Site
Map
Return to a Note on
Generator
Core Failure Mechanisms
Generator Core Fault Case History : Pacificorp
Hunter 1 :
24/11/2000
[Information source : Public
Service Commission of Utah and summarised by Dr. Antony
Anderson
C.Eng FIEE ]
According to information laid before the Public
Service
Commission of Utah, by Mr Barry Cunningham, Vice President of
Generation,
Pacificorp, a forced outage occurred on Hunter Unit 1 on November
24th 2000, in which the generator stator core was severely
damaged.
The rebuilt stator was restored to service on May 8th 2001.
Pacificorp's
attorneys claim that the Company incurred $270.1 million in net
purchased
power costs in relation to the outage. Summarising the
information in Mr Cunningham's
deposition :
- Hunter 1 is a 430MW 496 MVA hydrogen cooled
Westinghouse
generator operating at 24 kV. 28 similar machines have been built
during
the last 30 years and are in service in the US and Spain. The core is
104
ins (2.66 m) diameter 225 ins (5.76 m) long. The laminations are of
0.018
in (0.46 mm) varnished coreplate, 1/3 and 2/3 lapped, 9 segments
per layer, located on 27 semi-circular keybars. Each lamination locates
on 3 keybars. 27 insulated tie rods pass through axial holes in the
core
approximately mid-way between the tooth tips and the back of core. The
stator core is split into 3 inch packets separated by 0.125 in
(3.2mm)
radial cooling ducts.
- Unit 1 was operating at 415 MW at time of
fault.
First indications
of trouble were at 12:38:53, when the No. 5 bearing alarmed at
-262.6
degrees F (impossibly low). 40 seconds later the No. 6 bearing alarm
operated
because the vibration level had risen to 5.24 mils displacement (0.133
mm), which is over the alarm level of 5.0 mils (0.127mm). Shortly
afterwards
a winding cooling gas differential alarm annunciated and then returned
to normal. Sparks were noted at joints between the generator frame and
the cowling and heavy arcing around ground straps near the no 5
bearing.
The unit tripped automatically due to the loss of field relay
operating.
Total time to trip was 5.5 minutes.
- The damage was worst at the exciter end but
extended the
entire length of the core and formed a tunnel-like hole varying in
diameter
from 1.5 to 5 in diameter (38.1mm to 127 mm). The area involved was
between
the slot bottom and the through bolt hole. Molten metal from this hole
spilled out over the end windings. There was some collateral damage at
two other locations at the exciter end over axial lengths of 2 ft and 4
ft and a small amount of collateral damage at the turbine end in the
form
of melting between the flux shield and a through bolt.
- The causes of the failure have not been
determined.
Location
of initial fault is thought to be 5-6 feet from exciter end, in
the
area between the bottom of slot and a through bolt. Evidence of the
root
cause is considered likely to have been destroyed during the failure.
Comments by AFA:
Unfortunately, Mr Barry
Cunningham's deposition is no longer available on the Utah State
Government website. [See old references below].
The
Hunter 1 failure is however briefly mentioned on the Exponent website
at: http://www.exponent.com/practices/electrical/cases2.html.
It appears that Exponent were asked by Pacificorp to examine the
body of evidence after the unit had been repaired and was back on line.
Exponent concluded that "the failure was an unprecedented and
unpredictable event."
Hunter 1 is by no means the only recorded case of a core failure on a
machine with through-bolt core clamping. Therefore, in my view, the
core failure cannot strictly be said to be unprecedented. Nor could the
core failure be claimed to be unpredictable, since, should a
through bolt become loose, eventually a core fault would become a
distinct possibility. Only the timing of the core fault would be
unpredictable.
"On
line failure vs Preventive Maintenance" by Jim Taillon AGT Services
[Iris Rotating Machines Conference 2003 , Santa Barbara CA]
describes core damage at through-bolt positions on Machine
A , 10 years old, of 68.8 MVA and Machine B, 40 years old, of 64
MVA. Of particular note is Machine B on which, according to
Taillon page 3: " ...The thru bolt insulation also had core iron
impressions down its length. The bolt was apparently loose enough to
vibrate and hit the core hard enough to cut small grooves into the
insulation..." These observations of incipient core damage on
Machine B, suggest strongly the mechanisms at work in this type
of internal core fault to be: (1) a loose tie rod that
allows transverse vibration of the rod and its
insulation under the influence of the rotating
radial/circumferential magnetic field in the axial duct (2)
fretting of insulation because of vibration and eventual thinning
to the extent that the through bolt makes contact with
lamination edges at one or more points along its axial length (3) given
core to frame contacts that correspond with tie rod to lamination
contacts, a circulating radial/axial eddy current with high local
losses at tie rod to lamination and lamination to keybar points will
result (4) eventually a runaway core fault may be established.
Generally speaking core faults of this type in the body of the core are
unlikely in machines where core clamping is by a cantilever arrangement
involving the core end plate and the core fingers and where the through
bolts are dispensed with. The exception is when a conducting
foreign body lodges in one of the axial vent ducts. See Image
Note: 12 January
2006 : The following Utah Government references listed below seem to
have been removed from the Internet: -
Return to a Note on
Generator
Core Failure Mechanisms
Summary prepared 1st December 2001 and updated Jan 12th 2006
by Antony
Anderson