine systems, and power generators. Care must be taken not to use the
supplier's guaranteed data as these are conservative, but an attempt should
be made to obtain actual performance data. These are, in general, quite accu-
rate. Collect the company reliability data, particular for components in a spe-
cific service and utility supplies. It may be useful to confirm the utility data
with a detailed reliability study of the utility systems, complete with its redun-
dancy and load shedding system (see Chapter 7). Some overall data for pro-
cess equipment are shown in Table 6.2, though these are to be used for ref-
erence, and not for design.
. To collect repair and process interruption times, and to set mission time(s).
Repair times must be determined per item, including the time required for
spare parts to be available at location. Critical spare parts selected at its avail-
ability and failure rate might be available at the site. Repair times need to be
supported by maintenance, as they have an impact on the organization, since
maintenance people on call need time to arrive at the location. Process inter-
ruption times have to be determined with production per item. These times
include the time required to make the failed component available for mainte-
nance, and it also emphasizes the time required to get back into production.
Repair times are not included. The repair and process interruption time to-
gether set the system unavailability per component per failure.
. Mission time has to be set. For continuous plants, mission times of 4±5 years
have been shown to be achievable for world-scale plants. The mission time is
set by:
± Inspection time of specific items, such as pressure vessels.
± Maintenance on components which become subject to wear-out, for example
seals and bearings, and those subject to erosion and/or corrosion.
± Process improvements to upgrade economically the plant, together with evo-
lution of the technology.
. To quantify next to repair cost, any cost-related to process damage as a conse-
quence of a failure. Any process damage might result in high cost as per:
rework of product, additional production loss as a result of lost batches,
equipment damage which is sometimes applicable to high-temperature pro-
cesses or polymerization processes. These cost data are in addition to any lost
production capacity due to unavailability of the process. These cost data are
needed for the evaluation of project alternatives.
. The model has to run, to collect, and to accumulate the reliability and avail-
ability data. The probabilistic simulations must run for sufficient time to
obtain statistically relevant information. The experienced gained for a process
plant run time over 50 years and a random number of input data selection of
100 was sufficient, though this must be confirmed by making initially some
more runs. The following data are collected over the mission time: