Page 259 - Design of Simple and Robust Process Plants

6.4 Application of Reliability Studies for a Process and Utility Plant 245
. The calculation method assumes that all the steam-producing units are
operational, except when they are on scheduled or forced outage.
. Only one unit is scheduled for an outage at a time.
. Common mode failures are not assumed for the different units, including
electrical power losses.
. Electrical power reductions are assumed not to have an effect on the steam
generation.
The results of the reliability study are summarized in Table 6.7, the conclusions
being that:
. Case 4 has the lowest failures per year, as the number of spare units is still
one during a scheduled outage. The failure rate of a boiler unit is low com-
pared with the systems with a GT, which have high failure rates.
. Case 2, being the lowest performer, is analyzed. It is clear that there are rela-
tively cheap methods to decrease the number of forced outages. First, a triple
redundant control system halves the number of forced outages. Second, an
increase in the duty of the WHB with 33% reduces the number of failures
per year from 11.7 to 1.6 for the triplex control system, a decrease by a factor
10. These actions would bring the performance to the next best.
Table 6.7. Number of failures per year to supply 380 Mton/year steam.

Cases Failures per year Failures per year

Case 1 4.2
Case 2A 23.4 (Simplex) 11.7 (Triplex)
Case 2B 3.2(Simplex) 1.6 (Triplex)
Case 3 2.1
Case 4 0.22

Overall conclusions of the reliability study of the utility system
. The results form a good basis for an economic evaluation to select the opti-
mal utility system.
. The impact of process failures on a site might be reduced by the application
of a steam load-shedding system. Such a system should be part of the evalua-
tion, as it is relatively cheap to operate, and easy to implement with current
control techniques.
. The results depend heavily on its assumptions, to mention specifically com-
mon cause failures. The design of each system should be totally independent
to achieve the calculated failure numbers. The feed water system with its treat-
ment, e.g., is commonly shared with other boilers, though this would require a
specific reliability study. A detailed RBD and its quantification with all interact-
ing components will show the weak points in the design. This is also applic-
able for other boiler-related systems which are shared. An error here might
lead to total forced site outages, with associated very high cost penalty.