30 May 97
d. The process described above could be
Step 3 - Given the baseline SI and improved SI,
outage frequency and duration are calculated as a
improved by determining the weighting factors in a
function of SI. A simple arithmetic weighting
more objective fashion. For example, the weighting
scheme was used to estimate the SI for a unit.
factors could be determined through a fault tree
Estimates of frequency and duration of outages as a
analysis technique that would interrelate the various
function of SI were developed through opinions of
systems/components and rank them based on the
five hydroelectric power experts. This outage
risk associated with a failure of the system/
information was collected by distributing a series of
component. Systems/components with higher risk
worksheets to hydroelectric power experts. The
to the facility would be assigned a higher weighting
worksheets collected judgments of outage fre-
factor based on the relative magnitudes of the risks.
quency and duration for units with SIs of 90, 80,
e. Estimates of frequency and duration of
60, and 40. Respondents were asked to estimate the
25, 50, 75, and 99 percentile probabilities at each SI
outages as a function of SI were developed through
level. Five respondents successfully executed the
opinions of only five hydroelectric power experts.
exercise and their estimates were used to develop
Historical data were either not available or not used.
frequency SI and duration SI functions to estimate
This process is critical to the accuracy of the overall
outage frequency and duration given an SI. Finally,
QUADRANT process but is based on the sampling
this information was converted into an estimated
of only five experts. This process should be
improved or at the very least the number of experts
should be increased to reduce the uncertainties
Step 4 - The difference in costs with and
associated with expert opinion. Although the SI
without the proposed maintenance are compared to
does provide useful information, it may not be
determine the net benefit of performing the
definitive enough to use alone in this analysis.
maintenance. QUADRANT's output includes costs
There is no substitute for solid historical reliability
and damages for all years, rankings, and cumulative
data. Outage frequency should be based on
initial project costs. A PC version was developed
historical reliability data, not SI values. The SI
for quantification. A dynamic programming
values could be used to adjust the historical
technique was used to compare projects.
reliability information (combined with failure rate
data) to provide a better estimate of the reliability of
c. In addition to current unit condition and
the actual piece of equipment under consideration.
expected improved condition, HYD-QUAD input
f. Finally, the cost estimations are somewhat
includes cost of outage (energy and capacity costs),
interest rates, bowing factors, target SI, zero
simplistic and should be improved. QUADRANT
maintenance age, years to horizon, and total cost of
results only show highly summarized, cumulative
work items. Input information is provided either by
project costs. There is no consideration given to
the district or project office or HQUSACE. Other
repair/construction costs of collateral damages that
intermediate input may include the energy value
could occur from a given failure or the interest costs
plant factor. The QUADRANT methodology is
associated with construction costs.
based on a CI adapted from the REMR program.
Generally, the CI presents a "snapshot"
F-7. Risk Assessment for Nonroutine
representing the absolute condition of a piece of
Closure/Shutdown of Hydroelectric
equipment regardless of its age or maintenance
history. CIs are received from the field and are
based on testing, field observations, and inspector
a. The Department of Energy Pacific
opinion. The CIs are combined into an SI through a
simple weighted average process of five systems.
Northwest Lab under contract with the Corps is
The weighting factors were determined by an
performing a reliability and risk analysis for
opinion poll of 38 hydroelectric power managers.
evaluating nonroutine turbine shutdown scenarios at