01/09 ·
uptime
m
e
gazine
renewable
energy
department with the required input to
plan any required service actions.
Operator data access
The geographical separation of tur-
bines, monitoring centre, service of-
fice and turbine owner necessitates a
web-based multi-user access to the
recorded condition monitoring data
and the alarm status of the turbines.
In support of the Vestas condition
monitoring implementation, Brüel
& Kjær Vibro has developed a CMS
User Home Page which features ac-
cess to the alarm status on both tur-
bine and park level, and facilities to
review and compare trend curves of
the recorded scalar values [Figure 7
and Figure 8]. These functions allow
the service management department
instantaneous overview of the park
wear condition and enables identifi-
cation of possible outliers.
Furthermore the access to trend
curves makes it possible for the serv-
ice technicians to quickly check the
effect of their service actions "did
the vibration level drop after we
greased the bearing?".
Performance
monitoring
Like other business activities the
cost-efficiency of a condition monitor-
ing programme must be continuously
monitored and evaluated at regular
intervals. The condition monitoring
business is abundant with hear-say
evidence about either the wonders it
works or the opposite. The frequency
of failures and the progression of
the individual failure modes is a sta-
tistical phenomenon. To establish a
performance baseline it is therefore
important to implement systematic
collection of the data associated with
all failures, both detected and un-de-
tected. In Vestas this is implemented
by treating all condition monitoring
alarms as a transaction which is not
closed before the service department
has provided feedback to the alarm.
Thus all alarms and component in-
spection reports are logged along
with information about the diagnosed
severity class and failure mode.
These data allows Vestas to perform
statistical analyses about component
life time and frequency of failure
modes which can be applied as
input to component design improve-
ments. Also the efficiency of condi-
tion monitoring can be measured
from statistics over the achieved lead
times from fault detection to function
failure. Finally the technical perform-
ance of the condition monitoring
equipment is monitored by logging
the system availability and frequency
of hardware and system errors.
Conclusion
This paper has outlined the elements
which are required for a successful
implementation of a large scale con-
dition monitoring programme for wind
turbines. This includes centralised
monitoring with advanced alarm man-
agement which helps avoid alarm
flooding and only provides one new
alarm per detected physical fault or
changed severity level. Essential to
the handling of condition monitor-
ing in the service departments is the
transformation of data and machine
generated alarms into actionable in-
formation, that is, service technicians
get clear advice about what to do and
within what time frame. Furthermore,
to ensure continuous improvement
of the condition monitoring system
settings, it is very important that field
findings are reported back to the
diagnostic experts "was the condi-
tion of this component as severe as
diagnosed?" Finally, all data must
be logged so that the technical and
financial performance of the condition
monitoring programme can be moni-
tored.
n