'Another issue is the software used to schedule inspection, maintenance and
replacement for all components in a refinery. Having worked on one of the
more advanced products of this nature (up to a year ago) I can tell you that
the way it handled dates was not what I now know to be "compliant".
What does this mean to the refinery?
If a valve is scheduled for inspection in March 2000 and the inspection
tracking system understands it as 3/1900, the component may not get inspected
or may be shut down for having an overdue inspection.
If a "reading" on, say, a pipe wall thickness in 1/1/2000 is 1/8", and it was
7/16" in 1/1/1999, 4/8" in 1/1/98, and 9/16 in 1/1/97, the pattern of
thinning means that the pipe is deteriorating, deterioration is occurring at
an increasing rate, and that replacement should probably be scheduled at an
interval sooner than 1 year (rather than just scheduling another reading in a
year).
Rates of change may just be looked at for the last "n" readings. If the Y2K
reading is interpreted as 1900 and we just look at the rate of change for the
latest 3 readings, we get the impression that the pipe loses 1/16" each year,
so in 1/2001 we should still have a 5/16" pipe (that's 2 years from the
"last" reading in 1999, so 2/16" would be lost).
7/16" 1999
4/8" 1998
9/16" 1997
1/8" 1900 (not evaluated)
The actual progression would be:
1/8" 2000
7/16" 1999
4/8" 1998
9/16" 1997 (not evaluated)
Obviously the rate of deterioration has increased dramatically between 1/1999
and 1/2000, perhaps due to a new chemical or product temperature going
through this pipe. If the pipe is just scheduled for re-inspection until
1/2001 it is very likely to fail during 2000. Non-compliant date handling
resulted in failure to correctly interpret a critical trend.
Of course, if an inspector saw such thinness in an important pipe, tank, or
vessel, it would probably be reported as needing immediate attention. The
point is that trend analysis in general may be compromised, and in some
situations it could result in failure to predict when replacement or
maintenance is needed. Plants shut down, people die, the environment gets
trashed, whatever.
This same type of software may be used in other major process plants, as well
as for railroads and other infrastructure. The Y2K risk may not be in the
plant itself, but in the tools used to run the plant.
--Dawna
> > How many critical valves have to fail in order for a refinery or pipeline to
>
> And how many microprocessor-driven, critical valves fail every day that we
> never hear about?
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
dejanews.com Create Your Own Free Member Forum
____
From:
dclephas@my-dejanews.com
18:28
Subject:
Re: Y2K & refineries: maybe they won't explode |
